{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8007170600537795,
  "eval_steps": 500,
  "global_step": 5360,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00014938751120406333,
      "grad_norm": 0.5712769031524658,
      "learning_rate": 2e-05,
      "loss": 1.7864,
      "step": 1
    },
    {
      "epoch": 0.00029877502240812666,
      "grad_norm": 0.8328073620796204,
      "learning_rate": 4e-05,
      "loss": 1.954,
      "step": 2
    },
    {
      "epoch": 0.00044816253361219,
      "grad_norm": 0.3498156666755676,
      "learning_rate": 6e-05,
      "loss": 1.0704,
      "step": 3
    },
    {
      "epoch": 0.0005975500448162533,
      "grad_norm": 0.46848297119140625,
      "learning_rate": 8e-05,
      "loss": 1.3401,
      "step": 4
    },
    {
      "epoch": 0.0007469375560203167,
      "grad_norm": 0.5157446265220642,
      "learning_rate": 0.0001,
      "loss": 1.2173,
      "step": 5
    },
    {
      "epoch": 0.00089632506722438,
      "grad_norm": 0.34888526797294617,
      "learning_rate": 0.00012,
      "loss": 1.2156,
      "step": 6
    },
    {
      "epoch": 0.0010457125784284435,
      "grad_norm": 0.25781184434890747,
      "learning_rate": 0.00014,
      "loss": 1.1595,
      "step": 7
    },
    {
      "epoch": 0.0011951000896325067,
      "grad_norm": 0.39395585656166077,
      "learning_rate": 0.00016,
      "loss": 1.447,
      "step": 8
    },
    {
      "epoch": 0.00134448760083657,
      "grad_norm": 0.42594340443611145,
      "learning_rate": 0.00018,
      "loss": 0.8779,
      "step": 9
    },
    {
      "epoch": 0.0014938751120406335,
      "grad_norm": 0.29849162697792053,
      "learning_rate": 0.0002,
      "loss": 1.214,
      "step": 10
    },
    {
      "epoch": 0.0016432626232446967,
      "grad_norm": 0.206129252910614,
      "learning_rate": 0.00019999998895420804,
      "loss": 1.222,
      "step": 11
    },
    {
      "epoch": 0.00179265013444876,
      "grad_norm": 0.20143035054206848,
      "learning_rate": 0.0001999999558168346,
      "loss": 1.2729,
      "step": 12
    },
    {
      "epoch": 0.0019420376456528235,
      "grad_norm": 0.3609657883644104,
      "learning_rate": 0.00019999990058788703,
      "loss": 1.14,
      "step": 13
    },
    {
      "epoch": 0.002091425156856887,
      "grad_norm": 0.22608326375484467,
      "learning_rate": 0.00019999982326737747,
      "loss": 1.2138,
      "step": 14
    },
    {
      "epoch": 0.00224081266806095,
      "grad_norm": 0.40498849749565125,
      "learning_rate": 0.00019999972385532303,
      "loss": 1.4689,
      "step": 15
    },
    {
      "epoch": 0.0023902001792650133,
      "grad_norm": 0.215151846408844,
      "learning_rate": 0.0001999996023517457,
      "loss": 1.0072,
      "step": 16
    },
    {
      "epoch": 0.0025395876904690767,
      "grad_norm": 0.28476935625076294,
      "learning_rate": 0.0001999994587566723,
      "loss": 0.9636,
      "step": 17
    },
    {
      "epoch": 0.00268897520167314,
      "grad_norm": 0.2294227033853531,
      "learning_rate": 0.0001999992930701345,
      "loss": 0.8722,
      "step": 18
    },
    {
      "epoch": 0.0028383627128772035,
      "grad_norm": 0.24489282071590424,
      "learning_rate": 0.00019999910529216902,
      "loss": 0.913,
      "step": 19
    },
    {
      "epoch": 0.002987750224081267,
      "grad_norm": 0.24592871963977814,
      "learning_rate": 0.00019999889542281728,
      "loss": 1.2072,
      "step": 20
    },
    {
      "epoch": 0.00313713773528533,
      "grad_norm": 0.19988738000392914,
      "learning_rate": 0.0001999986634621256,
      "loss": 1.0463,
      "step": 21
    },
    {
      "epoch": 0.0032865252464893933,
      "grad_norm": 0.16961582005023956,
      "learning_rate": 0.00019999840941014525,
      "loss": 0.8549,
      "step": 22
    },
    {
      "epoch": 0.0034359127576934568,
      "grad_norm": 0.1631389707326889,
      "learning_rate": 0.0001999981332669324,
      "loss": 0.8112,
      "step": 23
    },
    {
      "epoch": 0.00358530026889752,
      "grad_norm": 0.1402096003293991,
      "learning_rate": 0.00019999783503254803,
      "loss": 1.0202,
      "step": 24
    },
    {
      "epoch": 0.0037346877801015836,
      "grad_norm": 0.3608367443084717,
      "learning_rate": 0.000199997514707058,
      "loss": 1.4076,
      "step": 25
    },
    {
      "epoch": 0.003884075291305647,
      "grad_norm": 0.15870416164398193,
      "learning_rate": 0.0001999971722905331,
      "loss": 0.9372,
      "step": 26
    },
    {
      "epoch": 0.00403346280250971,
      "grad_norm": 0.21479852497577667,
      "learning_rate": 0.00019999680778304897,
      "loss": 1.0237,
      "step": 27
    },
    {
      "epoch": 0.004182850313713774,
      "grad_norm": 0.6759325861930847,
      "learning_rate": 0.00019999642118468614,
      "loss": 1.7162,
      "step": 28
    },
    {
      "epoch": 0.004332237824917837,
      "grad_norm": 0.15937522053718567,
      "learning_rate": 0.00019999601249552998,
      "loss": 0.7851,
      "step": 29
    },
    {
      "epoch": 0.0044816253361219,
      "grad_norm": 0.12428741902112961,
      "learning_rate": 0.00019999558171567082,
      "loss": 0.9342,
      "step": 30
    },
    {
      "epoch": 0.004631012847325963,
      "grad_norm": 0.2223120778799057,
      "learning_rate": 0.0001999951288452038,
      "loss": 1.1871,
      "step": 31
    },
    {
      "epoch": 0.004780400358530027,
      "grad_norm": 0.19341352581977844,
      "learning_rate": 0.000199994653884229,
      "loss": 0.9797,
      "step": 32
    },
    {
      "epoch": 0.00492978786973409,
      "grad_norm": 0.41403812170028687,
      "learning_rate": 0.0001999941568328513,
      "loss": 0.9629,
      "step": 33
    },
    {
      "epoch": 0.005079175380938153,
      "grad_norm": 0.29959213733673096,
      "learning_rate": 0.00019999363769118055,
      "loss": 1.231,
      "step": 34
    },
    {
      "epoch": 0.005228562892142217,
      "grad_norm": 0.17178373038768768,
      "learning_rate": 0.00019999309645933142,
      "loss": 0.7834,
      "step": 35
    },
    {
      "epoch": 0.00537795040334628,
      "grad_norm": 0.2903209626674652,
      "learning_rate": 0.00019999253313742344,
      "loss": 1.2852,
      "step": 36
    },
    {
      "epoch": 0.005527337914550344,
      "grad_norm": 0.3148847818374634,
      "learning_rate": 0.00019999194772558112,
      "loss": 1.376,
      "step": 37
    },
    {
      "epoch": 0.005676725425754407,
      "grad_norm": 0.4141716957092285,
      "learning_rate": 0.00019999134022393375,
      "loss": 0.9684,
      "step": 38
    },
    {
      "epoch": 0.0058261129369584705,
      "grad_norm": 0.2509758770465851,
      "learning_rate": 0.00019999071063261554,
      "loss": 1.1722,
      "step": 39
    },
    {
      "epoch": 0.005975500448162534,
      "grad_norm": 0.1800830066204071,
      "learning_rate": 0.0001999900589517656,
      "loss": 1.0264,
      "step": 40
    },
    {
      "epoch": 0.006124887959366597,
      "grad_norm": 0.19982534646987915,
      "learning_rate": 0.00019998938518152787,
      "loss": 0.7343,
      "step": 41
    },
    {
      "epoch": 0.00627427547057066,
      "grad_norm": 0.1816824972629547,
      "learning_rate": 0.0001999886893220512,
      "loss": 0.9136,
      "step": 42
    },
    {
      "epoch": 0.006423662981774723,
      "grad_norm": 0.216154083609581,
      "learning_rate": 0.0001999879713734893,
      "loss": 1.0128,
      "step": 43
    },
    {
      "epoch": 0.006573050492978787,
      "grad_norm": 0.16445758938789368,
      "learning_rate": 0.0001999872313360008,
      "loss": 0.9593,
      "step": 44
    },
    {
      "epoch": 0.00672243800418285,
      "grad_norm": 0.14644479751586914,
      "learning_rate": 0.00019998646920974919,
      "loss": 0.9424,
      "step": 45
    },
    {
      "epoch": 0.0068718255153869135,
      "grad_norm": 0.3356267809867859,
      "learning_rate": 0.00019998568499490283,
      "loss": 0.8878,
      "step": 46
    },
    {
      "epoch": 0.007021213026590977,
      "grad_norm": 0.4974273443222046,
      "learning_rate": 0.00019998487869163497,
      "loss": 1.338,
      "step": 47
    },
    {
      "epoch": 0.00717060053779504,
      "grad_norm": 0.1881432980298996,
      "learning_rate": 0.00019998405030012371,
      "loss": 0.882,
      "step": 48
    },
    {
      "epoch": 0.007319988048999104,
      "grad_norm": 0.1419169157743454,
      "learning_rate": 0.0001999831998205521,
      "loss": 0.7255,
      "step": 49
    },
    {
      "epoch": 0.007469375560203167,
      "grad_norm": 0.16964636743068695,
      "learning_rate": 0.00019998232725310796,
      "loss": 1.0151,
      "step": 50
    },
    {
      "epoch": 0.007618763071407231,
      "grad_norm": 0.18110951781272888,
      "learning_rate": 0.0001999814325979841,
      "loss": 0.9705,
      "step": 51
    },
    {
      "epoch": 0.007768150582611294,
      "grad_norm": 0.13390210270881653,
      "learning_rate": 0.00019998051585537818,
      "loss": 0.6999,
      "step": 52
    },
    {
      "epoch": 0.007917538093815357,
      "grad_norm": 0.13011352717876434,
      "learning_rate": 0.00019997957702549269,
      "loss": 0.5496,
      "step": 53
    },
    {
      "epoch": 0.00806692560501942,
      "grad_norm": 0.30082619190216064,
      "learning_rate": 0.00019997861610853503,
      "loss": 1.1625,
      "step": 54
    },
    {
      "epoch": 0.008216313116223483,
      "grad_norm": 0.18591542541980743,
      "learning_rate": 0.00019997763310471752,
      "loss": 0.8686,
      "step": 55
    },
    {
      "epoch": 0.008365700627427548,
      "grad_norm": 0.20125310122966766,
      "learning_rate": 0.00019997662801425725,
      "loss": 0.7371,
      "step": 56
    },
    {
      "epoch": 0.00851508813863161,
      "grad_norm": 0.14571277797222137,
      "learning_rate": 0.00019997560083737632,
      "loss": 1.0109,
      "step": 57
    },
    {
      "epoch": 0.008664475649835674,
      "grad_norm": 0.1365530639886856,
      "learning_rate": 0.00019997455157430165,
      "loss": 0.7709,
      "step": 58
    },
    {
      "epoch": 0.008813863161039737,
      "grad_norm": 0.14306791126728058,
      "learning_rate": 0.000199973480225265,
      "loss": 0.7487,
      "step": 59
    },
    {
      "epoch": 0.0089632506722438,
      "grad_norm": 0.2880021929740906,
      "learning_rate": 0.00019997238679050308,
      "loss": 1.2318,
      "step": 60
    },
    {
      "epoch": 0.009112638183447864,
      "grad_norm": 0.1913180649280548,
      "learning_rate": 0.00019997127127025746,
      "loss": 0.8861,
      "step": 61
    },
    {
      "epoch": 0.009262025694651926,
      "grad_norm": 0.44290822744369507,
      "learning_rate": 0.00019997013366477453,
      "loss": 1.3866,
      "step": 62
    },
    {
      "epoch": 0.00941141320585599,
      "grad_norm": 0.12451759725809097,
      "learning_rate": 0.00019996897397430563,
      "loss": 0.855,
      "step": 63
    },
    {
      "epoch": 0.009560800717060053,
      "grad_norm": 0.2885231375694275,
      "learning_rate": 0.00019996779219910696,
      "loss": 1.147,
      "step": 64
    },
    {
      "epoch": 0.009710188228264117,
      "grad_norm": 0.14056627452373505,
      "learning_rate": 0.00019996658833943957,
      "loss": 0.7883,
      "step": 65
    },
    {
      "epoch": 0.00985957573946818,
      "grad_norm": 0.14305633306503296,
      "learning_rate": 0.00019996536239556942,
      "loss": 0.5779,
      "step": 66
    },
    {
      "epoch": 0.010008963250672244,
      "grad_norm": 0.17166224122047424,
      "learning_rate": 0.00019996411436776738,
      "loss": 0.9014,
      "step": 67
    },
    {
      "epoch": 0.010158350761876307,
      "grad_norm": 0.15985064208507538,
      "learning_rate": 0.0001999628442563091,
      "loss": 0.9639,
      "step": 68
    },
    {
      "epoch": 0.010307738273080371,
      "grad_norm": 0.21808995306491852,
      "learning_rate": 0.0001999615520614752,
      "loss": 0.7525,
      "step": 69
    },
    {
      "epoch": 0.010457125784284434,
      "grad_norm": 0.21660713851451874,
      "learning_rate": 0.00019996023778355113,
      "loss": 0.9379,
      "step": 70
    },
    {
      "epoch": 0.010606513295488498,
      "grad_norm": 0.20963691174983978,
      "learning_rate": 0.00019995890142282728,
      "loss": 0.6632,
      "step": 71
    },
    {
      "epoch": 0.01075590080669256,
      "grad_norm": 0.2161312997341156,
      "learning_rate": 0.00019995754297959882,
      "loss": 0.8932,
      "step": 72
    },
    {
      "epoch": 0.010905288317896623,
      "grad_norm": 0.17720285058021545,
      "learning_rate": 0.00019995616245416584,
      "loss": 0.5283,
      "step": 73
    },
    {
      "epoch": 0.011054675829100687,
      "grad_norm": 0.18077623844146729,
      "learning_rate": 0.0001999547598468334,
      "loss": 0.603,
      "step": 74
    },
    {
      "epoch": 0.01120406334030475,
      "grad_norm": 0.19094090163707733,
      "learning_rate": 0.00019995333515791125,
      "loss": 0.9712,
      "step": 75
    },
    {
      "epoch": 0.011353450851508814,
      "grad_norm": 0.2597976326942444,
      "learning_rate": 0.00019995188838771425,
      "loss": 0.7913,
      "step": 76
    },
    {
      "epoch": 0.011502838362712877,
      "grad_norm": 0.26235243678092957,
      "learning_rate": 0.00019995041953656194,
      "loss": 0.7578,
      "step": 77
    },
    {
      "epoch": 0.011652225873916941,
      "grad_norm": 0.1809563785791397,
      "learning_rate": 0.0001999489286047788,
      "loss": 0.9168,
      "step": 78
    },
    {
      "epoch": 0.011801613385121004,
      "grad_norm": 0.22458118200302124,
      "learning_rate": 0.0001999474155926942,
      "loss": 1.1456,
      "step": 79
    },
    {
      "epoch": 0.011951000896325068,
      "grad_norm": 0.49804338812828064,
      "learning_rate": 0.00019994588050064243,
      "loss": 1.1896,
      "step": 80
    },
    {
      "epoch": 0.01210038840752913,
      "grad_norm": 0.2247380167245865,
      "learning_rate": 0.00019994432332896258,
      "loss": 0.958,
      "step": 81
    },
    {
      "epoch": 0.012249775918733195,
      "grad_norm": 0.17649094760417938,
      "learning_rate": 0.00019994274407799872,
      "loss": 0.9417,
      "step": 82
    },
    {
      "epoch": 0.012399163429937257,
      "grad_norm": 0.24156750738620758,
      "learning_rate": 0.00019994114274809964,
      "loss": 0.9959,
      "step": 83
    },
    {
      "epoch": 0.01254855094114132,
      "grad_norm": 0.2841317355632782,
      "learning_rate": 0.00019993951933961913,
      "loss": 1.296,
      "step": 84
    },
    {
      "epoch": 0.012697938452345384,
      "grad_norm": 0.26792535185813904,
      "learning_rate": 0.00019993787385291588,
      "loss": 1.0296,
      "step": 85
    },
    {
      "epoch": 0.012847325963549447,
      "grad_norm": 0.11657895892858505,
      "learning_rate": 0.00019993620628835332,
      "loss": 0.6993,
      "step": 86
    },
    {
      "epoch": 0.01299671347475351,
      "grad_norm": 0.15463922917842865,
      "learning_rate": 0.0001999345166462999,
      "loss": 0.6925,
      "step": 87
    },
    {
      "epoch": 0.013146100985957573,
      "grad_norm": 0.1776818037033081,
      "learning_rate": 0.0001999328049271289,
      "loss": 0.7629,
      "step": 88
    },
    {
      "epoch": 0.013295488497161638,
      "grad_norm": 0.16185057163238525,
      "learning_rate": 0.00019993107113121844,
      "loss": 1.0736,
      "step": 89
    },
    {
      "epoch": 0.0134448760083657,
      "grad_norm": 0.3227018415927887,
      "learning_rate": 0.0001999293152589515,
      "loss": 0.8586,
      "step": 90
    },
    {
      "epoch": 0.013594263519569764,
      "grad_norm": 0.16486088931560516,
      "learning_rate": 0.00019992753731071602,
      "loss": 1.0614,
      "step": 91
    },
    {
      "epoch": 0.013743651030773827,
      "grad_norm": 0.1495029777288437,
      "learning_rate": 0.0001999257372869048,
      "loss": 0.8561,
      "step": 92
    },
    {
      "epoch": 0.013893038541977891,
      "grad_norm": 0.16044558584690094,
      "learning_rate": 0.00019992391518791546,
      "loss": 1.0294,
      "step": 93
    },
    {
      "epoch": 0.014042426053181954,
      "grad_norm": 0.16251197457313538,
      "learning_rate": 0.00019992207101415053,
      "loss": 0.838,
      "step": 94
    },
    {
      "epoch": 0.014191813564386018,
      "grad_norm": 0.47265177965164185,
      "learning_rate": 0.00019992020476601745,
      "loss": 1.4965,
      "step": 95
    },
    {
      "epoch": 0.01434120107559008,
      "grad_norm": 0.16032789647579193,
      "learning_rate": 0.00019991831644392848,
      "loss": 0.9541,
      "step": 96
    },
    {
      "epoch": 0.014490588586794143,
      "grad_norm": 0.1563224196434021,
      "learning_rate": 0.0001999164060483008,
      "loss": 0.6754,
      "step": 97
    },
    {
      "epoch": 0.014639976097998207,
      "grad_norm": 0.2469077855348587,
      "learning_rate": 0.00019991447357955639,
      "loss": 0.9911,
      "step": 98
    },
    {
      "epoch": 0.01478936360920227,
      "grad_norm": 0.13441061973571777,
      "learning_rate": 0.00019991251903812225,
      "loss": 0.7342,
      "step": 99
    },
    {
      "epoch": 0.014938751120406334,
      "grad_norm": 0.20193825662136078,
      "learning_rate": 0.00019991054242443008,
      "loss": 0.6834,
      "step": 100
    },
    {
      "epoch": 0.015088138631610397,
      "grad_norm": 0.1552024483680725,
      "learning_rate": 0.0001999085437389166,
      "loss": 0.8257,
      "step": 101
    },
    {
      "epoch": 0.015237526142814461,
      "grad_norm": 0.25607559084892273,
      "learning_rate": 0.00019990652298202335,
      "loss": 1.1036,
      "step": 102
    },
    {
      "epoch": 0.015386913654018524,
      "grad_norm": 0.11925558000802994,
      "learning_rate": 0.00019990448015419675,
      "loss": 0.6345,
      "step": 103
    },
    {
      "epoch": 0.015536301165222588,
      "grad_norm": 0.28443989157676697,
      "learning_rate": 0.00019990241525588804,
      "loss": 1.1217,
      "step": 104
    },
    {
      "epoch": 0.01568568867642665,
      "grad_norm": 0.15237168967723846,
      "learning_rate": 0.00019990032828755345,
      "loss": 0.7735,
      "step": 105
    },
    {
      "epoch": 0.015835076187630713,
      "grad_norm": 0.173554927110672,
      "learning_rate": 0.000199898219249654,
      "loss": 1.1093,
      "step": 106
    },
    {
      "epoch": 0.01598446369883478,
      "grad_norm": 0.14035388827323914,
      "learning_rate": 0.0001998960881426556,
      "loss": 0.8206,
      "step": 107
    },
    {
      "epoch": 0.01613385121003884,
      "grad_norm": 0.13699080049991608,
      "learning_rate": 0.00019989393496702907,
      "loss": 1.0426,
      "step": 108
    },
    {
      "epoch": 0.016283238721242904,
      "grad_norm": 0.1313813477754593,
      "learning_rate": 0.00019989175972325005,
      "loss": 0.7804,
      "step": 109
    },
    {
      "epoch": 0.016432626232446967,
      "grad_norm": 0.15459921956062317,
      "learning_rate": 0.00019988956241179912,
      "loss": 0.9949,
      "step": 110
    },
    {
      "epoch": 0.01658201374365103,
      "grad_norm": 0.20009522140026093,
      "learning_rate": 0.00019988734303316168,
      "loss": 1.0159,
      "step": 111
    },
    {
      "epoch": 0.016731401254855095,
      "grad_norm": 0.17909283936023712,
      "learning_rate": 0.00019988510158782804,
      "loss": 0.5281,
      "step": 112
    },
    {
      "epoch": 0.016880788766059158,
      "grad_norm": 0.6786065697669983,
      "learning_rate": 0.00019988283807629334,
      "loss": 1.2808,
      "step": 113
    },
    {
      "epoch": 0.01703017627726322,
      "grad_norm": 0.15410082042217255,
      "learning_rate": 0.00019988055249905767,
      "loss": 1.0755,
      "step": 114
    },
    {
      "epoch": 0.017179563788467283,
      "grad_norm": 0.19600564241409302,
      "learning_rate": 0.00019987824485662593,
      "loss": 1.2168,
      "step": 115
    },
    {
      "epoch": 0.01732895129967135,
      "grad_norm": 0.1903967261314392,
      "learning_rate": 0.00019987591514950787,
      "loss": 0.9987,
      "step": 116
    },
    {
      "epoch": 0.01747833881087541,
      "grad_norm": 0.13745927810668945,
      "learning_rate": 0.00019987356337821822,
      "loss": 0.9328,
      "step": 117
    },
    {
      "epoch": 0.017627726322079474,
      "grad_norm": 0.18400371074676514,
      "learning_rate": 0.00019987118954327654,
      "loss": 0.6686,
      "step": 118
    },
    {
      "epoch": 0.017777113833283537,
      "grad_norm": 0.29325294494628906,
      "learning_rate": 0.00019986879364520716,
      "loss": 1.173,
      "step": 119
    },
    {
      "epoch": 0.0179265013444876,
      "grad_norm": 0.15400448441505432,
      "learning_rate": 0.00019986637568453945,
      "loss": 0.897,
      "step": 120
    },
    {
      "epoch": 0.018075888855691665,
      "grad_norm": 0.2178712785243988,
      "learning_rate": 0.00019986393566180755,
      "loss": 1.2885,
      "step": 121
    },
    {
      "epoch": 0.018225276366895728,
      "grad_norm": 0.15217332541942596,
      "learning_rate": 0.00019986147357755048,
      "loss": 0.9847,
      "step": 122
    },
    {
      "epoch": 0.01837466387809979,
      "grad_norm": 0.35806745290756226,
      "learning_rate": 0.0001998589894323122,
      "loss": 1.0114,
      "step": 123
    },
    {
      "epoch": 0.018524051389303853,
      "grad_norm": 0.46978825330734253,
      "learning_rate": 0.00019985648322664145,
      "loss": 1.33,
      "step": 124
    },
    {
      "epoch": 0.01867343890050792,
      "grad_norm": 0.3191573917865753,
      "learning_rate": 0.00019985395496109192,
      "loss": 0.8691,
      "step": 125
    },
    {
      "epoch": 0.01882282641171198,
      "grad_norm": 0.6782920360565186,
      "learning_rate": 0.00019985140463622215,
      "loss": 1.7387,
      "step": 126
    },
    {
      "epoch": 0.018972213922916044,
      "grad_norm": 0.4556753933429718,
      "learning_rate": 0.0001998488322525955,
      "loss": 1.9508,
      "step": 127
    },
    {
      "epoch": 0.019121601434120106,
      "grad_norm": 0.1401677280664444,
      "learning_rate": 0.0001998462378107803,
      "loss": 0.8127,
      "step": 128
    },
    {
      "epoch": 0.019270988945324172,
      "grad_norm": 0.1187644675374031,
      "learning_rate": 0.00019984362131134968,
      "loss": 0.5421,
      "step": 129
    },
    {
      "epoch": 0.019420376456528235,
      "grad_norm": 0.17097711563110352,
      "learning_rate": 0.0001998409827548817,
      "loss": 0.7335,
      "step": 130
    },
    {
      "epoch": 0.019569763967732298,
      "grad_norm": 0.12986791133880615,
      "learning_rate": 0.00019983832214195917,
      "loss": 0.7151,
      "step": 131
    },
    {
      "epoch": 0.01971915147893636,
      "grad_norm": 0.14413940906524658,
      "learning_rate": 0.00019983563947316996,
      "loss": 0.6833,
      "step": 132
    },
    {
      "epoch": 0.019868538990140423,
      "grad_norm": 0.1424313336610794,
      "learning_rate": 0.00019983293474910667,
      "loss": 0.8199,
      "step": 133
    },
    {
      "epoch": 0.02001792650134449,
      "grad_norm": 0.11715007573366165,
      "learning_rate": 0.00019983020797036683,
      "loss": 0.6079,
      "step": 134
    },
    {
      "epoch": 0.02016731401254855,
      "grad_norm": 0.11654297262430191,
      "learning_rate": 0.00019982745913755282,
      "loss": 0.5302,
      "step": 135
    },
    {
      "epoch": 0.020316701523752614,
      "grad_norm": 0.1768021434545517,
      "learning_rate": 0.00019982468825127187,
      "loss": 0.5412,
      "step": 136
    },
    {
      "epoch": 0.020466089034956676,
      "grad_norm": 0.2620059549808502,
      "learning_rate": 0.00019982189531213618,
      "loss": 1.0321,
      "step": 137
    },
    {
      "epoch": 0.020615476546160742,
      "grad_norm": 0.1472969800233841,
      "learning_rate": 0.0001998190803207627,
      "loss": 1.0097,
      "step": 138
    },
    {
      "epoch": 0.020764864057364805,
      "grad_norm": 0.1889103651046753,
      "learning_rate": 0.00019981624327777332,
      "loss": 0.7423,
      "step": 139
    },
    {
      "epoch": 0.020914251568568867,
      "grad_norm": 0.20594674348831177,
      "learning_rate": 0.0001998133841837948,
      "loss": 1.3315,
      "step": 140
    },
    {
      "epoch": 0.02106363907977293,
      "grad_norm": 0.19783402979373932,
      "learning_rate": 0.00019981050303945877,
      "loss": 1.0435,
      "step": 141
    },
    {
      "epoch": 0.021213026590976996,
      "grad_norm": 0.12695230543613434,
      "learning_rate": 0.00019980759984540168,
      "loss": 0.8,
      "step": 142
    },
    {
      "epoch": 0.02136241410218106,
      "grad_norm": 0.18909701704978943,
      "learning_rate": 0.0001998046746022649,
      "loss": 0.7549,
      "step": 143
    },
    {
      "epoch": 0.02151180161338512,
      "grad_norm": 0.2600228488445282,
      "learning_rate": 0.0001998017273106947,
      "loss": 0.6657,
      "step": 144
    },
    {
      "epoch": 0.021661189124589184,
      "grad_norm": 0.2999661862850189,
      "learning_rate": 0.00019979875797134216,
      "loss": 0.9626,
      "step": 145
    },
    {
      "epoch": 0.021810576635793246,
      "grad_norm": 0.2698724567890167,
      "learning_rate": 0.00019979576658486325,
      "loss": 0.8856,
      "step": 146
    },
    {
      "epoch": 0.021959964146997312,
      "grad_norm": 0.14443442225456238,
      "learning_rate": 0.0001997927531519188,
      "loss": 0.9285,
      "step": 147
    },
    {
      "epoch": 0.022109351658201375,
      "grad_norm": 0.21873806416988373,
      "learning_rate": 0.00019978971767317457,
      "loss": 0.9565,
      "step": 148
    },
    {
      "epoch": 0.022258739169405437,
      "grad_norm": 0.2041054517030716,
      "learning_rate": 0.0001997866601493011,
      "loss": 0.9455,
      "step": 149
    },
    {
      "epoch": 0.0224081266806095,
      "grad_norm": 0.19350048899650574,
      "learning_rate": 0.00019978358058097388,
      "loss": 0.8047,
      "step": 150
    },
    {
      "epoch": 0.022557514191813566,
      "grad_norm": 0.25492972135543823,
      "learning_rate": 0.00019978047896887323,
      "loss": 1.234,
      "step": 151
    },
    {
      "epoch": 0.02270690170301763,
      "grad_norm": 0.1414925754070282,
      "learning_rate": 0.0001997773553136843,
      "loss": 0.9703,
      "step": 152
    },
    {
      "epoch": 0.02285628921422169,
      "grad_norm": 0.15996232628822327,
      "learning_rate": 0.00019977420961609721,
      "loss": 1.1475,
      "step": 153
    },
    {
      "epoch": 0.023005676725425753,
      "grad_norm": 0.1661243587732315,
      "learning_rate": 0.00019977104187680688,
      "loss": 0.8574,
      "step": 154
    },
    {
      "epoch": 0.02315506423662982,
      "grad_norm": 0.17797410488128662,
      "learning_rate": 0.00019976785209651309,
      "loss": 0.6742,
      "step": 155
    },
    {
      "epoch": 0.023304451747833882,
      "grad_norm": 0.1401919722557068,
      "learning_rate": 0.00019976464027592053,
      "loss": 0.7065,
      "step": 156
    },
    {
      "epoch": 0.023453839259037945,
      "grad_norm": 0.3743472695350647,
      "learning_rate": 0.00019976140641573875,
      "loss": 1.1391,
      "step": 157
    },
    {
      "epoch": 0.023603226770242007,
      "grad_norm": 0.12242946028709412,
      "learning_rate": 0.00019975815051668217,
      "loss": 0.7517,
      "step": 158
    },
    {
      "epoch": 0.02375261428144607,
      "grad_norm": 0.12851974368095398,
      "learning_rate": 0.00019975487257947004,
      "loss": 0.5795,
      "step": 159
    },
    {
      "epoch": 0.023902001792650136,
      "grad_norm": 0.15131576359272003,
      "learning_rate": 0.0001997515726048265,
      "loss": 0.5735,
      "step": 160
    },
    {
      "epoch": 0.024051389303854198,
      "grad_norm": 0.2606663703918457,
      "learning_rate": 0.00019974825059348062,
      "loss": 0.7677,
      "step": 161
    },
    {
      "epoch": 0.02420077681505826,
      "grad_norm": 0.3116661608219147,
      "learning_rate": 0.00019974490654616625,
      "loss": 1.1465,
      "step": 162
    },
    {
      "epoch": 0.024350164326262323,
      "grad_norm": 0.15296891331672668,
      "learning_rate": 0.00019974154046362212,
      "loss": 0.9154,
      "step": 163
    },
    {
      "epoch": 0.02449955183746639,
      "grad_norm": 0.12761889398097992,
      "learning_rate": 0.0001997381523465919,
      "loss": 0.7902,
      "step": 164
    },
    {
      "epoch": 0.024648939348670452,
      "grad_norm": 0.11566565185785294,
      "learning_rate": 0.00019973474219582405,
      "loss": 0.5558,
      "step": 165
    },
    {
      "epoch": 0.024798326859874514,
      "grad_norm": 0.12765946984291077,
      "learning_rate": 0.00019973131001207195,
      "loss": 0.7314,
      "step": 166
    },
    {
      "epoch": 0.024947714371078577,
      "grad_norm": 0.12568241357803345,
      "learning_rate": 0.00019972785579609376,
      "loss": 0.9476,
      "step": 167
    },
    {
      "epoch": 0.02509710188228264,
      "grad_norm": 0.21633781492710114,
      "learning_rate": 0.00019972437954865265,
      "loss": 1.3121,
      "step": 168
    },
    {
      "epoch": 0.025246489393486705,
      "grad_norm": 0.14913877844810486,
      "learning_rate": 0.00019972088127051657,
      "loss": 0.8247,
      "step": 169
    },
    {
      "epoch": 0.025395876904690768,
      "grad_norm": 0.16602329909801483,
      "learning_rate": 0.00019971736096245825,
      "loss": 1.064,
      "step": 170
    },
    {
      "epoch": 0.02554526441589483,
      "grad_norm": 0.28753912448883057,
      "learning_rate": 0.00019971381862525552,
      "loss": 1.4888,
      "step": 171
    },
    {
      "epoch": 0.025694651927098893,
      "grad_norm": 0.13800154626369476,
      "learning_rate": 0.00019971025425969083,
      "loss": 0.5945,
      "step": 172
    },
    {
      "epoch": 0.02584403943830296,
      "grad_norm": 0.22065165638923645,
      "learning_rate": 0.00019970666786655166,
      "loss": 0.8695,
      "step": 173
    },
    {
      "epoch": 0.02599342694950702,
      "grad_norm": 0.3128964900970459,
      "learning_rate": 0.0001997030594466303,
      "loss": 1.1673,
      "step": 174
    },
    {
      "epoch": 0.026142814460711084,
      "grad_norm": 0.20117323100566864,
      "learning_rate": 0.00019969942900072387,
      "loss": 1.0395,
      "step": 175
    },
    {
      "epoch": 0.026292201971915147,
      "grad_norm": 0.14214551448822021,
      "learning_rate": 0.00019969577652963444,
      "loss": 0.5135,
      "step": 176
    },
    {
      "epoch": 0.026441589483119213,
      "grad_norm": 0.20523157715797424,
      "learning_rate": 0.00019969210203416883,
      "loss": 1.0156,
      "step": 177
    },
    {
      "epoch": 0.026590976994323275,
      "grad_norm": 0.1385200172662735,
      "learning_rate": 0.0001996884055151389,
      "loss": 0.5202,
      "step": 178
    },
    {
      "epoch": 0.026740364505527338,
      "grad_norm": 0.14539465308189392,
      "learning_rate": 0.00019968468697336117,
      "loss": 0.6127,
      "step": 179
    },
    {
      "epoch": 0.0268897520167314,
      "grad_norm": 0.2886626720428467,
      "learning_rate": 0.00019968094640965717,
      "loss": 1.213,
      "step": 180
    },
    {
      "epoch": 0.027039139527935463,
      "grad_norm": 0.22695106267929077,
      "learning_rate": 0.00019967718382485323,
      "loss": 1.137,
      "step": 181
    },
    {
      "epoch": 0.02718852703913953,
      "grad_norm": 0.17426621913909912,
      "learning_rate": 0.00019967339921978062,
      "loss": 0.8978,
      "step": 182
    },
    {
      "epoch": 0.02733791455034359,
      "grad_norm": 0.2875833809375763,
      "learning_rate": 0.00019966959259527534,
      "loss": 0.8688,
      "step": 183
    },
    {
      "epoch": 0.027487302061547654,
      "grad_norm": 0.14763344824314117,
      "learning_rate": 0.00019966576395217837,
      "loss": 0.564,
      "step": 184
    },
    {
      "epoch": 0.027636689572751717,
      "grad_norm": 0.24988707900047302,
      "learning_rate": 0.00019966191329133555,
      "loss": 0.429,
      "step": 185
    },
    {
      "epoch": 0.027786077083955783,
      "grad_norm": 0.16043034195899963,
      "learning_rate": 0.0001996580406135975,
      "loss": 0.7251,
      "step": 186
    },
    {
      "epoch": 0.027935464595159845,
      "grad_norm": 0.2776719033718109,
      "learning_rate": 0.00019965414591981975,
      "loss": 1.3513,
      "step": 187
    },
    {
      "epoch": 0.028084852106363908,
      "grad_norm": 0.1749371886253357,
      "learning_rate": 0.00019965022921086275,
      "loss": 0.7383,
      "step": 188
    },
    {
      "epoch": 0.02823423961756797,
      "grad_norm": 0.2881135940551758,
      "learning_rate": 0.00019964629048759176,
      "loss": 1.0511,
      "step": 189
    },
    {
      "epoch": 0.028383627128772036,
      "grad_norm": 0.11646547168493271,
      "learning_rate": 0.00019964232975087687,
      "loss": 0.6575,
      "step": 190
    },
    {
      "epoch": 0.0285330146399761,
      "grad_norm": 0.1457265317440033,
      "learning_rate": 0.0001996383470015931,
      "loss": 0.7538,
      "step": 191
    },
    {
      "epoch": 0.02868240215118016,
      "grad_norm": 0.13400131464004517,
      "learning_rate": 0.00019963434224062025,
      "loss": 0.7773,
      "step": 192
    },
    {
      "epoch": 0.028831789662384224,
      "grad_norm": 0.23350481688976288,
      "learning_rate": 0.0001996303154688431,
      "loss": 1.2769,
      "step": 193
    },
    {
      "epoch": 0.028981177173588286,
      "grad_norm": 0.563207745552063,
      "learning_rate": 0.0001996262666871512,
      "loss": 1.5822,
      "step": 194
    },
    {
      "epoch": 0.029130564684792352,
      "grad_norm": 0.5377495884895325,
      "learning_rate": 0.00019962219589643898,
      "loss": 1.4911,
      "step": 195
    },
    {
      "epoch": 0.029279952195996415,
      "grad_norm": 0.15726317465305328,
      "learning_rate": 0.00019961810309760577,
      "loss": 0.7824,
      "step": 196
    },
    {
      "epoch": 0.029429339707200478,
      "grad_norm": 0.15040062367916107,
      "learning_rate": 0.00019961398829155568,
      "loss": 0.7541,
      "step": 197
    },
    {
      "epoch": 0.02957872721840454,
      "grad_norm": 0.12620890140533447,
      "learning_rate": 0.00019960985147919778,
      "loss": 0.932,
      "step": 198
    },
    {
      "epoch": 0.029728114729608606,
      "grad_norm": 0.15029945969581604,
      "learning_rate": 0.00019960569266144597,
      "loss": 0.9588,
      "step": 199
    },
    {
      "epoch": 0.02987750224081267,
      "grad_norm": 0.16770395636558533,
      "learning_rate": 0.00019960151183921897,
      "loss": 0.5964,
      "step": 200
    },
    {
      "epoch": 0.03002688975201673,
      "grad_norm": 0.12649193406105042,
      "learning_rate": 0.0001995973090134404,
      "loss": 0.8221,
      "step": 201
    },
    {
      "epoch": 0.030176277263220794,
      "grad_norm": 0.15091091394424438,
      "learning_rate": 0.00019959308418503877,
      "loss": 0.8811,
      "step": 202
    },
    {
      "epoch": 0.03032566477442486,
      "grad_norm": 0.12077690660953522,
      "learning_rate": 0.00019958883735494732,
      "loss": 0.6411,
      "step": 203
    },
    {
      "epoch": 0.030475052285628922,
      "grad_norm": 0.15071533620357513,
      "learning_rate": 0.00019958456852410433,
      "loss": 0.7742,
      "step": 204
    },
    {
      "epoch": 0.030624439796832985,
      "grad_norm": 0.1368575096130371,
      "learning_rate": 0.00019958027769345277,
      "loss": 1.1054,
      "step": 205
    },
    {
      "epoch": 0.030773827308037047,
      "grad_norm": 0.22218400239944458,
      "learning_rate": 0.0001995759648639406,
      "loss": 0.8127,
      "step": 206
    },
    {
      "epoch": 0.03092321481924111,
      "grad_norm": 0.12783220410346985,
      "learning_rate": 0.00019957163003652063,
      "loss": 0.6916,
      "step": 207
    },
    {
      "epoch": 0.031072602330445176,
      "grad_norm": 0.18626387417316437,
      "learning_rate": 0.00019956727321215044,
      "loss": 0.8217,
      "step": 208
    },
    {
      "epoch": 0.03122198984164924,
      "grad_norm": 0.12861424684524536,
      "learning_rate": 0.0001995628943917925,
      "loss": 1.0327,
      "step": 209
    },
    {
      "epoch": 0.0313713773528533,
      "grad_norm": 0.17585206031799316,
      "learning_rate": 0.00019955849357641424,
      "loss": 0.6836,
      "step": 210
    },
    {
      "epoch": 0.03152076486405737,
      "grad_norm": 0.1541229784488678,
      "learning_rate": 0.0001995540707669878,
      "loss": 1.1849,
      "step": 211
    },
    {
      "epoch": 0.031670152375261426,
      "grad_norm": 0.13117974996566772,
      "learning_rate": 0.00019954962596449024,
      "loss": 0.7779,
      "step": 212
    },
    {
      "epoch": 0.03181953988646549,
      "grad_norm": 0.3847130537033081,
      "learning_rate": 0.00019954515916990358,
      "loss": 1.4158,
      "step": 213
    },
    {
      "epoch": 0.03196892739766956,
      "grad_norm": 0.21019677817821503,
      "learning_rate": 0.0001995406703842145,
      "loss": 1.0807,
      "step": 214
    },
    {
      "epoch": 0.03211831490887362,
      "grad_norm": 0.10954124480485916,
      "learning_rate": 0.0001995361596084147,
      "loss": 0.7142,
      "step": 215
    },
    {
      "epoch": 0.03226770242007768,
      "grad_norm": 0.25598999857902527,
      "learning_rate": 0.0001995316268435007,
      "loss": 0.7528,
      "step": 216
    },
    {
      "epoch": 0.03241708993128174,
      "grad_norm": 0.16298946738243103,
      "learning_rate": 0.0001995270720904738,
      "loss": 0.6978,
      "step": 217
    },
    {
      "epoch": 0.03256647744248581,
      "grad_norm": 0.37723109126091003,
      "learning_rate": 0.00019952249535034025,
      "loss": 1.0603,
      "step": 218
    },
    {
      "epoch": 0.032715864953689874,
      "grad_norm": 0.1600012332201004,
      "learning_rate": 0.00019951789662411113,
      "loss": 0.7634,
      "step": 219
    },
    {
      "epoch": 0.03286525246489393,
      "grad_norm": 0.1618419587612152,
      "learning_rate": 0.00019951327591280236,
      "loss": 0.8068,
      "step": 220
    },
    {
      "epoch": 0.033014639976098,
      "grad_norm": 0.11850997805595398,
      "learning_rate": 0.00019950863321743475,
      "loss": 0.7883,
      "step": 221
    },
    {
      "epoch": 0.03316402748730206,
      "grad_norm": 0.18700255453586578,
      "learning_rate": 0.0001995039685390339,
      "loss": 0.5821,
      "step": 222
    },
    {
      "epoch": 0.033313414998506125,
      "grad_norm": 0.19264543056488037,
      "learning_rate": 0.00019949928187863036,
      "loss": 0.7578,
      "step": 223
    },
    {
      "epoch": 0.03346280250971019,
      "grad_norm": 0.23828521370887756,
      "learning_rate": 0.00019949457323725946,
      "loss": 1.1739,
      "step": 224
    },
    {
      "epoch": 0.03361219002091425,
      "grad_norm": 0.2178596705198288,
      "learning_rate": 0.0001994898426159614,
      "loss": 0.6075,
      "step": 225
    },
    {
      "epoch": 0.033761577532118316,
      "grad_norm": 0.13766655325889587,
      "learning_rate": 0.0001994850900157813,
      "loss": 0.6941,
      "step": 226
    },
    {
      "epoch": 0.03391096504332238,
      "grad_norm": 0.345782071352005,
      "learning_rate": 0.00019948031543776904,
      "loss": 1.0944,
      "step": 227
    },
    {
      "epoch": 0.03406035255452644,
      "grad_norm": 0.1483844369649887,
      "learning_rate": 0.0001994755188829794,
      "loss": 0.9752,
      "step": 228
    },
    {
      "epoch": 0.03420974006573051,
      "grad_norm": 0.16494938731193542,
      "learning_rate": 0.00019947070035247205,
      "loss": 0.7408,
      "step": 229
    },
    {
      "epoch": 0.034359127576934566,
      "grad_norm": 0.15683765709400177,
      "learning_rate": 0.00019946585984731142,
      "loss": 0.6498,
      "step": 230
    },
    {
      "epoch": 0.03450851508813863,
      "grad_norm": 0.17898918688297272,
      "learning_rate": 0.0001994609973685669,
      "loss": 0.7338,
      "step": 231
    },
    {
      "epoch": 0.0346579025993427,
      "grad_norm": 0.15439556539058685,
      "learning_rate": 0.00019945611291731274,
      "loss": 0.854,
      "step": 232
    },
    {
      "epoch": 0.03480729011054676,
      "grad_norm": 0.16742883622646332,
      "learning_rate": 0.0001994512064946279,
      "loss": 0.5851,
      "step": 233
    },
    {
      "epoch": 0.03495667762175082,
      "grad_norm": 0.17511384189128876,
      "learning_rate": 0.00019944627810159632,
      "loss": 0.569,
      "step": 234
    },
    {
      "epoch": 0.03510606513295488,
      "grad_norm": 0.26125359535217285,
      "learning_rate": 0.0001994413277393067,
      "loss": 0.862,
      "step": 235
    },
    {
      "epoch": 0.03525545264415895,
      "grad_norm": 0.1439584493637085,
      "learning_rate": 0.00019943635540885279,
      "loss": 1.1311,
      "step": 236
    },
    {
      "epoch": 0.035404840155363014,
      "grad_norm": 0.146185502409935,
      "learning_rate": 0.00019943136111133294,
      "loss": 0.9574,
      "step": 237
    },
    {
      "epoch": 0.03555422766656707,
      "grad_norm": 5.1866774559021,
      "learning_rate": 0.00019942634484785052,
      "loss": 3.2188,
      "step": 238
    },
    {
      "epoch": 0.03570361517777114,
      "grad_norm": 0.13537189364433289,
      "learning_rate": 0.00019942130661951372,
      "loss": 0.6154,
      "step": 239
    },
    {
      "epoch": 0.0358530026889752,
      "grad_norm": 0.13716812431812286,
      "learning_rate": 0.00019941624642743548,
      "loss": 0.7604,
      "step": 240
    },
    {
      "epoch": 0.036002390200179264,
      "grad_norm": 0.37018468976020813,
      "learning_rate": 0.0001994111642727338,
      "loss": 0.8598,
      "step": 241
    },
    {
      "epoch": 0.03615177771138333,
      "grad_norm": 0.1747826188802719,
      "learning_rate": 0.0001994060601565313,
      "loss": 0.6497,
      "step": 242
    },
    {
      "epoch": 0.03630116522258739,
      "grad_norm": 0.12225501239299774,
      "learning_rate": 0.0001994009340799556,
      "loss": 0.7278,
      "step": 243
    },
    {
      "epoch": 0.036450552733791455,
      "grad_norm": 0.17295411229133606,
      "learning_rate": 0.00019939578604413912,
      "loss": 0.8747,
      "step": 244
    },
    {
      "epoch": 0.03659994024499552,
      "grad_norm": 0.1815291941165924,
      "learning_rate": 0.00019939061605021917,
      "loss": 0.7242,
      "step": 245
    },
    {
      "epoch": 0.03674932775619958,
      "grad_norm": 0.25227025151252747,
      "learning_rate": 0.00019938542409933787,
      "loss": 0.7818,
      "step": 246
    },
    {
      "epoch": 0.036898715267403646,
      "grad_norm": 0.280819296836853,
      "learning_rate": 0.00019938021019264221,
      "loss": 0.7471,
      "step": 247
    },
    {
      "epoch": 0.037048102778607706,
      "grad_norm": 0.1746496856212616,
      "learning_rate": 0.000199374974331284,
      "loss": 0.9149,
      "step": 248
    },
    {
      "epoch": 0.03719749028981177,
      "grad_norm": 0.2506274878978729,
      "learning_rate": 0.00019936971651641995,
      "loss": 0.8666,
      "step": 249
    },
    {
      "epoch": 0.03734687780101584,
      "grad_norm": 0.12307952344417572,
      "learning_rate": 0.00019936443674921158,
      "loss": 0.7737,
      "step": 250
    },
    {
      "epoch": 0.0374962653122199,
      "grad_norm": 0.13615377247333527,
      "learning_rate": 0.0001993591350308253,
      "loss": 0.7592,
      "step": 251
    },
    {
      "epoch": 0.03764565282342396,
      "grad_norm": 0.16808447241783142,
      "learning_rate": 0.0001993538113624323,
      "loss": 0.8599,
      "step": 252
    },
    {
      "epoch": 0.03779504033462802,
      "grad_norm": 0.11544547975063324,
      "learning_rate": 0.00019934846574520872,
      "loss": 0.7348,
      "step": 253
    },
    {
      "epoch": 0.03794442784583209,
      "grad_norm": 0.20908010005950928,
      "learning_rate": 0.00019934309818033544,
      "loss": 0.6674,
      "step": 254
    },
    {
      "epoch": 0.038093815357036154,
      "grad_norm": 0.1379510909318924,
      "learning_rate": 0.00019933770866899825,
      "loss": 0.7295,
      "step": 255
    },
    {
      "epoch": 0.03824320286824021,
      "grad_norm": 0.15058402717113495,
      "learning_rate": 0.0001993322972123878,
      "loss": 1.1005,
      "step": 256
    },
    {
      "epoch": 0.03839259037944428,
      "grad_norm": 0.1941765695810318,
      "learning_rate": 0.00019932686381169955,
      "loss": 0.7658,
      "step": 257
    },
    {
      "epoch": 0.038541977890648345,
      "grad_norm": 0.1889600306749344,
      "learning_rate": 0.0001993214084681338,
      "loss": 0.9765,
      "step": 258
    },
    {
      "epoch": 0.038691365401852404,
      "grad_norm": 0.1466747522354126,
      "learning_rate": 0.00019931593118289578,
      "loss": 0.5899,
      "step": 259
    },
    {
      "epoch": 0.03884075291305647,
      "grad_norm": 0.2811049520969391,
      "learning_rate": 0.00019931043195719548,
      "loss": 0.8255,
      "step": 260
    },
    {
      "epoch": 0.03899014042426053,
      "grad_norm": 0.1345093995332718,
      "learning_rate": 0.00019930491079224772,
      "loss": 0.9015,
      "step": 261
    },
    {
      "epoch": 0.039139527935464595,
      "grad_norm": 0.9843081831932068,
      "learning_rate": 0.00019929936768927232,
      "loss": 1.5562,
      "step": 262
    },
    {
      "epoch": 0.03928891544666866,
      "grad_norm": 0.18343736231327057,
      "learning_rate": 0.00019929380264949376,
      "loss": 0.8783,
      "step": 263
    },
    {
      "epoch": 0.03943830295787272,
      "grad_norm": 0.29955071210861206,
      "learning_rate": 0.00019928821567414144,
      "loss": 1.0868,
      "step": 264
    },
    {
      "epoch": 0.039587690469076786,
      "grad_norm": 0.16872237622737885,
      "learning_rate": 0.00019928260676444965,
      "loss": 0.7375,
      "step": 265
    },
    {
      "epoch": 0.039737077980280845,
      "grad_norm": 0.1343865543603897,
      "learning_rate": 0.00019927697592165747,
      "loss": 1.0279,
      "step": 266
    },
    {
      "epoch": 0.03988646549148491,
      "grad_norm": 0.2587420642375946,
      "learning_rate": 0.00019927132314700885,
      "loss": 0.8529,
      "step": 267
    },
    {
      "epoch": 0.04003585300268898,
      "grad_norm": 0.23096689581871033,
      "learning_rate": 0.00019926564844175256,
      "loss": 0.8726,
      "step": 268
    },
    {
      "epoch": 0.040185240513893036,
      "grad_norm": 0.13639822602272034,
      "learning_rate": 0.00019925995180714224,
      "loss": 0.6957,
      "step": 269
    },
    {
      "epoch": 0.0403346280250971,
      "grad_norm": 0.13751177489757538,
      "learning_rate": 0.00019925423324443638,
      "loss": 0.7239,
      "step": 270
    },
    {
      "epoch": 0.04048401553630117,
      "grad_norm": 0.12963712215423584,
      "learning_rate": 0.0001992484927548983,
      "loss": 0.686,
      "step": 271
    },
    {
      "epoch": 0.04063340304750523,
      "grad_norm": 0.13808754086494446,
      "learning_rate": 0.00019924273033979613,
      "loss": 0.973,
      "step": 272
    },
    {
      "epoch": 0.04078279055870929,
      "grad_norm": 0.10808500647544861,
      "learning_rate": 0.0001992369460004029,
      "loss": 0.6725,
      "step": 273
    },
    {
      "epoch": 0.04093217806991335,
      "grad_norm": 0.14959432184696198,
      "learning_rate": 0.0001992311397379965,
      "loss": 0.9753,
      "step": 274
    },
    {
      "epoch": 0.04108156558111742,
      "grad_norm": 0.21491862833499908,
      "learning_rate": 0.00019922531155385954,
      "loss": 0.9506,
      "step": 275
    },
    {
      "epoch": 0.041230953092321485,
      "grad_norm": 0.16956715285778046,
      "learning_rate": 0.00019921946144927966,
      "loss": 1.0777,
      "step": 276
    },
    {
      "epoch": 0.041380340603525544,
      "grad_norm": 0.17658768594264984,
      "learning_rate": 0.00019921358942554917,
      "loss": 0.6994,
      "step": 277
    },
    {
      "epoch": 0.04152972811472961,
      "grad_norm": 0.20894332230091095,
      "learning_rate": 0.0001992076954839653,
      "loss": 1.0217,
      "step": 278
    },
    {
      "epoch": 0.04167911562593367,
      "grad_norm": 0.15689632296562195,
      "learning_rate": 0.00019920177962583015,
      "loss": 0.4844,
      "step": 279
    },
    {
      "epoch": 0.041828503137137735,
      "grad_norm": 0.24510236084461212,
      "learning_rate": 0.00019919584185245062,
      "loss": 0.8981,
      "step": 280
    },
    {
      "epoch": 0.0419778906483418,
      "grad_norm": 0.1385307013988495,
      "learning_rate": 0.00019918988216513844,
      "loss": 0.9278,
      "step": 281
    },
    {
      "epoch": 0.04212727815954586,
      "grad_norm": 0.15266434848308563,
      "learning_rate": 0.00019918390056521018,
      "loss": 0.7803,
      "step": 282
    },
    {
      "epoch": 0.042276665670749926,
      "grad_norm": 0.920637309551239,
      "learning_rate": 0.00019917789705398728,
      "loss": 2.0334,
      "step": 283
    },
    {
      "epoch": 0.04242605318195399,
      "grad_norm": 0.16596724092960358,
      "learning_rate": 0.00019917187163279605,
      "loss": 0.5195,
      "step": 284
    },
    {
      "epoch": 0.04257544069315805,
      "grad_norm": 0.21556362509727478,
      "learning_rate": 0.00019916582430296758,
      "loss": 1.0858,
      "step": 285
    },
    {
      "epoch": 0.04272482820436212,
      "grad_norm": 0.13357070088386536,
      "learning_rate": 0.00019915975506583778,
      "loss": 0.7042,
      "step": 286
    },
    {
      "epoch": 0.042874215715566176,
      "grad_norm": 0.13493870198726654,
      "learning_rate": 0.00019915366392274752,
      "loss": 0.7329,
      "step": 287
    },
    {
      "epoch": 0.04302360322677024,
      "grad_norm": 0.13088954985141754,
      "learning_rate": 0.00019914755087504236,
      "loss": 0.6911,
      "step": 288
    },
    {
      "epoch": 0.04317299073797431,
      "grad_norm": 0.12659801542758942,
      "learning_rate": 0.0001991414159240728,
      "loss": 0.9972,
      "step": 289
    },
    {
      "epoch": 0.04332237824917837,
      "grad_norm": 0.14888939261436462,
      "learning_rate": 0.00019913525907119418,
      "loss": 0.917,
      "step": 290
    },
    {
      "epoch": 0.04347176576038243,
      "grad_norm": 0.15505638718605042,
      "learning_rate": 0.00019912908031776655,
      "loss": 0.5189,
      "step": 291
    },
    {
      "epoch": 0.04362115327158649,
      "grad_norm": 0.1389274001121521,
      "learning_rate": 0.000199122879665155,
      "loss": 0.9571,
      "step": 292
    },
    {
      "epoch": 0.04377054078279056,
      "grad_norm": 0.5212099552154541,
      "learning_rate": 0.0001991166571147293,
      "loss": 1.7933,
      "step": 293
    },
    {
      "epoch": 0.043919928293994624,
      "grad_norm": 0.15124285221099854,
      "learning_rate": 0.0001991104126678641,
      "loss": 0.8451,
      "step": 294
    },
    {
      "epoch": 0.04406931580519868,
      "grad_norm": 0.12665098905563354,
      "learning_rate": 0.0001991041463259389,
      "loss": 0.8399,
      "step": 295
    },
    {
      "epoch": 0.04421870331640275,
      "grad_norm": 0.41863369941711426,
      "learning_rate": 0.00019909785809033806,
      "loss": 0.6476,
      "step": 296
    },
    {
      "epoch": 0.044368090827606815,
      "grad_norm": 0.12471672147512436,
      "learning_rate": 0.00019909154796245076,
      "loss": 0.8806,
      "step": 297
    },
    {
      "epoch": 0.044517478338810874,
      "grad_norm": 0.21696336567401886,
      "learning_rate": 0.00019908521594367098,
      "loss": 1.0237,
      "step": 298
    },
    {
      "epoch": 0.04466686585001494,
      "grad_norm": 0.10489077866077423,
      "learning_rate": 0.00019907886203539757,
      "loss": 0.5955,
      "step": 299
    },
    {
      "epoch": 0.044816253361219,
      "grad_norm": 0.17574696242809296,
      "learning_rate": 0.0001990724862390342,
      "loss": 0.669,
      "step": 300
    },
    {
      "epoch": 0.044965640872423066,
      "grad_norm": 0.16877828538417816,
      "learning_rate": 0.00019906608855598939,
      "loss": 0.8991,
      "step": 301
    },
    {
      "epoch": 0.04511502838362713,
      "grad_norm": 0.16386249661445618,
      "learning_rate": 0.0001990596689876765,
      "loss": 0.9378,
      "step": 302
    },
    {
      "epoch": 0.04526441589483119,
      "grad_norm": 0.1148744598031044,
      "learning_rate": 0.00019905322753551368,
      "loss": 0.5595,
      "step": 303
    },
    {
      "epoch": 0.04541380340603526,
      "grad_norm": 0.1777895838022232,
      "learning_rate": 0.00019904676420092404,
      "loss": 0.9161,
      "step": 304
    },
    {
      "epoch": 0.045563190917239316,
      "grad_norm": 0.14002953469753265,
      "learning_rate": 0.0001990402789853353,
      "loss": 0.6209,
      "step": 305
    },
    {
      "epoch": 0.04571257842844338,
      "grad_norm": 0.3308524489402771,
      "learning_rate": 0.00019903377189018024,
      "loss": 0.9947,
      "step": 306
    },
    {
      "epoch": 0.04586196593964745,
      "grad_norm": 0.16171899437904358,
      "learning_rate": 0.00019902724291689637,
      "loss": 0.9893,
      "step": 307
    },
    {
      "epoch": 0.04601135345085151,
      "grad_norm": 0.1441776603460312,
      "learning_rate": 0.000199020692066926,
      "loss": 0.8799,
      "step": 308
    },
    {
      "epoch": 0.04616074096205557,
      "grad_norm": 0.12858013808727264,
      "learning_rate": 0.00019901411934171638,
      "loss": 0.6444,
      "step": 309
    },
    {
      "epoch": 0.04631012847325964,
      "grad_norm": 0.14368069171905518,
      "learning_rate": 0.00019900752474271945,
      "loss": 0.4694,
      "step": 310
    },
    {
      "epoch": 0.0464595159844637,
      "grad_norm": 0.16599714756011963,
      "learning_rate": 0.00019900090827139214,
      "loss": 0.7362,
      "step": 311
    },
    {
      "epoch": 0.046608903495667764,
      "grad_norm": 0.18922938406467438,
      "learning_rate": 0.0001989942699291961,
      "loss": 0.9204,
      "step": 312
    },
    {
      "epoch": 0.04675829100687182,
      "grad_norm": 0.17513027787208557,
      "learning_rate": 0.00019898760971759783,
      "loss": 0.5341,
      "step": 313
    },
    {
      "epoch": 0.04690767851807589,
      "grad_norm": 0.19131267070770264,
      "learning_rate": 0.0001989809276380687,
      "loss": 1.0719,
      "step": 314
    },
    {
      "epoch": 0.047057066029279955,
      "grad_norm": 0.1646454632282257,
      "learning_rate": 0.00019897422369208488,
      "loss": 0.9124,
      "step": 315
    },
    {
      "epoch": 0.047206453540484014,
      "grad_norm": 0.17317859828472137,
      "learning_rate": 0.00019896749788112737,
      "loss": 0.7388,
      "step": 316
    },
    {
      "epoch": 0.04735584105168808,
      "grad_norm": 0.2774486839771271,
      "learning_rate": 0.00019896075020668202,
      "loss": 0.8183,
      "step": 317
    },
    {
      "epoch": 0.04750522856289214,
      "grad_norm": 0.14470195770263672,
      "learning_rate": 0.0001989539806702395,
      "loss": 0.8357,
      "step": 318
    },
    {
      "epoch": 0.047654616074096205,
      "grad_norm": 0.1452193409204483,
      "learning_rate": 0.00019894718927329524,
      "loss": 0.9126,
      "step": 319
    },
    {
      "epoch": 0.04780400358530027,
      "grad_norm": 0.15406997501850128,
      "learning_rate": 0.0001989403760173497,
      "loss": 0.6407,
      "step": 320
    },
    {
      "epoch": 0.04795339109650433,
      "grad_norm": 0.2753758728504181,
      "learning_rate": 0.00019893354090390791,
      "loss": 1.1197,
      "step": 321
    },
    {
      "epoch": 0.048102778607708396,
      "grad_norm": 0.14286638796329498,
      "learning_rate": 0.00019892668393447997,
      "loss": 0.7519,
      "step": 322
    },
    {
      "epoch": 0.04825216611891246,
      "grad_norm": 0.27783840894699097,
      "learning_rate": 0.0001989198051105806,
      "loss": 0.8173,
      "step": 323
    },
    {
      "epoch": 0.04840155363011652,
      "grad_norm": 0.14072780311107635,
      "learning_rate": 0.00019891290443372944,
      "loss": 0.7387,
      "step": 324
    },
    {
      "epoch": 0.04855094114132059,
      "grad_norm": 0.17845518887043,
      "learning_rate": 0.00019890598190545102,
      "loss": 0.9028,
      "step": 325
    },
    {
      "epoch": 0.048700328652524646,
      "grad_norm": 0.16201937198638916,
      "learning_rate": 0.0001988990375272746,
      "loss": 0.6775,
      "step": 326
    },
    {
      "epoch": 0.04884971616372871,
      "grad_norm": 0.3842596709728241,
      "learning_rate": 0.00019889207130073432,
      "loss": 1.166,
      "step": 327
    },
    {
      "epoch": 0.04899910367493278,
      "grad_norm": 0.1811336725950241,
      "learning_rate": 0.0001988850832273691,
      "loss": 0.5225,
      "step": 328
    },
    {
      "epoch": 0.04914849118613684,
      "grad_norm": 0.1348968744277954,
      "learning_rate": 0.0001988780733087228,
      "loss": 0.6588,
      "step": 329
    },
    {
      "epoch": 0.049297878697340904,
      "grad_norm": 0.15417031943798065,
      "learning_rate": 0.0001988710415463439,
      "loss": 1.0178,
      "step": 330
    },
    {
      "epoch": 0.04944726620854496,
      "grad_norm": 0.1732081174850464,
      "learning_rate": 0.0001988639879417859,
      "loss": 0.9023,
      "step": 331
    },
    {
      "epoch": 0.04959665371974903,
      "grad_norm": 0.20192794501781464,
      "learning_rate": 0.00019885691249660702,
      "loss": 0.9105,
      "step": 332
    },
    {
      "epoch": 0.049746041230953095,
      "grad_norm": 0.17755641043186188,
      "learning_rate": 0.0001988498152123704,
      "loss": 0.9222,
      "step": 333
    },
    {
      "epoch": 0.049895428742157154,
      "grad_norm": 0.10902781784534454,
      "learning_rate": 0.00019884269609064386,
      "loss": 0.6073,
      "step": 334
    },
    {
      "epoch": 0.05004481625336122,
      "grad_norm": 0.19006772339344025,
      "learning_rate": 0.00019883555513300019,
      "loss": 1.1671,
      "step": 335
    },
    {
      "epoch": 0.05019420376456528,
      "grad_norm": 0.1430109441280365,
      "learning_rate": 0.0001988283923410169,
      "loss": 0.9732,
      "step": 336
    },
    {
      "epoch": 0.050343591275769345,
      "grad_norm": 0.2234259843826294,
      "learning_rate": 0.00019882120771627638,
      "loss": 1.3184,
      "step": 337
    },
    {
      "epoch": 0.05049297878697341,
      "grad_norm": 0.1166071966290474,
      "learning_rate": 0.00019881400126036582,
      "loss": 0.6163,
      "step": 338
    },
    {
      "epoch": 0.05064236629817747,
      "grad_norm": 0.12911243736743927,
      "learning_rate": 0.0001988067729748773,
      "loss": 0.9934,
      "step": 339
    },
    {
      "epoch": 0.050791753809381536,
      "grad_norm": 0.14424997568130493,
      "learning_rate": 0.00019879952286140754,
      "loss": 0.6436,
      "step": 340
    },
    {
      "epoch": 0.0509411413205856,
      "grad_norm": 0.17660249769687653,
      "learning_rate": 0.00019879225092155834,
      "loss": 0.8772,
      "step": 341
    },
    {
      "epoch": 0.05109052883178966,
      "grad_norm": 0.126225546002388,
      "learning_rate": 0.0001987849571569361,
      "loss": 0.5238,
      "step": 342
    },
    {
      "epoch": 0.05123991634299373,
      "grad_norm": 0.1732265204191208,
      "learning_rate": 0.00019877764156915213,
      "loss": 0.7972,
      "step": 343
    },
    {
      "epoch": 0.051389303854197786,
      "grad_norm": 0.2713650166988373,
      "learning_rate": 0.0001987703041598226,
      "loss": 0.9757,
      "step": 344
    },
    {
      "epoch": 0.05153869136540185,
      "grad_norm": 0.1454247683286667,
      "learning_rate": 0.00019876294493056845,
      "loss": 0.9633,
      "step": 345
    },
    {
      "epoch": 0.05168807887660592,
      "grad_norm": 0.12023693323135376,
      "learning_rate": 0.00019875556388301543,
      "loss": 0.5894,
      "step": 346
    },
    {
      "epoch": 0.05183746638780998,
      "grad_norm": 0.15992896258831024,
      "learning_rate": 0.0001987481610187941,
      "loss": 0.7281,
      "step": 347
    },
    {
      "epoch": 0.05198685389901404,
      "grad_norm": 0.15995372831821442,
      "learning_rate": 0.00019874073633953997,
      "loss": 0.6077,
      "step": 348
    },
    {
      "epoch": 0.0521362414102181,
      "grad_norm": 0.12626518309116364,
      "learning_rate": 0.0001987332898468932,
      "loss": 0.8621,
      "step": 349
    },
    {
      "epoch": 0.05228562892142217,
      "grad_norm": 0.1364617496728897,
      "learning_rate": 0.00019872582154249884,
      "loss": 0.7432,
      "step": 350
    },
    {
      "epoch": 0.052435016432626234,
      "grad_norm": 0.17311328649520874,
      "learning_rate": 0.00019871833142800675,
      "loss": 0.7599,
      "step": 351
    },
    {
      "epoch": 0.052584403943830293,
      "grad_norm": 0.18369294703006744,
      "learning_rate": 0.00019871081950507163,
      "loss": 0.7234,
      "step": 352
    },
    {
      "epoch": 0.05273379145503436,
      "grad_norm": 0.12449700385332108,
      "learning_rate": 0.00019870328577535303,
      "loss": 0.8287,
      "step": 353
    },
    {
      "epoch": 0.052883178966238426,
      "grad_norm": 0.10272317379713058,
      "learning_rate": 0.00019869573024051517,
      "loss": 0.5652,
      "step": 354
    },
    {
      "epoch": 0.053032566477442485,
      "grad_norm": 0.16783328354358673,
      "learning_rate": 0.00019868815290222726,
      "loss": 0.5231,
      "step": 355
    },
    {
      "epoch": 0.05318195398864655,
      "grad_norm": 0.26311105489730835,
      "learning_rate": 0.00019868055376216323,
      "loss": 0.6964,
      "step": 356
    },
    {
      "epoch": 0.05333134149985061,
      "grad_norm": 0.32151147723197937,
      "learning_rate": 0.00019867293282200188,
      "loss": 1.0218,
      "step": 357
    },
    {
      "epoch": 0.053480729011054676,
      "grad_norm": 0.23724707961082458,
      "learning_rate": 0.00019866529008342673,
      "loss": 1.0303,
      "step": 358
    },
    {
      "epoch": 0.05363011652225874,
      "grad_norm": 0.1722903996706009,
      "learning_rate": 0.00019865762554812624,
      "loss": 0.9047,
      "step": 359
    },
    {
      "epoch": 0.0537795040334628,
      "grad_norm": 0.20011122524738312,
      "learning_rate": 0.00019864993921779361,
      "loss": 0.8151,
      "step": 360
    },
    {
      "epoch": 0.05392889154466687,
      "grad_norm": 0.13171933591365814,
      "learning_rate": 0.0001986422310941269,
      "loss": 0.7425,
      "step": 361
    },
    {
      "epoch": 0.054078279055870926,
      "grad_norm": 0.14532333612442017,
      "learning_rate": 0.0001986345011788289,
      "loss": 0.8868,
      "step": 362
    },
    {
      "epoch": 0.05422766656707499,
      "grad_norm": 0.17356187105178833,
      "learning_rate": 0.00019862674947360729,
      "loss": 0.4887,
      "step": 363
    },
    {
      "epoch": 0.05437705407827906,
      "grad_norm": 0.24108925461769104,
      "learning_rate": 0.00019861897598017457,
      "loss": 1.1633,
      "step": 364
    },
    {
      "epoch": 0.05452644158948312,
      "grad_norm": 0.16384924948215485,
      "learning_rate": 0.00019861118070024802,
      "loss": 0.8454,
      "step": 365
    },
    {
      "epoch": 0.05467582910068718,
      "grad_norm": 0.1604813188314438,
      "learning_rate": 0.00019860336363554973,
      "loss": 0.6332,
      "step": 366
    },
    {
      "epoch": 0.05482521661189125,
      "grad_norm": 0.19130012392997742,
      "learning_rate": 0.00019859552478780659,
      "loss": 0.9221,
      "step": 367
    },
    {
      "epoch": 0.05497460412309531,
      "grad_norm": 0.22768980264663696,
      "learning_rate": 0.0001985876641587504,
      "loss": 0.9983,
      "step": 368
    },
    {
      "epoch": 0.055123991634299374,
      "grad_norm": 0.13694654405117035,
      "learning_rate": 0.0001985797817501176,
      "loss": 0.633,
      "step": 369
    },
    {
      "epoch": 0.05527337914550343,
      "grad_norm": 0.1287558525800705,
      "learning_rate": 0.00019857187756364958,
      "loss": 0.6729,
      "step": 370
    },
    {
      "epoch": 0.0554227666567075,
      "grad_norm": 0.12864962220191956,
      "learning_rate": 0.00019856395160109256,
      "loss": 0.8324,
      "step": 371
    },
    {
      "epoch": 0.055572154167911565,
      "grad_norm": 0.15732339024543762,
      "learning_rate": 0.00019855600386419744,
      "loss": 0.7524,
      "step": 372
    },
    {
      "epoch": 0.055721541679115624,
      "grad_norm": 0.19376401603221893,
      "learning_rate": 0.00019854803435472,
      "loss": 0.9524,
      "step": 373
    },
    {
      "epoch": 0.05587092919031969,
      "grad_norm": 0.17447201907634735,
      "learning_rate": 0.00019854004307442088,
      "loss": 0.9536,
      "step": 374
    },
    {
      "epoch": 0.05602031670152375,
      "grad_norm": 0.1653999537229538,
      "learning_rate": 0.00019853203002506543,
      "loss": 0.6763,
      "step": 375
    },
    {
      "epoch": 0.056169704212727815,
      "grad_norm": 0.21575714647769928,
      "learning_rate": 0.0001985239952084239,
      "loss": 1.2311,
      "step": 376
    },
    {
      "epoch": 0.05631909172393188,
      "grad_norm": 0.18763162195682526,
      "learning_rate": 0.0001985159386262713,
      "loss": 0.7904,
      "step": 377
    },
    {
      "epoch": 0.05646847923513594,
      "grad_norm": 0.14876295626163483,
      "learning_rate": 0.0001985078602803874,
      "loss": 0.7818,
      "step": 378
    },
    {
      "epoch": 0.056617866746340006,
      "grad_norm": 0.15621663630008698,
      "learning_rate": 0.0001984997601725569,
      "loss": 0.8913,
      "step": 379
    },
    {
      "epoch": 0.05676725425754407,
      "grad_norm": 0.154473677277565,
      "learning_rate": 0.00019849163830456922,
      "loss": 0.5824,
      "step": 380
    },
    {
      "epoch": 0.05691664176874813,
      "grad_norm": 0.17189285159111023,
      "learning_rate": 0.00019848349467821864,
      "loss": 0.7574,
      "step": 381
    },
    {
      "epoch": 0.0570660292799522,
      "grad_norm": 0.17504605650901794,
      "learning_rate": 0.00019847532929530415,
      "loss": 0.746,
      "step": 382
    },
    {
      "epoch": 0.05721541679115626,
      "grad_norm": 0.14172236621379852,
      "learning_rate": 0.00019846714215762966,
      "loss": 0.9479,
      "step": 383
    },
    {
      "epoch": 0.05736480430236032,
      "grad_norm": 0.18013326823711395,
      "learning_rate": 0.00019845893326700384,
      "loss": 0.7438,
      "step": 384
    },
    {
      "epoch": 0.05751419181356439,
      "grad_norm": 0.12909965217113495,
      "learning_rate": 0.00019845070262524016,
      "loss": 0.6857,
      "step": 385
    },
    {
      "epoch": 0.05766357932476845,
      "grad_norm": 0.15466244518756866,
      "learning_rate": 0.00019844245023415685,
      "loss": 0.9448,
      "step": 386
    },
    {
      "epoch": 0.057812966835972514,
      "grad_norm": 0.6459704637527466,
      "learning_rate": 0.0001984341760955771,
      "loss": 1.7624,
      "step": 387
    },
    {
      "epoch": 0.05796235434717657,
      "grad_norm": 0.12828199565410614,
      "learning_rate": 0.0001984258802113287,
      "loss": 0.6321,
      "step": 388
    },
    {
      "epoch": 0.05811174185838064,
      "grad_norm": 0.19720108807086945,
      "learning_rate": 0.0001984175625832444,
      "loss": 0.742,
      "step": 389
    },
    {
      "epoch": 0.058261129369584705,
      "grad_norm": 0.1498088240623474,
      "learning_rate": 0.0001984092232131616,
      "loss": 0.8326,
      "step": 390
    },
    {
      "epoch": 0.058410516880788764,
      "grad_norm": 0.21480423212051392,
      "learning_rate": 0.0001984008621029227,
      "loss": 0.7597,
      "step": 391
    },
    {
      "epoch": 0.05855990439199283,
      "grad_norm": 0.29573026299476624,
      "learning_rate": 0.0001983924792543748,
      "loss": 1.3511,
      "step": 392
    },
    {
      "epoch": 0.058709291903196896,
      "grad_norm": 0.1259469836950302,
      "learning_rate": 0.0001983840746693698,
      "loss": 0.7294,
      "step": 393
    },
    {
      "epoch": 0.058858679414400955,
      "grad_norm": 0.14026756584644318,
      "learning_rate": 0.00019837564834976432,
      "loss": 0.6073,
      "step": 394
    },
    {
      "epoch": 0.05900806692560502,
      "grad_norm": 0.1396723836660385,
      "learning_rate": 0.00019836720029741995,
      "loss": 0.532,
      "step": 395
    },
    {
      "epoch": 0.05915745443680908,
      "grad_norm": 0.33731967210769653,
      "learning_rate": 0.000198358730514203,
      "loss": 1.0597,
      "step": 396
    },
    {
      "epoch": 0.059306841948013146,
      "grad_norm": 0.14965583384037018,
      "learning_rate": 0.00019835023900198454,
      "loss": 0.9816,
      "step": 397
    },
    {
      "epoch": 0.05945622945921721,
      "grad_norm": 0.18267478048801422,
      "learning_rate": 0.0001983417257626405,
      "loss": 0.7945,
      "step": 398
    },
    {
      "epoch": 0.05960561697042127,
      "grad_norm": 0.39048805832862854,
      "learning_rate": 0.0001983331907980516,
      "loss": 1.1221,
      "step": 399
    },
    {
      "epoch": 0.05975500448162534,
      "grad_norm": 0.11438261717557907,
      "learning_rate": 0.00019832463411010331,
      "loss": 0.6491,
      "step": 400
    },
    {
      "epoch": 0.059904391992829396,
      "grad_norm": 0.1477927714586258,
      "learning_rate": 0.00019831605570068596,
      "loss": 0.8062,
      "step": 401
    },
    {
      "epoch": 0.06005377950403346,
      "grad_norm": 0.12166056036949158,
      "learning_rate": 0.0001983074555716947,
      "loss": 0.8609,
      "step": 402
    },
    {
      "epoch": 0.06020316701523753,
      "grad_norm": 0.13299763202667236,
      "learning_rate": 0.00019829883372502935,
      "loss": 0.9124,
      "step": 403
    },
    {
      "epoch": 0.06035255452644159,
      "grad_norm": 0.5741954445838928,
      "learning_rate": 0.00019829019016259468,
      "loss": 1.4774,
      "step": 404
    },
    {
      "epoch": 0.060501942037645653,
      "grad_norm": 0.12422844022512436,
      "learning_rate": 0.00019828152488630016,
      "loss": 0.7628,
      "step": 405
    },
    {
      "epoch": 0.06065132954884972,
      "grad_norm": 0.22834278643131256,
      "learning_rate": 0.00019827283789806011,
      "loss": 1.0135,
      "step": 406
    },
    {
      "epoch": 0.06080071706005378,
      "grad_norm": 0.2153419405221939,
      "learning_rate": 0.00019826412919979358,
      "loss": 1.0723,
      "step": 407
    },
    {
      "epoch": 0.060950104571257845,
      "grad_norm": 0.18992449343204498,
      "learning_rate": 0.0001982553987934245,
      "loss": 0.5903,
      "step": 408
    },
    {
      "epoch": 0.061099492082461904,
      "grad_norm": 0.14883364737033844,
      "learning_rate": 0.00019824664668088155,
      "loss": 0.7192,
      "step": 409
    },
    {
      "epoch": 0.06124887959366597,
      "grad_norm": 0.12347893416881561,
      "learning_rate": 0.0001982378728640982,
      "loss": 0.785,
      "step": 410
    },
    {
      "epoch": 0.061398267104870036,
      "grad_norm": 0.14353862404823303,
      "learning_rate": 0.0001982290773450127,
      "loss": 0.6293,
      "step": 411
    },
    {
      "epoch": 0.061547654616074095,
      "grad_norm": 0.13001669943332672,
      "learning_rate": 0.00019822026012556818,
      "loss": 0.7083,
      "step": 412
    },
    {
      "epoch": 0.06169704212727816,
      "grad_norm": 0.15990330278873444,
      "learning_rate": 0.00019821142120771246,
      "loss": 0.6908,
      "step": 413
    },
    {
      "epoch": 0.06184642963848222,
      "grad_norm": 0.27101776003837585,
      "learning_rate": 0.0001982025605933982,
      "loss": 0.7311,
      "step": 414
    },
    {
      "epoch": 0.061995817149686286,
      "grad_norm": 0.3568134903907776,
      "learning_rate": 0.00019819367828458287,
      "loss": 1.0127,
      "step": 415
    },
    {
      "epoch": 0.06214520466089035,
      "grad_norm": 0.13109736144542694,
      "learning_rate": 0.0001981847742832287,
      "loss": 0.8776,
      "step": 416
    },
    {
      "epoch": 0.06229459217209441,
      "grad_norm": 0.1522371917963028,
      "learning_rate": 0.0001981758485913027,
      "loss": 0.9513,
      "step": 417
    },
    {
      "epoch": 0.06244397968329848,
      "grad_norm": 0.15717531740665436,
      "learning_rate": 0.00019816690121077674,
      "loss": 1.084,
      "step": 418
    },
    {
      "epoch": 0.06259336719450254,
      "grad_norm": 0.26732301712036133,
      "learning_rate": 0.00019815793214362742,
      "loss": 1.0219,
      "step": 419
    },
    {
      "epoch": 0.0627427547057066,
      "grad_norm": 0.1595582216978073,
      "learning_rate": 0.00019814894139183614,
      "loss": 0.7484,
      "step": 420
    },
    {
      "epoch": 0.06289214221691067,
      "grad_norm": 0.24443097412586212,
      "learning_rate": 0.00019813992895738908,
      "loss": 1.1702,
      "step": 421
    },
    {
      "epoch": 0.06304152972811473,
      "grad_norm": 0.12789739668369293,
      "learning_rate": 0.00019813089484227732,
      "loss": 0.9188,
      "step": 422
    },
    {
      "epoch": 0.06319091723931879,
      "grad_norm": 0.13589276373386383,
      "learning_rate": 0.00019812183904849653,
      "loss": 0.5864,
      "step": 423
    },
    {
      "epoch": 0.06334030475052285,
      "grad_norm": 0.20833753049373627,
      "learning_rate": 0.00019811276157804733,
      "loss": 0.9182,
      "step": 424
    },
    {
      "epoch": 0.06348969226172692,
      "grad_norm": 0.16936369240283966,
      "learning_rate": 0.0001981036624329351,
      "loss": 0.7968,
      "step": 425
    },
    {
      "epoch": 0.06363907977293098,
      "grad_norm": 0.21851451694965363,
      "learning_rate": 0.00019809454161516993,
      "loss": 0.6956,
      "step": 426
    },
    {
      "epoch": 0.06378846728413505,
      "grad_norm": 0.15087386965751648,
      "learning_rate": 0.0001980853991267668,
      "loss": 1.0246,
      "step": 427
    },
    {
      "epoch": 0.06393785479533912,
      "grad_norm": 0.12114626169204712,
      "learning_rate": 0.00019807623496974537,
      "loss": 0.7678,
      "step": 428
    },
    {
      "epoch": 0.06408724230654317,
      "grad_norm": 0.20948128402233124,
      "learning_rate": 0.00019806704914613024,
      "loss": 1.1529,
      "step": 429
    },
    {
      "epoch": 0.06423662981774723,
      "grad_norm": 0.1780691146850586,
      "learning_rate": 0.0001980578416579506,
      "loss": 0.9683,
      "step": 430
    },
    {
      "epoch": 0.0643860173289513,
      "grad_norm": 0.14259085059165955,
      "learning_rate": 0.00019804861250724063,
      "loss": 0.4694,
      "step": 431
    },
    {
      "epoch": 0.06453540484015537,
      "grad_norm": 0.2003784030675888,
      "learning_rate": 0.00019803936169603912,
      "loss": 0.6101,
      "step": 432
    },
    {
      "epoch": 0.06468479235135943,
      "grad_norm": 0.19409967958927155,
      "learning_rate": 0.00019803008922638976,
      "loss": 1.2219,
      "step": 433
    },
    {
      "epoch": 0.06483417986256348,
      "grad_norm": 0.12308470904827118,
      "learning_rate": 0.00019802079510034096,
      "loss": 0.8568,
      "step": 434
    },
    {
      "epoch": 0.06498356737376755,
      "grad_norm": 0.3875686228275299,
      "learning_rate": 0.00019801147931994596,
      "loss": 1.027,
      "step": 435
    },
    {
      "epoch": 0.06513295488497162,
      "grad_norm": 0.21796062588691711,
      "learning_rate": 0.00019800214188726276,
      "loss": 1.2517,
      "step": 436
    },
    {
      "epoch": 0.06528234239617568,
      "grad_norm": 0.14813588559627533,
      "learning_rate": 0.00019799278280435413,
      "loss": 0.8185,
      "step": 437
    },
    {
      "epoch": 0.06543172990737975,
      "grad_norm": 0.215727299451828,
      "learning_rate": 0.00019798340207328766,
      "loss": 1.3439,
      "step": 438
    },
    {
      "epoch": 0.0655811174185838,
      "grad_norm": 0.25380611419677734,
      "learning_rate": 0.0001979739996961357,
      "loss": 1.2215,
      "step": 439
    },
    {
      "epoch": 0.06573050492978787,
      "grad_norm": 0.16782647371292114,
      "learning_rate": 0.00019796457567497537,
      "loss": 0.9321,
      "step": 440
    },
    {
      "epoch": 0.06587989244099193,
      "grad_norm": 0.13334587216377258,
      "learning_rate": 0.0001979551300118886,
      "loss": 0.8722,
      "step": 441
    },
    {
      "epoch": 0.066029279952196,
      "grad_norm": 0.27328580617904663,
      "learning_rate": 0.0001979456627089621,
      "loss": 0.9013,
      "step": 442
    },
    {
      "epoch": 0.06617866746340006,
      "grad_norm": 0.10864201933145523,
      "learning_rate": 0.0001979361737682873,
      "loss": 0.6201,
      "step": 443
    },
    {
      "epoch": 0.06632805497460412,
      "grad_norm": 0.1800958216190338,
      "learning_rate": 0.0001979266631919605,
      "loss": 0.727,
      "step": 444
    },
    {
      "epoch": 0.06647744248580818,
      "grad_norm": 0.14912551641464233,
      "learning_rate": 0.00019791713098208272,
      "loss": 0.6877,
      "step": 445
    },
    {
      "epoch": 0.06662682999701225,
      "grad_norm": 0.3135051727294922,
      "learning_rate": 0.00019790757714075979,
      "loss": 0.9517,
      "step": 446
    },
    {
      "epoch": 0.06677621750821632,
      "grad_norm": 0.23976337909698486,
      "learning_rate": 0.0001978980016701023,
      "loss": 0.9952,
      "step": 447
    },
    {
      "epoch": 0.06692560501942038,
      "grad_norm": 0.16540992259979248,
      "learning_rate": 0.00019788840457222556,
      "loss": 0.796,
      "step": 448
    },
    {
      "epoch": 0.06707499253062443,
      "grad_norm": 0.19656099379062653,
      "learning_rate": 0.00019787878584924984,
      "loss": 0.7593,
      "step": 449
    },
    {
      "epoch": 0.0672243800418285,
      "grad_norm": 0.12927167117595673,
      "learning_rate": 0.0001978691455033,
      "loss": 0.819,
      "step": 450
    },
    {
      "epoch": 0.06737376755303257,
      "grad_norm": 0.19154123961925507,
      "learning_rate": 0.00019785948353650572,
      "loss": 0.9713,
      "step": 451
    },
    {
      "epoch": 0.06752315506423663,
      "grad_norm": 0.12065298110246658,
      "learning_rate": 0.0001978497999510015,
      "loss": 0.7942,
      "step": 452
    },
    {
      "epoch": 0.0676725425754407,
      "grad_norm": 0.1475251317024231,
      "learning_rate": 0.00019784009474892666,
      "loss": 0.8598,
      "step": 453
    },
    {
      "epoch": 0.06782193008664476,
      "grad_norm": 0.14072950184345245,
      "learning_rate": 0.00019783036793242516,
      "loss": 0.8956,
      "step": 454
    },
    {
      "epoch": 0.06797131759784882,
      "grad_norm": 0.20561757683753967,
      "learning_rate": 0.00019782061950364584,
      "loss": 0.9214,
      "step": 455
    },
    {
      "epoch": 0.06812070510905288,
      "grad_norm": 0.1547221839427948,
      "learning_rate": 0.00019781084946474226,
      "loss": 0.9042,
      "step": 456
    },
    {
      "epoch": 0.06827009262025695,
      "grad_norm": 0.233870729804039,
      "learning_rate": 0.0001978010578178728,
      "loss": 1.0727,
      "step": 457
    },
    {
      "epoch": 0.06841948013146101,
      "grad_norm": 0.19565890729427338,
      "learning_rate": 0.00019779124456520056,
      "loss": 0.8425,
      "step": 458
    },
    {
      "epoch": 0.06856886764266508,
      "grad_norm": 0.15683157742023468,
      "learning_rate": 0.00019778140970889348,
      "loss": 0.7989,
      "step": 459
    },
    {
      "epoch": 0.06871825515386913,
      "grad_norm": 0.12861517071723938,
      "learning_rate": 0.0001977715532511242,
      "loss": 0.8849,
      "step": 460
    },
    {
      "epoch": 0.0688676426650732,
      "grad_norm": 0.12733756005764008,
      "learning_rate": 0.00019776167519407022,
      "loss": 0.7984,
      "step": 461
    },
    {
      "epoch": 0.06901703017627726,
      "grad_norm": 0.14494583010673523,
      "learning_rate": 0.0001977517755399137,
      "loss": 0.8234,
      "step": 462
    },
    {
      "epoch": 0.06916641768748133,
      "grad_norm": 0.15694382786750793,
      "learning_rate": 0.00019774185429084166,
      "loss": 0.8476,
      "step": 463
    },
    {
      "epoch": 0.0693158051986854,
      "grad_norm": 0.2537318766117096,
      "learning_rate": 0.00019773191144904586,
      "loss": 0.9928,
      "step": 464
    },
    {
      "epoch": 0.06946519270988945,
      "grad_norm": 0.45019641518592834,
      "learning_rate": 0.0001977219470167228,
      "loss": 1.5181,
      "step": 465
    },
    {
      "epoch": 0.06961458022109351,
      "grad_norm": 0.18655377626419067,
      "learning_rate": 0.00019771196099607386,
      "loss": 0.7463,
      "step": 466
    },
    {
      "epoch": 0.06976396773229758,
      "grad_norm": 0.1978912502527237,
      "learning_rate": 0.00019770195338930503,
      "loss": 1.0358,
      "step": 467
    },
    {
      "epoch": 0.06991335524350165,
      "grad_norm": 0.15748754143714905,
      "learning_rate": 0.00019769192419862716,
      "loss": 0.6154,
      "step": 468
    },
    {
      "epoch": 0.07006274275470571,
      "grad_norm": 0.17752663791179657,
      "learning_rate": 0.00019768187342625592,
      "loss": 0.6289,
      "step": 469
    },
    {
      "epoch": 0.07021213026590976,
      "grad_norm": 0.12329550832509995,
      "learning_rate": 0.0001976718010744116,
      "loss": 0.9018,
      "step": 470
    },
    {
      "epoch": 0.07036151777711383,
      "grad_norm": 0.17692221701145172,
      "learning_rate": 0.00019766170714531937,
      "loss": 1.0348,
      "step": 471
    },
    {
      "epoch": 0.0705109052883179,
      "grad_norm": 0.2912501096725464,
      "learning_rate": 0.00019765159164120916,
      "loss": 0.9097,
      "step": 472
    },
    {
      "epoch": 0.07066029279952196,
      "grad_norm": 0.118867889046669,
      "learning_rate": 0.00019764145456431566,
      "loss": 0.6933,
      "step": 473
    },
    {
      "epoch": 0.07080968031072603,
      "grad_norm": 0.15778154134750366,
      "learning_rate": 0.00019763129591687827,
      "loss": 0.5504,
      "step": 474
    },
    {
      "epoch": 0.07095906782193008,
      "grad_norm": 0.14886140823364258,
      "learning_rate": 0.00019762111570114122,
      "loss": 0.8209,
      "step": 475
    },
    {
      "epoch": 0.07110845533313415,
      "grad_norm": 0.10714101046323776,
      "learning_rate": 0.00019761091391935347,
      "loss": 0.6473,
      "step": 476
    },
    {
      "epoch": 0.07125784284433821,
      "grad_norm": 0.15396277606487274,
      "learning_rate": 0.00019760069057376875,
      "loss": 0.8179,
      "step": 477
    },
    {
      "epoch": 0.07140723035554228,
      "grad_norm": 0.1531289666891098,
      "learning_rate": 0.00019759044566664558,
      "loss": 0.9508,
      "step": 478
    },
    {
      "epoch": 0.07155661786674634,
      "grad_norm": 0.11772772669792175,
      "learning_rate": 0.0001975801792002472,
      "loss": 0.6606,
      "step": 479
    },
    {
      "epoch": 0.0717060053779504,
      "grad_norm": 0.16291484236717224,
      "learning_rate": 0.00019756989117684164,
      "loss": 0.6476,
      "step": 480
    },
    {
      "epoch": 0.07185539288915446,
      "grad_norm": 0.15869931876659393,
      "learning_rate": 0.00019755958159870172,
      "loss": 0.923,
      "step": 481
    },
    {
      "epoch": 0.07200478040035853,
      "grad_norm": 0.1591719686985016,
      "learning_rate": 0.00019754925046810493,
      "loss": 0.7225,
      "step": 482
    },
    {
      "epoch": 0.0721541679115626,
      "grad_norm": 0.1619904637336731,
      "learning_rate": 0.00019753889778733363,
      "loss": 0.7185,
      "step": 483
    },
    {
      "epoch": 0.07230355542276666,
      "grad_norm": 0.12438057363033295,
      "learning_rate": 0.00019752852355867486,
      "loss": 0.7659,
      "step": 484
    },
    {
      "epoch": 0.07245294293397073,
      "grad_norm": 0.11964958906173706,
      "learning_rate": 0.00019751812778442046,
      "loss": 0.7564,
      "step": 485
    },
    {
      "epoch": 0.07260233044517478,
      "grad_norm": 0.2836957573890686,
      "learning_rate": 0.00019750771046686704,
      "loss": 1.4225,
      "step": 486
    },
    {
      "epoch": 0.07275171795637884,
      "grad_norm": 0.13266032934188843,
      "learning_rate": 0.00019749727160831593,
      "loss": 0.7952,
      "step": 487
    },
    {
      "epoch": 0.07290110546758291,
      "grad_norm": 0.12412168830633163,
      "learning_rate": 0.00019748681121107325,
      "loss": 0.9047,
      "step": 488
    },
    {
      "epoch": 0.07305049297878698,
      "grad_norm": 0.20590822398662567,
      "learning_rate": 0.00019747632927744982,
      "loss": 0.7496,
      "step": 489
    },
    {
      "epoch": 0.07319988048999104,
      "grad_norm": 0.18052905797958374,
      "learning_rate": 0.00019746582580976136,
      "loss": 0.6724,
      "step": 490
    },
    {
      "epoch": 0.0733492680011951,
      "grad_norm": 0.24314431846141815,
      "learning_rate": 0.0001974553008103282,
      "loss": 0.8552,
      "step": 491
    },
    {
      "epoch": 0.07349865551239916,
      "grad_norm": 0.14932356774806976,
      "learning_rate": 0.00019744475428147546,
      "loss": 0.8878,
      "step": 492
    },
    {
      "epoch": 0.07364804302360323,
      "grad_norm": 0.12878401577472687,
      "learning_rate": 0.00019743418622553303,
      "loss": 0.7097,
      "step": 493
    },
    {
      "epoch": 0.07379743053480729,
      "grad_norm": 0.1838715523481369,
      "learning_rate": 0.00019742359664483563,
      "loss": 1.0581,
      "step": 494
    },
    {
      "epoch": 0.07394681804601136,
      "grad_norm": 0.13782045245170593,
      "learning_rate": 0.0001974129855417226,
      "loss": 1.0439,
      "step": 495
    },
    {
      "epoch": 0.07409620555721541,
      "grad_norm": 0.179475337266922,
      "learning_rate": 0.00019740235291853812,
      "loss": 0.7708,
      "step": 496
    },
    {
      "epoch": 0.07424559306841948,
      "grad_norm": 0.1903340369462967,
      "learning_rate": 0.0001973916987776311,
      "loss": 1.3356,
      "step": 497
    },
    {
      "epoch": 0.07439498057962354,
      "grad_norm": 0.13480018079280853,
      "learning_rate": 0.00019738102312135523,
      "loss": 0.5876,
      "step": 498
    },
    {
      "epoch": 0.07454436809082761,
      "grad_norm": 0.10856112837791443,
      "learning_rate": 0.0001973703259520689,
      "loss": 0.6991,
      "step": 499
    },
    {
      "epoch": 0.07469375560203168,
      "grad_norm": 0.3036371171474457,
      "learning_rate": 0.0001973596072721353,
      "loss": 1.0438,
      "step": 500
    },
    {
      "epoch": 0.07484314311323573,
      "grad_norm": 0.12173090875148773,
      "learning_rate": 0.0001973488670839224,
      "loss": 0.5836,
      "step": 501
    },
    {
      "epoch": 0.0749925306244398,
      "grad_norm": 0.2667033076286316,
      "learning_rate": 0.00019733810538980281,
      "loss": 0.49,
      "step": 502
    },
    {
      "epoch": 0.07514191813564386,
      "grad_norm": 0.092626191675663,
      "learning_rate": 0.00019732732219215397,
      "loss": 0.3618,
      "step": 503
    },
    {
      "epoch": 0.07529130564684793,
      "grad_norm": 0.12614378333091736,
      "learning_rate": 0.0001973165174933581,
      "loss": 0.8307,
      "step": 504
    },
    {
      "epoch": 0.07544069315805199,
      "grad_norm": 0.14259332418441772,
      "learning_rate": 0.00019730569129580206,
      "loss": 0.8972,
      "step": 505
    },
    {
      "epoch": 0.07559008066925604,
      "grad_norm": 0.12121133506298065,
      "learning_rate": 0.0001972948436018776,
      "loss": 0.8446,
      "step": 506
    },
    {
      "epoch": 0.07573946818046011,
      "grad_norm": 0.16819995641708374,
      "learning_rate": 0.00019728397441398112,
      "loss": 0.8122,
      "step": 507
    },
    {
      "epoch": 0.07588885569166418,
      "grad_norm": 0.14065992832183838,
      "learning_rate": 0.00019727308373451377,
      "loss": 0.457,
      "step": 508
    },
    {
      "epoch": 0.07603824320286824,
      "grad_norm": 0.16299694776535034,
      "learning_rate": 0.0001972621715658815,
      "loss": 0.878,
      "step": 509
    },
    {
      "epoch": 0.07618763071407231,
      "grad_norm": 0.1941678524017334,
      "learning_rate": 0.000197251237910495,
      "loss": 0.9534,
      "step": 510
    },
    {
      "epoch": 0.07633701822527637,
      "grad_norm": 0.14302954077720642,
      "learning_rate": 0.00019724028277076964,
      "loss": 0.8577,
      "step": 511
    },
    {
      "epoch": 0.07648640573648043,
      "grad_norm": 0.19309553503990173,
      "learning_rate": 0.00019722930614912563,
      "loss": 0.9864,
      "step": 512
    },
    {
      "epoch": 0.07663579324768449,
      "grad_norm": 0.1626858413219452,
      "learning_rate": 0.00019721830804798787,
      "loss": 0.8104,
      "step": 513
    },
    {
      "epoch": 0.07678518075888856,
      "grad_norm": 0.11680503189563751,
      "learning_rate": 0.00019720728846978598,
      "loss": 0.7225,
      "step": 514
    },
    {
      "epoch": 0.07693456827009262,
      "grad_norm": 0.11943230032920837,
      "learning_rate": 0.0001971962474169544,
      "loss": 0.895,
      "step": 515
    },
    {
      "epoch": 0.07708395578129669,
      "grad_norm": 0.2065763920545578,
      "learning_rate": 0.00019718518489193225,
      "loss": 0.7482,
      "step": 516
    },
    {
      "epoch": 0.07723334329250074,
      "grad_norm": 0.13309244811534882,
      "learning_rate": 0.0001971741008971634,
      "loss": 0.5227,
      "step": 517
    },
    {
      "epoch": 0.07738273080370481,
      "grad_norm": 0.13642869889736176,
      "learning_rate": 0.00019716299543509654,
      "loss": 0.6441,
      "step": 518
    },
    {
      "epoch": 0.07753211831490887,
      "grad_norm": 0.18893510103225708,
      "learning_rate": 0.00019715186850818498,
      "loss": 0.7619,
      "step": 519
    },
    {
      "epoch": 0.07768150582611294,
      "grad_norm": 0.3364134132862091,
      "learning_rate": 0.00019714072011888686,
      "loss": 1.0618,
      "step": 520
    },
    {
      "epoch": 0.077830893337317,
      "grad_norm": 0.1486874371767044,
      "learning_rate": 0.00019712955026966506,
      "loss": 0.7846,
      "step": 521
    },
    {
      "epoch": 0.07798028084852106,
      "grad_norm": 0.21529722213745117,
      "learning_rate": 0.00019711835896298713,
      "loss": 0.4896,
      "step": 522
    },
    {
      "epoch": 0.07812966835972512,
      "grad_norm": 0.1832679957151413,
      "learning_rate": 0.00019710714620132546,
      "loss": 0.8258,
      "step": 523
    },
    {
      "epoch": 0.07827905587092919,
      "grad_norm": 0.189329594373703,
      "learning_rate": 0.00019709591198715707,
      "loss": 0.9139,
      "step": 524
    },
    {
      "epoch": 0.07842844338213326,
      "grad_norm": 0.13310536742210388,
      "learning_rate": 0.0001970846563229638,
      "loss": 0.9531,
      "step": 525
    },
    {
      "epoch": 0.07857783089333732,
      "grad_norm": 0.5338667631149292,
      "learning_rate": 0.00019707337921123221,
      "loss": 1.0734,
      "step": 526
    },
    {
      "epoch": 0.07872721840454137,
      "grad_norm": 0.6397843360900879,
      "learning_rate": 0.0001970620806544536,
      "loss": 1.4513,
      "step": 527
    },
    {
      "epoch": 0.07887660591574544,
      "grad_norm": 0.14354926347732544,
      "learning_rate": 0.00019705076065512398,
      "loss": 0.96,
      "step": 528
    },
    {
      "epoch": 0.0790259934269495,
      "grad_norm": 0.11550958454608917,
      "learning_rate": 0.00019703941921574413,
      "loss": 0.7829,
      "step": 529
    },
    {
      "epoch": 0.07917538093815357,
      "grad_norm": 0.14045141637325287,
      "learning_rate": 0.00019702805633881957,
      "loss": 0.6976,
      "step": 530
    },
    {
      "epoch": 0.07932476844935764,
      "grad_norm": 0.18493744730949402,
      "learning_rate": 0.00019701667202686048,
      "loss": 1.1065,
      "step": 531
    },
    {
      "epoch": 0.07947415596056169,
      "grad_norm": 0.2148275524377823,
      "learning_rate": 0.0001970052662823819,
      "loss": 1.1823,
      "step": 532
    },
    {
      "epoch": 0.07962354347176576,
      "grad_norm": 0.1770048886537552,
      "learning_rate": 0.0001969938391079035,
      "loss": 0.8534,
      "step": 533
    },
    {
      "epoch": 0.07977293098296982,
      "grad_norm": 0.11045973002910614,
      "learning_rate": 0.00019698239050594977,
      "loss": 0.628,
      "step": 534
    },
    {
      "epoch": 0.07992231849417389,
      "grad_norm": 0.113344706594944,
      "learning_rate": 0.0001969709204790498,
      "loss": 0.7655,
      "step": 535
    },
    {
      "epoch": 0.08007170600537795,
      "grad_norm": 0.13493701815605164,
      "learning_rate": 0.0001969594290297376,
      "loss": 0.6173,
      "step": 536
    },
    {
      "epoch": 0.08022109351658202,
      "grad_norm": 0.10995621234178543,
      "learning_rate": 0.00019694791616055177,
      "loss": 0.8091,
      "step": 537
    },
    {
      "epoch": 0.08037048102778607,
      "grad_norm": 0.1691836416721344,
      "learning_rate": 0.00019693638187403563,
      "loss": 0.9478,
      "step": 538
    },
    {
      "epoch": 0.08051986853899014,
      "grad_norm": 0.1570545732975006,
      "learning_rate": 0.0001969248261727374,
      "loss": 0.8514,
      "step": 539
    },
    {
      "epoch": 0.0806692560501942,
      "grad_norm": 0.14363858103752136,
      "learning_rate": 0.00019691324905920984,
      "loss": 0.5037,
      "step": 540
    },
    {
      "epoch": 0.08081864356139827,
      "grad_norm": 0.12334268540143967,
      "learning_rate": 0.00019690165053601056,
      "loss": 0.9084,
      "step": 541
    },
    {
      "epoch": 0.08096803107260234,
      "grad_norm": 0.3128170073032379,
      "learning_rate": 0.0001968900306057018,
      "loss": 0.737,
      "step": 542
    },
    {
      "epoch": 0.08111741858380639,
      "grad_norm": 0.16364172101020813,
      "learning_rate": 0.00019687838927085066,
      "loss": 0.4789,
      "step": 543
    },
    {
      "epoch": 0.08126680609501045,
      "grad_norm": 0.15874239802360535,
      "learning_rate": 0.0001968667265340288,
      "loss": 1.0205,
      "step": 544
    },
    {
      "epoch": 0.08141619360621452,
      "grad_norm": 0.13250084221363068,
      "learning_rate": 0.00019685504239781278,
      "loss": 0.8377,
      "step": 545
    },
    {
      "epoch": 0.08156558111741859,
      "grad_norm": 0.13961811363697052,
      "learning_rate": 0.00019684333686478383,
      "loss": 0.857,
      "step": 546
    },
    {
      "epoch": 0.08171496862862265,
      "grad_norm": 0.1638427972793579,
      "learning_rate": 0.0001968316099375278,
      "loss": 0.6918,
      "step": 547
    },
    {
      "epoch": 0.0818643561398267,
      "grad_norm": 0.260032057762146,
      "learning_rate": 0.00019681986161863542,
      "loss": 0.965,
      "step": 548
    },
    {
      "epoch": 0.08201374365103077,
      "grad_norm": 0.12642204761505127,
      "learning_rate": 0.00019680809191070203,
      "loss": 0.5236,
      "step": 549
    },
    {
      "epoch": 0.08216313116223484,
      "grad_norm": 0.18395595252513885,
      "learning_rate": 0.00019679630081632782,
      "loss": 1.0722,
      "step": 550
    },
    {
      "epoch": 0.0823125186734389,
      "grad_norm": 0.14553199708461761,
      "learning_rate": 0.0001967844883381176,
      "loss": 0.7844,
      "step": 551
    },
    {
      "epoch": 0.08246190618464297,
      "grad_norm": 0.13094641268253326,
      "learning_rate": 0.00019677265447868086,
      "loss": 0.5937,
      "step": 552
    },
    {
      "epoch": 0.08261129369584702,
      "grad_norm": 0.1621766984462738,
      "learning_rate": 0.00019676079924063196,
      "loss": 0.7064,
      "step": 553
    },
    {
      "epoch": 0.08276068120705109,
      "grad_norm": 0.16490310430526733,
      "learning_rate": 0.0001967489226265899,
      "loss": 0.654,
      "step": 554
    },
    {
      "epoch": 0.08291006871825515,
      "grad_norm": 0.1863924264907837,
      "learning_rate": 0.00019673702463917842,
      "loss": 0.8091,
      "step": 555
    },
    {
      "epoch": 0.08305945622945922,
      "grad_norm": 0.1565409004688263,
      "learning_rate": 0.00019672510528102597,
      "loss": 0.583,
      "step": 556
    },
    {
      "epoch": 0.08320884374066329,
      "grad_norm": 0.2025175541639328,
      "learning_rate": 0.0001967131645547657,
      "loss": 0.8749,
      "step": 557
    },
    {
      "epoch": 0.08335823125186734,
      "grad_norm": 0.23709741234779358,
      "learning_rate": 0.0001967012024630355,
      "loss": 1.1301,
      "step": 558
    },
    {
      "epoch": 0.0835076187630714,
      "grad_norm": 0.33181461691856384,
      "learning_rate": 0.00019668921900847805,
      "loss": 1.0527,
      "step": 559
    },
    {
      "epoch": 0.08365700627427547,
      "grad_norm": 0.11158733814954758,
      "learning_rate": 0.00019667721419374065,
      "loss": 0.6433,
      "step": 560
    },
    {
      "epoch": 0.08380639378547954,
      "grad_norm": 0.21710413694381714,
      "learning_rate": 0.00019666518802147534,
      "loss": 0.7117,
      "step": 561
    },
    {
      "epoch": 0.0839557812966836,
      "grad_norm": 0.1480075567960739,
      "learning_rate": 0.00019665314049433888,
      "loss": 0.8019,
      "step": 562
    },
    {
      "epoch": 0.08410516880788767,
      "grad_norm": 0.15463611483573914,
      "learning_rate": 0.00019664107161499277,
      "loss": 0.679,
      "step": 563
    },
    {
      "epoch": 0.08425455631909172,
      "grad_norm": 0.13593655824661255,
      "learning_rate": 0.00019662898138610323,
      "loss": 0.7527,
      "step": 564
    },
    {
      "epoch": 0.08440394383029579,
      "grad_norm": 0.1662757396697998,
      "learning_rate": 0.0001966168698103412,
      "loss": 0.7486,
      "step": 565
    },
    {
      "epoch": 0.08455333134149985,
      "grad_norm": 0.10743851959705353,
      "learning_rate": 0.00019660473689038228,
      "loss": 0.6341,
      "step": 566
    },
    {
      "epoch": 0.08470271885270392,
      "grad_norm": 0.15268459916114807,
      "learning_rate": 0.00019659258262890683,
      "loss": 0.8236,
      "step": 567
    },
    {
      "epoch": 0.08485210636390798,
      "grad_norm": 0.14038819074630737,
      "learning_rate": 0.00019658040702859997,
      "loss": 0.8065,
      "step": 568
    },
    {
      "epoch": 0.08500149387511204,
      "grad_norm": 0.15066532790660858,
      "learning_rate": 0.0001965682100921514,
      "loss": 0.9532,
      "step": 569
    },
    {
      "epoch": 0.0851508813863161,
      "grad_norm": 0.1580052673816681,
      "learning_rate": 0.00019655599182225565,
      "loss": 1.0969,
      "step": 570
    },
    {
      "epoch": 0.08530026889752017,
      "grad_norm": 0.16247299313545227,
      "learning_rate": 0.000196543752221612,
      "loss": 0.791,
      "step": 571
    },
    {
      "epoch": 0.08544965640872423,
      "grad_norm": 0.1083194687962532,
      "learning_rate": 0.00019653149129292426,
      "loss": 0.6203,
      "step": 572
    },
    {
      "epoch": 0.0855990439199283,
      "grad_norm": 0.19076959788799286,
      "learning_rate": 0.0001965192090389011,
      "loss": 0.6709,
      "step": 573
    },
    {
      "epoch": 0.08574843143113235,
      "grad_norm": 0.15673895180225372,
      "learning_rate": 0.00019650690546225592,
      "loss": 0.7816,
      "step": 574
    },
    {
      "epoch": 0.08589781894233642,
      "grad_norm": 0.13381899893283844,
      "learning_rate": 0.00019649458056570672,
      "loss": 0.8227,
      "step": 575
    },
    {
      "epoch": 0.08604720645354048,
      "grad_norm": 0.10087165981531143,
      "learning_rate": 0.00019648223435197627,
      "loss": 0.5839,
      "step": 576
    },
    {
      "epoch": 0.08619659396474455,
      "grad_norm": 0.17345178127288818,
      "learning_rate": 0.00019646986682379206,
      "loss": 0.8435,
      "step": 577
    },
    {
      "epoch": 0.08634598147594862,
      "grad_norm": 0.16065159440040588,
      "learning_rate": 0.00019645747798388628,
      "loss": 0.8124,
      "step": 578
    },
    {
      "epoch": 0.08649536898715267,
      "grad_norm": 0.32011693716049194,
      "learning_rate": 0.0001964450678349958,
      "loss": 0.8119,
      "step": 579
    },
    {
      "epoch": 0.08664475649835673,
      "grad_norm": 0.13616526126861572,
      "learning_rate": 0.0001964326363798622,
      "loss": 0.8056,
      "step": 580
    },
    {
      "epoch": 0.0867941440095608,
      "grad_norm": 0.28961271047592163,
      "learning_rate": 0.00019642018362123182,
      "loss": 1.0182,
      "step": 581
    },
    {
      "epoch": 0.08694353152076487,
      "grad_norm": 0.16200341284275055,
      "learning_rate": 0.00019640770956185567,
      "loss": 0.9604,
      "step": 582
    },
    {
      "epoch": 0.08709291903196893,
      "grad_norm": 0.1426459699869156,
      "learning_rate": 0.00019639521420448947,
      "loss": 0.6555,
      "step": 583
    },
    {
      "epoch": 0.08724230654317298,
      "grad_norm": 0.11389955133199692,
      "learning_rate": 0.0001963826975518936,
      "loss": 0.6878,
      "step": 584
    },
    {
      "epoch": 0.08739169405437705,
      "grad_norm": 0.12384461611509323,
      "learning_rate": 0.00019637015960683322,
      "loss": 0.612,
      "step": 585
    },
    {
      "epoch": 0.08754108156558112,
      "grad_norm": 0.1436391919851303,
      "learning_rate": 0.00019635760037207817,
      "loss": 0.7069,
      "step": 586
    },
    {
      "epoch": 0.08769046907678518,
      "grad_norm": 0.1199880838394165,
      "learning_rate": 0.00019634501985040296,
      "loss": 0.7547,
      "step": 587
    },
    {
      "epoch": 0.08783985658798925,
      "grad_norm": 0.16157324612140656,
      "learning_rate": 0.00019633241804458687,
      "loss": 0.7839,
      "step": 588
    },
    {
      "epoch": 0.0879892440991933,
      "grad_norm": 0.16362355649471283,
      "learning_rate": 0.00019631979495741378,
      "loss": 0.7197,
      "step": 589
    },
    {
      "epoch": 0.08813863161039737,
      "grad_norm": 0.1794954091310501,
      "learning_rate": 0.00019630715059167238,
      "loss": 0.6936,
      "step": 590
    },
    {
      "epoch": 0.08828801912160143,
      "grad_norm": 0.3147851228713989,
      "learning_rate": 0.00019629448495015597,
      "loss": 1.8882,
      "step": 591
    },
    {
      "epoch": 0.0884374066328055,
      "grad_norm": 0.1880149394273758,
      "learning_rate": 0.0001962817980356626,
      "loss": 0.7354,
      "step": 592
    },
    {
      "epoch": 0.08858679414400956,
      "grad_norm": 0.13019341230392456,
      "learning_rate": 0.00019626908985099503,
      "loss": 0.9714,
      "step": 593
    },
    {
      "epoch": 0.08873618165521363,
      "grad_norm": 0.18782681226730347,
      "learning_rate": 0.0001962563603989607,
      "loss": 0.9334,
      "step": 594
    },
    {
      "epoch": 0.08888556916641768,
      "grad_norm": 0.1618777960538864,
      "learning_rate": 0.00019624360968237172,
      "loss": 0.8143,
      "step": 595
    },
    {
      "epoch": 0.08903495667762175,
      "grad_norm": 0.13201208412647247,
      "learning_rate": 0.00019623083770404492,
      "loss": 0.9077,
      "step": 596
    },
    {
      "epoch": 0.08918434418882581,
      "grad_norm": 0.11902808398008347,
      "learning_rate": 0.0001962180444668019,
      "loss": 1.014,
      "step": 597
    },
    {
      "epoch": 0.08933373170002988,
      "grad_norm": 0.17474393546581268,
      "learning_rate": 0.0001962052299734688,
      "loss": 0.8466,
      "step": 598
    },
    {
      "epoch": 0.08948311921123395,
      "grad_norm": 0.2875848412513733,
      "learning_rate": 0.00019619239422687663,
      "loss": 0.8023,
      "step": 599
    },
    {
      "epoch": 0.089632506722438,
      "grad_norm": 0.13769172132015228,
      "learning_rate": 0.00019617953722986096,
      "loss": 0.9026,
      "step": 600
    },
    {
      "epoch": 0.08978189423364207,
      "grad_norm": 0.30760055780410767,
      "learning_rate": 0.00019616665898526206,
      "loss": 1.0103,
      "step": 601
    },
    {
      "epoch": 0.08993128174484613,
      "grad_norm": 0.5774983763694763,
      "learning_rate": 0.00019615375949592504,
      "loss": 1.4516,
      "step": 602
    },
    {
      "epoch": 0.0900806692560502,
      "grad_norm": 0.13711205124855042,
      "learning_rate": 0.00019614083876469954,
      "loss": 0.7543,
      "step": 603
    },
    {
      "epoch": 0.09023005676725426,
      "grad_norm": 0.1261071413755417,
      "learning_rate": 0.00019612789679443997,
      "loss": 0.6952,
      "step": 604
    },
    {
      "epoch": 0.09037944427845832,
      "grad_norm": 0.13197720050811768,
      "learning_rate": 0.00019611493358800538,
      "loss": 0.906,
      "step": 605
    },
    {
      "epoch": 0.09052883178966238,
      "grad_norm": 0.12066524475812912,
      "learning_rate": 0.00019610194914825962,
      "loss": 0.7621,
      "step": 606
    },
    {
      "epoch": 0.09067821930086645,
      "grad_norm": 0.1722782999277115,
      "learning_rate": 0.00019608894347807108,
      "loss": 0.8798,
      "step": 607
    },
    {
      "epoch": 0.09082760681207051,
      "grad_norm": 0.1058247983455658,
      "learning_rate": 0.000196075916580313,
      "loss": 0.6246,
      "step": 608
    },
    {
      "epoch": 0.09097699432327458,
      "grad_norm": 0.17862388491630554,
      "learning_rate": 0.00019606286845786315,
      "loss": 0.6607,
      "step": 609
    },
    {
      "epoch": 0.09112638183447863,
      "grad_norm": 0.1704971194267273,
      "learning_rate": 0.0001960497991136041,
      "loss": 0.8648,
      "step": 610
    },
    {
      "epoch": 0.0912757693456827,
      "grad_norm": 0.12244229018688202,
      "learning_rate": 0.00019603670855042308,
      "loss": 0.7546,
      "step": 611
    },
    {
      "epoch": 0.09142515685688676,
      "grad_norm": 0.1924651712179184,
      "learning_rate": 0.00019602359677121199,
      "loss": 0.7931,
      "step": 612
    },
    {
      "epoch": 0.09157454436809083,
      "grad_norm": 0.1722334921360016,
      "learning_rate": 0.00019601046377886746,
      "loss": 1.259,
      "step": 613
    },
    {
      "epoch": 0.0917239318792949,
      "grad_norm": 0.11828108131885529,
      "learning_rate": 0.0001959973095762907,
      "loss": 0.8418,
      "step": 614
    },
    {
      "epoch": 0.09187331939049895,
      "grad_norm": 0.10462518781423569,
      "learning_rate": 0.0001959841341663878,
      "loss": 0.5707,
      "step": 615
    },
    {
      "epoch": 0.09202270690170301,
      "grad_norm": 0.14686742424964905,
      "learning_rate": 0.00019597093755206936,
      "loss": 1.0242,
      "step": 616
    },
    {
      "epoch": 0.09217209441290708,
      "grad_norm": 0.1107390895485878,
      "learning_rate": 0.00019595771973625068,
      "loss": 0.6829,
      "step": 617
    },
    {
      "epoch": 0.09232148192411115,
      "grad_norm": 0.18590034544467926,
      "learning_rate": 0.00019594448072185182,
      "loss": 0.7426,
      "step": 618
    },
    {
      "epoch": 0.09247086943531521,
      "grad_norm": 0.12690337002277374,
      "learning_rate": 0.00019593122051179748,
      "loss": 0.978,
      "step": 619
    },
    {
      "epoch": 0.09262025694651928,
      "grad_norm": 0.19712156057357788,
      "learning_rate": 0.00019591793910901707,
      "loss": 0.8153,
      "step": 620
    },
    {
      "epoch": 0.09276964445772333,
      "grad_norm": 0.16727258265018463,
      "learning_rate": 0.00019590463651644464,
      "loss": 0.8827,
      "step": 621
    },
    {
      "epoch": 0.0929190319689274,
      "grad_norm": 0.21075357496738434,
      "learning_rate": 0.00019589131273701894,
      "loss": 0.624,
      "step": 622
    },
    {
      "epoch": 0.09306841948013146,
      "grad_norm": 0.5324035286903381,
      "learning_rate": 0.00019587796777368347,
      "loss": 1.808,
      "step": 623
    },
    {
      "epoch": 0.09321780699133553,
      "grad_norm": 0.12788911163806915,
      "learning_rate": 0.00019586460162938622,
      "loss": 0.8374,
      "step": 624
    },
    {
      "epoch": 0.0933671945025396,
      "grad_norm": 0.1606101393699646,
      "learning_rate": 0.00019585121430708012,
      "loss": 0.6427,
      "step": 625
    },
    {
      "epoch": 0.09351658201374365,
      "grad_norm": 0.1087862104177475,
      "learning_rate": 0.00019583780580972253,
      "loss": 0.5977,
      "step": 626
    },
    {
      "epoch": 0.09366596952494771,
      "grad_norm": 0.1577569991350174,
      "learning_rate": 0.00019582437614027565,
      "loss": 0.7819,
      "step": 627
    },
    {
      "epoch": 0.09381535703615178,
      "grad_norm": 0.17475225031375885,
      "learning_rate": 0.00019581092530170633,
      "loss": 1.0605,
      "step": 628
    },
    {
      "epoch": 0.09396474454735584,
      "grad_norm": 0.12946583330631256,
      "learning_rate": 0.000195797453296986,
      "loss": 0.832,
      "step": 629
    },
    {
      "epoch": 0.09411413205855991,
      "grad_norm": 0.14866258203983307,
      "learning_rate": 0.00019578396012909092,
      "loss": 0.7355,
      "step": 630
    },
    {
      "epoch": 0.09426351956976396,
      "grad_norm": 0.13466285169124603,
      "learning_rate": 0.00019577044580100189,
      "loss": 0.6529,
      "step": 631
    },
    {
      "epoch": 0.09441290708096803,
      "grad_norm": 0.11548906564712524,
      "learning_rate": 0.00019575691031570446,
      "loss": 0.7818,
      "step": 632
    },
    {
      "epoch": 0.0945622945921721,
      "grad_norm": 0.18528424203395844,
      "learning_rate": 0.00019574335367618883,
      "loss": 1.0789,
      "step": 633
    },
    {
      "epoch": 0.09471168210337616,
      "grad_norm": 0.1468898057937622,
      "learning_rate": 0.00019572977588544986,
      "loss": 0.9756,
      "step": 634
    },
    {
      "epoch": 0.09486106961458023,
      "grad_norm": 0.16121335327625275,
      "learning_rate": 0.00019571617694648713,
      "loss": 0.7288,
      "step": 635
    },
    {
      "epoch": 0.09501045712578428,
      "grad_norm": 0.12135611474514008,
      "learning_rate": 0.00019570255686230485,
      "loss": 0.5934,
      "step": 636
    },
    {
      "epoch": 0.09515984463698834,
      "grad_norm": 0.12789905071258545,
      "learning_rate": 0.0001956889156359119,
      "loss": 0.6069,
      "step": 637
    },
    {
      "epoch": 0.09530923214819241,
      "grad_norm": 0.359639436006546,
      "learning_rate": 0.00019567525327032187,
      "loss": 1.3909,
      "step": 638
    },
    {
      "epoch": 0.09545861965939648,
      "grad_norm": 0.18724483251571655,
      "learning_rate": 0.00019566156976855297,
      "loss": 0.6935,
      "step": 639
    },
    {
      "epoch": 0.09560800717060054,
      "grad_norm": 0.19026364386081696,
      "learning_rate": 0.0001956478651336281,
      "loss": 0.7966,
      "step": 640
    },
    {
      "epoch": 0.0957573946818046,
      "grad_norm": 0.26914387941360474,
      "learning_rate": 0.00019563413936857484,
      "loss": 0.8772,
      "step": 641
    },
    {
      "epoch": 0.09590678219300866,
      "grad_norm": 0.1765991449356079,
      "learning_rate": 0.00019562039247642546,
      "loss": 0.7672,
      "step": 642
    },
    {
      "epoch": 0.09605616970421273,
      "grad_norm": 0.2534937560558319,
      "learning_rate": 0.00019560662446021677,
      "loss": 1.2049,
      "step": 643
    },
    {
      "epoch": 0.09620555721541679,
      "grad_norm": 0.14177022874355316,
      "learning_rate": 0.00019559283532299043,
      "loss": 0.5386,
      "step": 644
    },
    {
      "epoch": 0.09635494472662086,
      "grad_norm": 0.1188812330365181,
      "learning_rate": 0.00019557902506779268,
      "loss": 0.6466,
      "step": 645
    },
    {
      "epoch": 0.09650433223782492,
      "grad_norm": 0.12482491880655289,
      "learning_rate": 0.00019556519369767438,
      "loss": 0.6614,
      "step": 646
    },
    {
      "epoch": 0.09665371974902898,
      "grad_norm": 0.1450030505657196,
      "learning_rate": 0.00019555134121569112,
      "loss": 0.7014,
      "step": 647
    },
    {
      "epoch": 0.09680310726023304,
      "grad_norm": 0.12003795802593231,
      "learning_rate": 0.0001955374676249031,
      "loss": 0.8168,
      "step": 648
    },
    {
      "epoch": 0.09695249477143711,
      "grad_norm": 0.1340424120426178,
      "learning_rate": 0.0001955235729283753,
      "loss": 0.9613,
      "step": 649
    },
    {
      "epoch": 0.09710188228264117,
      "grad_norm": 0.19842645525932312,
      "learning_rate": 0.0001955096571291772,
      "loss": 0.8049,
      "step": 650
    },
    {
      "epoch": 0.09725126979384524,
      "grad_norm": 0.256247341632843,
      "learning_rate": 0.00019549572023038305,
      "loss": 1.3097,
      "step": 651
    },
    {
      "epoch": 0.09740065730504929,
      "grad_norm": 0.4698760509490967,
      "learning_rate": 0.0001954817622350717,
      "loss": 1.6247,
      "step": 652
    },
    {
      "epoch": 0.09755004481625336,
      "grad_norm": 0.16017255187034607,
      "learning_rate": 0.00019546778314632674,
      "loss": 0.9872,
      "step": 653
    },
    {
      "epoch": 0.09769943232745743,
      "grad_norm": 0.13327959179878235,
      "learning_rate": 0.00019545378296723635,
      "loss": 0.8846,
      "step": 654
    },
    {
      "epoch": 0.09784881983866149,
      "grad_norm": 0.3522357642650604,
      "learning_rate": 0.0001954397617008934,
      "loss": 1.2639,
      "step": 655
    },
    {
      "epoch": 0.09799820734986556,
      "grad_norm": 0.14365145564079285,
      "learning_rate": 0.0001954257193503954,
      "loss": 0.6805,
      "step": 656
    },
    {
      "epoch": 0.09814759486106961,
      "grad_norm": 0.09565582871437073,
      "learning_rate": 0.00019541165591884454,
      "loss": 0.5522,
      "step": 657
    },
    {
      "epoch": 0.09829698237227368,
      "grad_norm": 0.11068779975175858,
      "learning_rate": 0.0001953975714093476,
      "loss": 0.6697,
      "step": 658
    },
    {
      "epoch": 0.09844636988347774,
      "grad_norm": 0.15267115831375122,
      "learning_rate": 0.00019538346582501616,
      "loss": 0.743,
      "step": 659
    },
    {
      "epoch": 0.09859575739468181,
      "grad_norm": 0.13380476832389832,
      "learning_rate": 0.00019536933916896633,
      "loss": 0.7273,
      "step": 660
    },
    {
      "epoch": 0.09874514490588587,
      "grad_norm": 0.37871164083480835,
      "learning_rate": 0.0001953551914443189,
      "loss": 0.9902,
      "step": 661
    },
    {
      "epoch": 0.09889453241708993,
      "grad_norm": 0.16891950368881226,
      "learning_rate": 0.00019534102265419932,
      "loss": 1.0287,
      "step": 662
    },
    {
      "epoch": 0.09904391992829399,
      "grad_norm": 0.15239520370960236,
      "learning_rate": 0.00019532683280173768,
      "loss": 0.9578,
      "step": 663
    },
    {
      "epoch": 0.09919330743949806,
      "grad_norm": 0.1469845324754715,
      "learning_rate": 0.00019531262189006882,
      "loss": 0.8758,
      "step": 664
    },
    {
      "epoch": 0.09934269495070212,
      "grad_norm": 0.15531109273433685,
      "learning_rate": 0.00019529838992233208,
      "loss": 1.0518,
      "step": 665
    },
    {
      "epoch": 0.09949208246190619,
      "grad_norm": 0.19832894206047058,
      "learning_rate": 0.0001952841369016716,
      "loss": 0.6697,
      "step": 666
    },
    {
      "epoch": 0.09964146997311024,
      "grad_norm": 0.12545038759708405,
      "learning_rate": 0.00019526986283123601,
      "loss": 0.9188,
      "step": 667
    },
    {
      "epoch": 0.09979085748431431,
      "grad_norm": 0.15024740993976593,
      "learning_rate": 0.00019525556771417875,
      "loss": 0.9019,
      "step": 668
    },
    {
      "epoch": 0.09994024499551837,
      "grad_norm": 0.17119628190994263,
      "learning_rate": 0.0001952412515536578,
      "loss": 0.7931,
      "step": 669
    },
    {
      "epoch": 0.10008963250672244,
      "grad_norm": 0.2737690806388855,
      "learning_rate": 0.00019522691435283585,
      "loss": 1.1618,
      "step": 670
    },
    {
      "epoch": 0.1002390200179265,
      "grad_norm": 0.16071948409080505,
      "learning_rate": 0.00019521255611488022,
      "loss": 0.4868,
      "step": 671
    },
    {
      "epoch": 0.10038840752913056,
      "grad_norm": 0.14076560735702515,
      "learning_rate": 0.00019519817684296285,
      "loss": 0.6771,
      "step": 672
    },
    {
      "epoch": 0.10053779504033462,
      "grad_norm": 0.1131826639175415,
      "learning_rate": 0.00019518377654026032,
      "loss": 0.768,
      "step": 673
    },
    {
      "epoch": 0.10068718255153869,
      "grad_norm": 0.15727178752422333,
      "learning_rate": 0.00019516935520995393,
      "loss": 0.8779,
      "step": 674
    },
    {
      "epoch": 0.10083657006274276,
      "grad_norm": 0.14138628542423248,
      "learning_rate": 0.00019515491285522957,
      "loss": 0.8227,
      "step": 675
    },
    {
      "epoch": 0.10098595757394682,
      "grad_norm": 0.2561928331851959,
      "learning_rate": 0.0001951404494792778,
      "loss": 0.9701,
      "step": 676
    },
    {
      "epoch": 0.10113534508515089,
      "grad_norm": 0.22002875804901123,
      "learning_rate": 0.00019512596508529378,
      "loss": 0.8702,
      "step": 677
    },
    {
      "epoch": 0.10128473259635494,
      "grad_norm": 0.2049333155155182,
      "learning_rate": 0.00019511145967647737,
      "loss": 0.8744,
      "step": 678
    },
    {
      "epoch": 0.101434120107559,
      "grad_norm": 0.19357016682624817,
      "learning_rate": 0.00019509693325603302,
      "loss": 0.6567,
      "step": 679
    },
    {
      "epoch": 0.10158350761876307,
      "grad_norm": 0.14756688475608826,
      "learning_rate": 0.00019508238582716984,
      "loss": 0.7318,
      "step": 680
    },
    {
      "epoch": 0.10173289512996714,
      "grad_norm": 0.16614054143428802,
      "learning_rate": 0.00019506781739310163,
      "loss": 0.9397,
      "step": 681
    },
    {
      "epoch": 0.1018822826411712,
      "grad_norm": 0.11248838156461716,
      "learning_rate": 0.00019505322795704676,
      "loss": 0.7042,
      "step": 682
    },
    {
      "epoch": 0.10203167015237526,
      "grad_norm": 0.14995478093624115,
      "learning_rate": 0.00019503861752222826,
      "loss": 0.7761,
      "step": 683
    },
    {
      "epoch": 0.10218105766357932,
      "grad_norm": 0.15232588350772858,
      "learning_rate": 0.0001950239860918738,
      "loss": 0.7433,
      "step": 684
    },
    {
      "epoch": 0.10233044517478339,
      "grad_norm": 0.17888033390045166,
      "learning_rate": 0.00019500933366921571,
      "loss": 1.0783,
      "step": 685
    },
    {
      "epoch": 0.10247983268598745,
      "grad_norm": 0.16491816937923431,
      "learning_rate": 0.00019499466025749097,
      "loss": 0.6902,
      "step": 686
    },
    {
      "epoch": 0.10262922019719152,
      "grad_norm": 0.29704946279525757,
      "learning_rate": 0.00019497996585994112,
      "loss": 0.8224,
      "step": 687
    },
    {
      "epoch": 0.10277860770839557,
      "grad_norm": 0.13493022322654724,
      "learning_rate": 0.00019496525047981242,
      "loss": 0.6601,
      "step": 688
    },
    {
      "epoch": 0.10292799521959964,
      "grad_norm": 0.24183045327663422,
      "learning_rate": 0.0001949505141203557,
      "loss": 1.0734,
      "step": 689
    },
    {
      "epoch": 0.1030773827308037,
      "grad_norm": 0.41064098477363586,
      "learning_rate": 0.00019493575678482649,
      "loss": 1.1831,
      "step": 690
    },
    {
      "epoch": 0.10322677024200777,
      "grad_norm": 0.21319490671157837,
      "learning_rate": 0.0001949209784764849,
      "loss": 0.9954,
      "step": 691
    },
    {
      "epoch": 0.10337615775321184,
      "grad_norm": 0.15545840561389923,
      "learning_rate": 0.0001949061791985957,
      "loss": 0.831,
      "step": 692
    },
    {
      "epoch": 0.10352554526441589,
      "grad_norm": 0.13938447833061218,
      "learning_rate": 0.00019489135895442826,
      "loss": 0.8778,
      "step": 693
    },
    {
      "epoch": 0.10367493277561995,
      "grad_norm": 0.27140700817108154,
      "learning_rate": 0.00019487651774725663,
      "loss": 0.7168,
      "step": 694
    },
    {
      "epoch": 0.10382432028682402,
      "grad_norm": 0.17713186144828796,
      "learning_rate": 0.00019486165558035948,
      "loss": 0.7476,
      "step": 695
    },
    {
      "epoch": 0.10397370779802809,
      "grad_norm": 0.19945883750915527,
      "learning_rate": 0.00019484677245702004,
      "loss": 0.5964,
      "step": 696
    },
    {
      "epoch": 0.10412309530923215,
      "grad_norm": 0.27336767315864563,
      "learning_rate": 0.0001948318683805263,
      "loss": 1.0354,
      "step": 697
    },
    {
      "epoch": 0.1042724828204362,
      "grad_norm": 0.15666867792606354,
      "learning_rate": 0.0001948169433541708,
      "loss": 0.6364,
      "step": 698
    },
    {
      "epoch": 0.10442187033164027,
      "grad_norm": 0.17508071660995483,
      "learning_rate": 0.00019480199738125068,
      "loss": 0.7329,
      "step": 699
    },
    {
      "epoch": 0.10457125784284434,
      "grad_norm": 0.16016682982444763,
      "learning_rate": 0.00019478703046506773,
      "loss": 0.6733,
      "step": 700
    },
    {
      "epoch": 0.1047206453540484,
      "grad_norm": 0.10869868099689484,
      "learning_rate": 0.0001947720426089284,
      "loss": 0.6218,
      "step": 701
    },
    {
      "epoch": 0.10487003286525247,
      "grad_norm": 0.37434107065200806,
      "learning_rate": 0.00019475703381614375,
      "loss": 1.1511,
      "step": 702
    },
    {
      "epoch": 0.10501942037645653,
      "grad_norm": 0.2148595005273819,
      "learning_rate": 0.00019474200409002945,
      "loss": 1.0165,
      "step": 703
    },
    {
      "epoch": 0.10516880788766059,
      "grad_norm": 0.1994418203830719,
      "learning_rate": 0.00019472695343390585,
      "loss": 0.9581,
      "step": 704
    },
    {
      "epoch": 0.10531819539886465,
      "grad_norm": 0.14765793085098267,
      "learning_rate": 0.0001947118818510978,
      "loss": 0.9704,
      "step": 705
    },
    {
      "epoch": 0.10546758291006872,
      "grad_norm": 0.3004849851131439,
      "learning_rate": 0.00019469678934493488,
      "loss": 1.3147,
      "step": 706
    },
    {
      "epoch": 0.10561697042127279,
      "grad_norm": 0.14334464073181152,
      "learning_rate": 0.0001946816759187513,
      "loss": 0.5802,
      "step": 707
    },
    {
      "epoch": 0.10576635793247685,
      "grad_norm": 0.20262622833251953,
      "learning_rate": 0.0001946665415758858,
      "loss": 1.0572,
      "step": 708
    },
    {
      "epoch": 0.1059157454436809,
      "grad_norm": 0.1506635844707489,
      "learning_rate": 0.00019465138631968184,
      "loss": 1.264,
      "step": 709
    },
    {
      "epoch": 0.10606513295488497,
      "grad_norm": 0.1906273066997528,
      "learning_rate": 0.00019463621015348748,
      "loss": 0.9111,
      "step": 710
    },
    {
      "epoch": 0.10621452046608904,
      "grad_norm": 0.14025966823101044,
      "learning_rate": 0.0001946210130806553,
      "loss": 0.7685,
      "step": 711
    },
    {
      "epoch": 0.1063639079772931,
      "grad_norm": 0.13425306975841522,
      "learning_rate": 0.00019460579510454263,
      "loss": 0.8569,
      "step": 712
    },
    {
      "epoch": 0.10651329548849717,
      "grad_norm": 0.2626805305480957,
      "learning_rate": 0.0001945905562285113,
      "loss": 1.1358,
      "step": 713
    },
    {
      "epoch": 0.10666268299970122,
      "grad_norm": 0.12032942473888397,
      "learning_rate": 0.00019457529645592792,
      "loss": 0.9112,
      "step": 714
    },
    {
      "epoch": 0.10681207051090529,
      "grad_norm": 0.16017121076583862,
      "learning_rate": 0.0001945600157901635,
      "loss": 0.4619,
      "step": 715
    },
    {
      "epoch": 0.10696145802210935,
      "grad_norm": 0.1190473660826683,
      "learning_rate": 0.00019454471423459389,
      "loss": 0.7032,
      "step": 716
    },
    {
      "epoch": 0.10711084553331342,
      "grad_norm": 0.15523146092891693,
      "learning_rate": 0.00019452939179259937,
      "loss": 0.6421,
      "step": 717
    },
    {
      "epoch": 0.10726023304451748,
      "grad_norm": 0.27652865648269653,
      "learning_rate": 0.00019451404846756494,
      "loss": 0.8187,
      "step": 718
    },
    {
      "epoch": 0.10740962055572154,
      "grad_norm": 0.2987957000732422,
      "learning_rate": 0.00019449868426288017,
      "loss": 0.8263,
      "step": 719
    },
    {
      "epoch": 0.1075590080669256,
      "grad_norm": 0.143573597073555,
      "learning_rate": 0.00019448329918193927,
      "loss": 0.7988,
      "step": 720
    },
    {
      "epoch": 0.10770839557812967,
      "grad_norm": 0.1809718906879425,
      "learning_rate": 0.00019446789322814106,
      "loss": 0.7172,
      "step": 721
    },
    {
      "epoch": 0.10785778308933373,
      "grad_norm": 0.3178151845932007,
      "learning_rate": 0.00019445246640488893,
      "loss": 0.8902,
      "step": 722
    },
    {
      "epoch": 0.1080071706005378,
      "grad_norm": 0.17539192736148834,
      "learning_rate": 0.00019443701871559092,
      "loss": 1.0598,
      "step": 723
    },
    {
      "epoch": 0.10815655811174185,
      "grad_norm": 0.1401323676109314,
      "learning_rate": 0.00019442155016365965,
      "loss": 0.772,
      "step": 724
    },
    {
      "epoch": 0.10830594562294592,
      "grad_norm": 0.14516496658325195,
      "learning_rate": 0.0001944060607525124,
      "loss": 1.0332,
      "step": 725
    },
    {
      "epoch": 0.10845533313414998,
      "grad_norm": 0.12078884243965149,
      "learning_rate": 0.00019439055048557101,
      "loss": 0.4937,
      "step": 726
    },
    {
      "epoch": 0.10860472064535405,
      "grad_norm": 0.12402236461639404,
      "learning_rate": 0.00019437501936626198,
      "loss": 0.8666,
      "step": 727
    },
    {
      "epoch": 0.10875410815655812,
      "grad_norm": 0.2173183113336563,
      "learning_rate": 0.00019435946739801633,
      "loss": 0.7866,
      "step": 728
    },
    {
      "epoch": 0.10890349566776218,
      "grad_norm": 0.18823803961277008,
      "learning_rate": 0.00019434389458426976,
      "loss": 0.6379,
      "step": 729
    },
    {
      "epoch": 0.10905288317896623,
      "grad_norm": 0.1408422738313675,
      "learning_rate": 0.00019432830092846253,
      "loss": 0.7941,
      "step": 730
    },
    {
      "epoch": 0.1092022706901703,
      "grad_norm": 0.11038701981306076,
      "learning_rate": 0.00019431268643403958,
      "loss": 0.7818,
      "step": 731
    },
    {
      "epoch": 0.10935165820137437,
      "grad_norm": 0.2575507164001465,
      "learning_rate": 0.0001942970511044503,
      "loss": 0.8922,
      "step": 732
    },
    {
      "epoch": 0.10950104571257843,
      "grad_norm": 0.17680765688419342,
      "learning_rate": 0.00019428139494314888,
      "loss": 0.9351,
      "step": 733
    },
    {
      "epoch": 0.1096504332237825,
      "grad_norm": 0.13490507006645203,
      "learning_rate": 0.00019426571795359398,
      "loss": 0.8378,
      "step": 734
    },
    {
      "epoch": 0.10979982073498655,
      "grad_norm": 0.28408023715019226,
      "learning_rate": 0.0001942500201392489,
      "loss": 0.7867,
      "step": 735
    },
    {
      "epoch": 0.10994920824619062,
      "grad_norm": 0.16506126523017883,
      "learning_rate": 0.0001942343015035815,
      "loss": 1.1124,
      "step": 736
    },
    {
      "epoch": 0.11009859575739468,
      "grad_norm": 0.2039964199066162,
      "learning_rate": 0.00019421856205006433,
      "loss": 0.9931,
      "step": 737
    },
    {
      "epoch": 0.11024798326859875,
      "grad_norm": 0.14856329560279846,
      "learning_rate": 0.00019420280178217443,
      "loss": 0.7827,
      "step": 738
    },
    {
      "epoch": 0.11039737077980281,
      "grad_norm": 0.1818045973777771,
      "learning_rate": 0.00019418702070339355,
      "loss": 0.5399,
      "step": 739
    },
    {
      "epoch": 0.11054675829100687,
      "grad_norm": 0.2717982232570648,
      "learning_rate": 0.00019417121881720793,
      "loss": 1.2213,
      "step": 740
    },
    {
      "epoch": 0.11069614580221093,
      "grad_norm": 0.12363403290510178,
      "learning_rate": 0.0001941553961271085,
      "loss": 0.7174,
      "step": 741
    },
    {
      "epoch": 0.110845533313415,
      "grad_norm": 0.22691577672958374,
      "learning_rate": 0.0001941395526365907,
      "loss": 1.1225,
      "step": 742
    },
    {
      "epoch": 0.11099492082461906,
      "grad_norm": 0.16864041984081268,
      "learning_rate": 0.00019412368834915462,
      "loss": 0.5674,
      "step": 743
    },
    {
      "epoch": 0.11114430833582313,
      "grad_norm": 0.20602700114250183,
      "learning_rate": 0.00019410780326830498,
      "loss": 0.9028,
      "step": 744
    },
    {
      "epoch": 0.11129369584702718,
      "grad_norm": 0.13901367783546448,
      "learning_rate": 0.00019409189739755096,
      "loss": 0.6309,
      "step": 745
    },
    {
      "epoch": 0.11144308335823125,
      "grad_norm": 0.20725879073143005,
      "learning_rate": 0.0001940759707404065,
      "loss": 0.5478,
      "step": 746
    },
    {
      "epoch": 0.11159247086943531,
      "grad_norm": 0.14670120179653168,
      "learning_rate": 0.00019406002330039001,
      "loss": 0.5592,
      "step": 747
    },
    {
      "epoch": 0.11174185838063938,
      "grad_norm": 0.17827396094799042,
      "learning_rate": 0.00019404405508102455,
      "loss": 0.8181,
      "step": 748
    },
    {
      "epoch": 0.11189124589184345,
      "grad_norm": 0.1741906702518463,
      "learning_rate": 0.00019402806608583774,
      "loss": 1.1636,
      "step": 749
    },
    {
      "epoch": 0.1120406334030475,
      "grad_norm": 0.12306499481201172,
      "learning_rate": 0.00019401205631836178,
      "loss": 0.7564,
      "step": 750
    },
    {
      "epoch": 0.11219002091425156,
      "grad_norm": 0.12863090634346008,
      "learning_rate": 0.00019399602578213353,
      "loss": 0.739,
      "step": 751
    },
    {
      "epoch": 0.11233940842545563,
      "grad_norm": 0.11944539844989777,
      "learning_rate": 0.00019397997448069435,
      "loss": 0.5797,
      "step": 752
    },
    {
      "epoch": 0.1124887959366597,
      "grad_norm": 0.11944981664419174,
      "learning_rate": 0.00019396390241759024,
      "loss": 0.7503,
      "step": 753
    },
    {
      "epoch": 0.11263818344786376,
      "grad_norm": 0.12803085148334503,
      "learning_rate": 0.00019394780959637177,
      "loss": 0.984,
      "step": 754
    },
    {
      "epoch": 0.11278757095906783,
      "grad_norm": 8.864607810974121,
      "learning_rate": 0.0001939316960205941,
      "loss": 4.3697,
      "step": 755
    },
    {
      "epoch": 0.11293695847027188,
      "grad_norm": 0.15747593343257904,
      "learning_rate": 0.000193915561693817,
      "loss": 0.7344,
      "step": 756
    },
    {
      "epoch": 0.11308634598147595,
      "grad_norm": 0.13283927738666534,
      "learning_rate": 0.00019389940661960478,
      "loss": 0.5302,
      "step": 757
    },
    {
      "epoch": 0.11323573349268001,
      "grad_norm": 0.12874341011047363,
      "learning_rate": 0.00019388323080152633,
      "loss": 0.828,
      "step": 758
    },
    {
      "epoch": 0.11338512100388408,
      "grad_norm": 0.4698888063430786,
      "learning_rate": 0.00019386703424315518,
      "loss": 1.5101,
      "step": 759
    },
    {
      "epoch": 0.11353450851508815,
      "grad_norm": 0.12892121076583862,
      "learning_rate": 0.00019385081694806936,
      "loss": 0.7772,
      "step": 760
    },
    {
      "epoch": 0.1136838960262922,
      "grad_norm": 0.28520870208740234,
      "learning_rate": 0.00019383457891985158,
      "loss": 0.928,
      "step": 761
    },
    {
      "epoch": 0.11383328353749626,
      "grad_norm": 1.7955225706100464,
      "learning_rate": 0.00019381832016208904,
      "loss": 0.955,
      "step": 762
    },
    {
      "epoch": 0.11398267104870033,
      "grad_norm": 0.14767761528491974,
      "learning_rate": 0.00019380204067837356,
      "loss": 0.6337,
      "step": 763
    },
    {
      "epoch": 0.1141320585599044,
      "grad_norm": 0.171101912856102,
      "learning_rate": 0.0001937857404723016,
      "loss": 0.7905,
      "step": 764
    },
    {
      "epoch": 0.11428144607110846,
      "grad_norm": 0.33381524682044983,
      "learning_rate": 0.000193769419547474,
      "loss": 0.9788,
      "step": 765
    },
    {
      "epoch": 0.11443083358231251,
      "grad_norm": 0.1214352399110794,
      "learning_rate": 0.00019375307790749647,
      "loss": 0.9152,
      "step": 766
    },
    {
      "epoch": 0.11458022109351658,
      "grad_norm": 0.3238249123096466,
      "learning_rate": 0.00019373671555597902,
      "loss": 0.9923,
      "step": 767
    },
    {
      "epoch": 0.11472960860472065,
      "grad_norm": 0.14854952692985535,
      "learning_rate": 0.0001937203324965364,
      "loss": 0.7194,
      "step": 768
    },
    {
      "epoch": 0.11487899611592471,
      "grad_norm": 0.13817191123962402,
      "learning_rate": 0.00019370392873278784,
      "loss": 0.7175,
      "step": 769
    },
    {
      "epoch": 0.11502838362712878,
      "grad_norm": 0.24061405658721924,
      "learning_rate": 0.0001936875042683573,
      "loss": 0.9647,
      "step": 770
    },
    {
      "epoch": 0.11517777113833283,
      "grad_norm": 0.1889171153306961,
      "learning_rate": 0.00019367105910687307,
      "loss": 0.9805,
      "step": 771
    },
    {
      "epoch": 0.1153271586495369,
      "grad_norm": 0.2141283005475998,
      "learning_rate": 0.00019365459325196825,
      "loss": 0.7467,
      "step": 772
    },
    {
      "epoch": 0.11547654616074096,
      "grad_norm": 0.20832307636737823,
      "learning_rate": 0.00019363810670728032,
      "loss": 0.8881,
      "step": 773
    },
    {
      "epoch": 0.11562593367194503,
      "grad_norm": 0.19508862495422363,
      "learning_rate": 0.00019362159947645152,
      "loss": 0.9594,
      "step": 774
    },
    {
      "epoch": 0.1157753211831491,
      "grad_norm": 0.16393844783306122,
      "learning_rate": 0.00019360507156312848,
      "loss": 0.8358,
      "step": 775
    },
    {
      "epoch": 0.11592470869435315,
      "grad_norm": 0.17555637657642365,
      "learning_rate": 0.00019358852297096253,
      "loss": 0.7137,
      "step": 776
    },
    {
      "epoch": 0.11607409620555721,
      "grad_norm": 0.16748878359794617,
      "learning_rate": 0.00019357195370360946,
      "loss": 0.8945,
      "step": 777
    },
    {
      "epoch": 0.11622348371676128,
      "grad_norm": 0.11165950447320938,
      "learning_rate": 0.00019355536376472972,
      "loss": 0.5322,
      "step": 778
    },
    {
      "epoch": 0.11637287122796534,
      "grad_norm": 0.12036951631307602,
      "learning_rate": 0.00019353875315798828,
      "loss": 0.824,
      "step": 779
    },
    {
      "epoch": 0.11652225873916941,
      "grad_norm": 0.2941713333129883,
      "learning_rate": 0.0001935221218870547,
      "loss": 1.1819,
      "step": 780
    },
    {
      "epoch": 0.11667164625037346,
      "grad_norm": 0.16016024351119995,
      "learning_rate": 0.0001935054699556031,
      "loss": 0.8635,
      "step": 781
    },
    {
      "epoch": 0.11682103376157753,
      "grad_norm": 0.16375142335891724,
      "learning_rate": 0.0001934887973673121,
      "loss": 1.0148,
      "step": 782
    },
    {
      "epoch": 0.1169704212727816,
      "grad_norm": 0.3563995659351349,
      "learning_rate": 0.000193472104125865,
      "loss": 0.9712,
      "step": 783
    },
    {
      "epoch": 0.11711980878398566,
      "grad_norm": 0.19017907977104187,
      "learning_rate": 0.0001934553902349496,
      "loss": 0.9978,
      "step": 784
    },
    {
      "epoch": 0.11726919629518973,
      "grad_norm": 0.14762148261070251,
      "learning_rate": 0.00019343865569825818,
      "loss": 0.8258,
      "step": 785
    },
    {
      "epoch": 0.11741858380639379,
      "grad_norm": 0.19044718146324158,
      "learning_rate": 0.00019342190051948777,
      "loss": 1.0279,
      "step": 786
    },
    {
      "epoch": 0.11756797131759784,
      "grad_norm": 0.15434981882572174,
      "learning_rate": 0.0001934051247023398,
      "loss": 0.6936,
      "step": 787
    },
    {
      "epoch": 0.11771735882880191,
      "grad_norm": 0.21782897412776947,
      "learning_rate": 0.0001933883282505203,
      "loss": 0.8074,
      "step": 788
    },
    {
      "epoch": 0.11786674634000598,
      "grad_norm": 0.1382855623960495,
      "learning_rate": 0.00019337151116773993,
      "loss": 0.76,
      "step": 789
    },
    {
      "epoch": 0.11801613385121004,
      "grad_norm": 0.12988682091236115,
      "learning_rate": 0.00019335467345771377,
      "loss": 0.7619,
      "step": 790
    },
    {
      "epoch": 0.11816552136241411,
      "grad_norm": 0.14587444067001343,
      "learning_rate": 0.0001933378151241616,
      "loss": 0.6618,
      "step": 791
    },
    {
      "epoch": 0.11831490887361816,
      "grad_norm": 0.13079139590263367,
      "learning_rate": 0.0001933209361708077,
      "loss": 0.7573,
      "step": 792
    },
    {
      "epoch": 0.11846429638482223,
      "grad_norm": 0.27695533633232117,
      "learning_rate": 0.00019330403660138085,
      "loss": 1.0731,
      "step": 793
    },
    {
      "epoch": 0.11861368389602629,
      "grad_norm": 0.1361035704612732,
      "learning_rate": 0.00019328711641961445,
      "loss": 0.8251,
      "step": 794
    },
    {
      "epoch": 0.11876307140723036,
      "grad_norm": 0.16837474703788757,
      "learning_rate": 0.00019327017562924644,
      "loss": 0.9535,
      "step": 795
    },
    {
      "epoch": 0.11891245891843442,
      "grad_norm": 0.31326931715011597,
      "learning_rate": 0.00019325321423401933,
      "loss": 0.9828,
      "step": 796
    },
    {
      "epoch": 0.11906184642963848,
      "grad_norm": 0.16937793791294098,
      "learning_rate": 0.00019323623223768015,
      "loss": 0.9445,
      "step": 797
    },
    {
      "epoch": 0.11921123394084254,
      "grad_norm": 0.15119615197181702,
      "learning_rate": 0.00019321922964398046,
      "loss": 0.7483,
      "step": 798
    },
    {
      "epoch": 0.11936062145204661,
      "grad_norm": 0.1750727891921997,
      "learning_rate": 0.00019320220645667645,
      "loss": 0.9364,
      "step": 799
    },
    {
      "epoch": 0.11951000896325067,
      "grad_norm": 0.22479920089244843,
      "learning_rate": 0.00019318516267952874,
      "loss": 0.854,
      "step": 800
    },
    {
      "epoch": 0.11965939647445474,
      "grad_norm": 0.12636275589466095,
      "learning_rate": 0.00019316809831630265,
      "loss": 0.8127,
      "step": 801
    },
    {
      "epoch": 0.11980878398565879,
      "grad_norm": 0.1225171908736229,
      "learning_rate": 0.00019315101337076792,
      "loss": 0.7809,
      "step": 802
    },
    {
      "epoch": 0.11995817149686286,
      "grad_norm": 0.24531179666519165,
      "learning_rate": 0.00019313390784669895,
      "loss": 1.0615,
      "step": 803
    },
    {
      "epoch": 0.12010755900806692,
      "grad_norm": 0.15142390131950378,
      "learning_rate": 0.0001931167817478745,
      "loss": 0.913,
      "step": 804
    },
    {
      "epoch": 0.12025694651927099,
      "grad_norm": 0.13152721524238586,
      "learning_rate": 0.00019309963507807811,
      "loss": 0.6284,
      "step": 805
    },
    {
      "epoch": 0.12040633403047506,
      "grad_norm": 0.10537417232990265,
      "learning_rate": 0.0001930824678410977,
      "loss": 0.6071,
      "step": 806
    },
    {
      "epoch": 0.12055572154167911,
      "grad_norm": 0.13866813480854034,
      "learning_rate": 0.0001930652800407258,
      "loss": 0.7461,
      "step": 807
    },
    {
      "epoch": 0.12070510905288317,
      "grad_norm": 0.16239172220230103,
      "learning_rate": 0.00019304807168075944,
      "loss": 0.8089,
      "step": 808
    },
    {
      "epoch": 0.12085449656408724,
      "grad_norm": 0.12834633886814117,
      "learning_rate": 0.00019303084276500027,
      "loss": 0.8459,
      "step": 809
    },
    {
      "epoch": 0.12100388407529131,
      "grad_norm": 0.17036044597625732,
      "learning_rate": 0.00019301359329725436,
      "loss": 0.8131,
      "step": 810
    },
    {
      "epoch": 0.12115327158649537,
      "grad_norm": 0.14274178445339203,
      "learning_rate": 0.00019299632328133247,
      "loss": 0.9325,
      "step": 811
    },
    {
      "epoch": 0.12130265909769944,
      "grad_norm": 0.16712737083435059,
      "learning_rate": 0.00019297903272104977,
      "loss": 0.88,
      "step": 812
    },
    {
      "epoch": 0.12145204660890349,
      "grad_norm": 0.1521551012992859,
      "learning_rate": 0.00019296172162022604,
      "loss": 0.9838,
      "step": 813
    },
    {
      "epoch": 0.12160143412010756,
      "grad_norm": 0.1645859032869339,
      "learning_rate": 0.00019294438998268554,
      "loss": 0.6344,
      "step": 814
    },
    {
      "epoch": 0.12175082163131162,
      "grad_norm": 0.15957693755626678,
      "learning_rate": 0.00019292703781225717,
      "loss": 0.926,
      "step": 815
    },
    {
      "epoch": 0.12190020914251569,
      "grad_norm": 0.13982924818992615,
      "learning_rate": 0.00019290966511277422,
      "loss": 0.849,
      "step": 816
    },
    {
      "epoch": 0.12204959665371976,
      "grad_norm": 0.16864308714866638,
      "learning_rate": 0.00019289227188807467,
      "loss": 0.6237,
      "step": 817
    },
    {
      "epoch": 0.12219898416492381,
      "grad_norm": 0.14285649359226227,
      "learning_rate": 0.00019287485814200087,
      "loss": 0.6809,
      "step": 818
    },
    {
      "epoch": 0.12234837167612787,
      "grad_norm": 0.6627392768859863,
      "learning_rate": 0.00019285742387839988,
      "loss": 1.9625,
      "step": 819
    },
    {
      "epoch": 0.12249775918733194,
      "grad_norm": 0.13897117972373962,
      "learning_rate": 0.00019283996910112318,
      "loss": 1.0312,
      "step": 820
    },
    {
      "epoch": 0.122647146698536,
      "grad_norm": 0.15520478785037994,
      "learning_rate": 0.00019282249381402677,
      "loss": 0.8191,
      "step": 821
    },
    {
      "epoch": 0.12279653420974007,
      "grad_norm": 0.24286359548568726,
      "learning_rate": 0.00019280499802097126,
      "loss": 1.1401,
      "step": 822
    },
    {
      "epoch": 0.12294592172094412,
      "grad_norm": 0.15701743960380554,
      "learning_rate": 0.00019278748172582173,
      "loss": 0.7389,
      "step": 823
    },
    {
      "epoch": 0.12309530923214819,
      "grad_norm": 0.16320520639419556,
      "learning_rate": 0.0001927699449324478,
      "loss": 0.9232,
      "step": 824
    },
    {
      "epoch": 0.12324469674335226,
      "grad_norm": 0.13166747987270355,
      "learning_rate": 0.00019275238764472364,
      "loss": 0.648,
      "step": 825
    },
    {
      "epoch": 0.12339408425455632,
      "grad_norm": 0.1490585058927536,
      "learning_rate": 0.00019273480986652794,
      "loss": 0.8485,
      "step": 826
    },
    {
      "epoch": 0.12354347176576039,
      "grad_norm": 0.1607678383588791,
      "learning_rate": 0.00019271721160174388,
      "loss": 0.7415,
      "step": 827
    },
    {
      "epoch": 0.12369285927696444,
      "grad_norm": 0.13538645207881927,
      "learning_rate": 0.0001926995928542592,
      "loss": 0.702,
      "step": 828
    },
    {
      "epoch": 0.1238422467881685,
      "grad_norm": 0.17759506404399872,
      "learning_rate": 0.00019268195362796622,
      "loss": 0.819,
      "step": 829
    },
    {
      "epoch": 0.12399163429937257,
      "grad_norm": 0.14529088139533997,
      "learning_rate": 0.00019266429392676164,
      "loss": 0.5262,
      "step": 830
    },
    {
      "epoch": 0.12414102181057664,
      "grad_norm": 0.16848066449165344,
      "learning_rate": 0.0001926466137545468,
      "loss": 0.6761,
      "step": 831
    },
    {
      "epoch": 0.1242904093217807,
      "grad_norm": 0.33541253209114075,
      "learning_rate": 0.00019262891311522755,
      "loss": 0.9293,
      "step": 832
    },
    {
      "epoch": 0.12443979683298476,
      "grad_norm": 0.2218908667564392,
      "learning_rate": 0.00019261119201271422,
      "loss": 1.0644,
      "step": 833
    },
    {
      "epoch": 0.12458918434418882,
      "grad_norm": 0.1724868267774582,
      "learning_rate": 0.0001925934504509217,
      "loss": 0.8411,
      "step": 834
    },
    {
      "epoch": 0.12473857185539289,
      "grad_norm": 0.12873922288417816,
      "learning_rate": 0.00019257568843376936,
      "loss": 0.7466,
      "step": 835
    },
    {
      "epoch": 0.12488795936659695,
      "grad_norm": 0.13556797802448273,
      "learning_rate": 0.00019255790596518112,
      "loss": 0.675,
      "step": 836
    },
    {
      "epoch": 0.12503734687780102,
      "grad_norm": 0.12460656464099884,
      "learning_rate": 0.00019254010304908543,
      "loss": 0.6401,
      "step": 837
    },
    {
      "epoch": 0.12518673438900507,
      "grad_norm": 0.16490042209625244,
      "learning_rate": 0.00019252227968941522,
      "loss": 0.8934,
      "step": 838
    },
    {
      "epoch": 0.12533612190020915,
      "grad_norm": 0.13448311388492584,
      "learning_rate": 0.00019250443589010792,
      "loss": 0.7896,
      "step": 839
    },
    {
      "epoch": 0.1254855094114132,
      "grad_norm": 0.16792979836463928,
      "learning_rate": 0.00019248657165510556,
      "loss": 0.74,
      "step": 840
    },
    {
      "epoch": 0.12563489692261726,
      "grad_norm": 0.1711784452199936,
      "learning_rate": 0.00019246868698835458,
      "loss": 0.5992,
      "step": 841
    },
    {
      "epoch": 0.12578428443382134,
      "grad_norm": 0.12499507516622543,
      "learning_rate": 0.00019245078189380604,
      "loss": 0.8517,
      "step": 842
    },
    {
      "epoch": 0.1259336719450254,
      "grad_norm": 0.09736009687185287,
      "learning_rate": 0.00019243285637541544,
      "loss": 0.6209,
      "step": 843
    },
    {
      "epoch": 0.12608305945622947,
      "grad_norm": 0.21913480758666992,
      "learning_rate": 0.0001924149104371428,
      "loss": 0.9499,
      "step": 844
    },
    {
      "epoch": 0.12623244696743352,
      "grad_norm": 0.23238122463226318,
      "learning_rate": 0.00019239694408295266,
      "loss": 1.1666,
      "step": 845
    },
    {
      "epoch": 0.12638183447863757,
      "grad_norm": 0.24529285728931427,
      "learning_rate": 0.00019237895731681408,
      "loss": 0.734,
      "step": 846
    },
    {
      "epoch": 0.12653122198984165,
      "grad_norm": 0.16577094793319702,
      "learning_rate": 0.00019236095014270064,
      "loss": 1.0808,
      "step": 847
    },
    {
      "epoch": 0.1266806095010457,
      "grad_norm": 0.08479466289281845,
      "learning_rate": 0.0001923429225645904,
      "loss": 0.4277,
      "step": 848
    },
    {
      "epoch": 0.12682999701224978,
      "grad_norm": 0.15633448958396912,
      "learning_rate": 0.00019232487458646587,
      "loss": 0.8547,
      "step": 849
    },
    {
      "epoch": 0.12697938452345384,
      "grad_norm": 0.2966431975364685,
      "learning_rate": 0.00019230680621231425,
      "loss": 0.7691,
      "step": 850
    },
    {
      "epoch": 0.12712877203465792,
      "grad_norm": 0.21817810833454132,
      "learning_rate": 0.00019228871744612704,
      "loss": 0.9089,
      "step": 851
    },
    {
      "epoch": 0.12727815954586197,
      "grad_norm": 0.1577853262424469,
      "learning_rate": 0.0001922706082919004,
      "loss": 0.7345,
      "step": 852
    },
    {
      "epoch": 0.12742754705706602,
      "grad_norm": 0.14371559023857117,
      "learning_rate": 0.00019225247875363487,
      "loss": 0.9727,
      "step": 853
    },
    {
      "epoch": 0.1275769345682701,
      "grad_norm": 0.10418539494276047,
      "learning_rate": 0.0001922343288353356,
      "loss": 0.6287,
      "step": 854
    },
    {
      "epoch": 0.12772632207947415,
      "grad_norm": 0.14414125680923462,
      "learning_rate": 0.00019221615854101215,
      "loss": 1.0211,
      "step": 855
    },
    {
      "epoch": 0.12787570959067823,
      "grad_norm": 0.14058874547481537,
      "learning_rate": 0.00019219796787467867,
      "loss": 0.9388,
      "step": 856
    },
    {
      "epoch": 0.12802509710188228,
      "grad_norm": 0.24741698801517487,
      "learning_rate": 0.00019217975684035374,
      "loss": 0.8997,
      "step": 857
    },
    {
      "epoch": 0.12817448461308634,
      "grad_norm": 0.1672530174255371,
      "learning_rate": 0.00019216152544206049,
      "loss": 0.9332,
      "step": 858
    },
    {
      "epoch": 0.12832387212429042,
      "grad_norm": 0.30283141136169434,
      "learning_rate": 0.00019214327368382645,
      "loss": 1.1119,
      "step": 859
    },
    {
      "epoch": 0.12847325963549447,
      "grad_norm": 0.41509029269218445,
      "learning_rate": 0.00019212500156968383,
      "loss": 1.0261,
      "step": 860
    },
    {
      "epoch": 0.12862264714669855,
      "grad_norm": 0.1252591907978058,
      "learning_rate": 0.00019210670910366917,
      "loss": 0.7836,
      "step": 861
    },
    {
      "epoch": 0.1287720346579026,
      "grad_norm": 0.1359296441078186,
      "learning_rate": 0.00019208839628982358,
      "loss": 0.9255,
      "step": 862
    },
    {
      "epoch": 0.12892142216910665,
      "grad_norm": 0.22711493074893951,
      "learning_rate": 0.0001920700631321926,
      "loss": 1.1905,
      "step": 863
    },
    {
      "epoch": 0.12907080968031073,
      "grad_norm": 0.13732872903347015,
      "learning_rate": 0.00019205170963482643,
      "loss": 0.7808,
      "step": 864
    },
    {
      "epoch": 0.12922019719151479,
      "grad_norm": 0.3582336902618408,
      "learning_rate": 0.00019203333580177954,
      "loss": 0.7557,
      "step": 865
    },
    {
      "epoch": 0.12936958470271887,
      "grad_norm": 0.14303429424762726,
      "learning_rate": 0.00019201494163711104,
      "loss": 0.898,
      "step": 866
    },
    {
      "epoch": 0.12951897221392292,
      "grad_norm": 0.11724136024713516,
      "learning_rate": 0.00019199652714488446,
      "loss": 0.6867,
      "step": 867
    },
    {
      "epoch": 0.12966835972512697,
      "grad_norm": 0.14605864882469177,
      "learning_rate": 0.00019197809232916795,
      "loss": 0.7361,
      "step": 868
    },
    {
      "epoch": 0.12981774723633105,
      "grad_norm": 0.15446977317333221,
      "learning_rate": 0.00019195963719403393,
      "loss": 0.7941,
      "step": 869
    },
    {
      "epoch": 0.1299671347475351,
      "grad_norm": 0.17148366570472717,
      "learning_rate": 0.00019194116174355954,
      "loss": 0.7493,
      "step": 870
    },
    {
      "epoch": 0.13011652225873918,
      "grad_norm": 0.12092535197734833,
      "learning_rate": 0.0001919226659818262,
      "loss": 0.8044,
      "step": 871
    },
    {
      "epoch": 0.13026590976994323,
      "grad_norm": 0.174177348613739,
      "learning_rate": 0.00019190414991291998,
      "loss": 0.8565,
      "step": 872
    },
    {
      "epoch": 0.13041529728114729,
      "grad_norm": 0.14217911660671234,
      "learning_rate": 0.0001918856135409314,
      "loss": 0.7967,
      "step": 873
    },
    {
      "epoch": 0.13056468479235137,
      "grad_norm": 0.2161906510591507,
      "learning_rate": 0.00019186705686995533,
      "loss": 1.0536,
      "step": 874
    },
    {
      "epoch": 0.13071407230355542,
      "grad_norm": 0.11044277995824814,
      "learning_rate": 0.00019184847990409134,
      "loss": 0.7443,
      "step": 875
    },
    {
      "epoch": 0.1308634598147595,
      "grad_norm": 0.4287470877170563,
      "learning_rate": 0.0001918298826474433,
      "loss": 1.1078,
      "step": 876
    },
    {
      "epoch": 0.13101284732596355,
      "grad_norm": 0.313516229391098,
      "learning_rate": 0.00019181126510411974,
      "loss": 1.1343,
      "step": 877
    },
    {
      "epoch": 0.1311622348371676,
      "grad_norm": 0.15673229098320007,
      "learning_rate": 0.0001917926272782334,
      "loss": 1.0282,
      "step": 878
    },
    {
      "epoch": 0.13131162234837168,
      "grad_norm": 0.14207200706005096,
      "learning_rate": 0.00019177396917390187,
      "loss": 0.7487,
      "step": 879
    },
    {
      "epoch": 0.13146100985957573,
      "grad_norm": 0.21555255353450775,
      "learning_rate": 0.00019175529079524687,
      "loss": 0.8816,
      "step": 880
    },
    {
      "epoch": 0.1316103973707798,
      "grad_norm": 0.10295391082763672,
      "learning_rate": 0.00019173659214639482,
      "loss": 0.7433,
      "step": 881
    },
    {
      "epoch": 0.13175978488198387,
      "grad_norm": 0.12901347875595093,
      "learning_rate": 0.00019171787323147654,
      "loss": 0.828,
      "step": 882
    },
    {
      "epoch": 0.13190917239318792,
      "grad_norm": 0.1917589008808136,
      "learning_rate": 0.00019169913405462733,
      "loss": 0.8845,
      "step": 883
    },
    {
      "epoch": 0.132058559904392,
      "grad_norm": 0.14522455632686615,
      "learning_rate": 0.00019168037461998695,
      "loss": 0.8876,
      "step": 884
    },
    {
      "epoch": 0.13220794741559605,
      "grad_norm": 0.33967098593711853,
      "learning_rate": 0.0001916615949316997,
      "loss": 0.93,
      "step": 885
    },
    {
      "epoch": 0.13235733492680013,
      "grad_norm": 0.12859250605106354,
      "learning_rate": 0.00019164279499391427,
      "loss": 0.8351,
      "step": 886
    },
    {
      "epoch": 0.13250672243800418,
      "grad_norm": 0.11414708197116852,
      "learning_rate": 0.00019162397481078386,
      "loss": 0.5253,
      "step": 887
    },
    {
      "epoch": 0.13265610994920823,
      "grad_norm": 0.6967369914054871,
      "learning_rate": 0.00019160513438646617,
      "loss": 1.0658,
      "step": 888
    },
    {
      "epoch": 0.13280549746041231,
      "grad_norm": 0.2920320928096771,
      "learning_rate": 0.00019158627372512337,
      "loss": 1.0202,
      "step": 889
    },
    {
      "epoch": 0.13295488497161637,
      "grad_norm": 0.1285315304994583,
      "learning_rate": 0.00019156739283092205,
      "loss": 0.5447,
      "step": 890
    },
    {
      "epoch": 0.13310427248282045,
      "grad_norm": 0.6523221135139465,
      "learning_rate": 0.00019154849170803327,
      "loss": 1.2813,
      "step": 891
    },
    {
      "epoch": 0.1332536599940245,
      "grad_norm": 0.11954151839017868,
      "learning_rate": 0.00019152957036063265,
      "loss": 0.7564,
      "step": 892
    },
    {
      "epoch": 0.13340304750522855,
      "grad_norm": 0.10571929812431335,
      "learning_rate": 0.0001915106287929002,
      "loss": 0.4487,
      "step": 893
    },
    {
      "epoch": 0.13355243501643263,
      "grad_norm": 0.24934682250022888,
      "learning_rate": 0.00019149166700902032,
      "loss": 0.7512,
      "step": 894
    },
    {
      "epoch": 0.13370182252763668,
      "grad_norm": 0.1705646514892578,
      "learning_rate": 0.00019147268501318212,
      "loss": 0.6647,
      "step": 895
    },
    {
      "epoch": 0.13385121003884076,
      "grad_norm": 0.1977798193693161,
      "learning_rate": 0.0001914536828095789,
      "loss": 1.0058,
      "step": 896
    },
    {
      "epoch": 0.13400059755004481,
      "grad_norm": 0.16323387622833252,
      "learning_rate": 0.00019143466040240863,
      "loss": 1.0301,
      "step": 897
    },
    {
      "epoch": 0.13414998506124887,
      "grad_norm": 0.15375447273254395,
      "learning_rate": 0.0001914156177958736,
      "loss": 0.5928,
      "step": 898
    },
    {
      "epoch": 0.13429937257245295,
      "grad_norm": 0.2192535251379013,
      "learning_rate": 0.00019139655499418067,
      "loss": 0.6105,
      "step": 899
    },
    {
      "epoch": 0.134448760083657,
      "grad_norm": 0.35383331775665283,
      "learning_rate": 0.0001913774720015411,
      "loss": 0.9193,
      "step": 900
    },
    {
      "epoch": 0.13459814759486108,
      "grad_norm": 0.2478116750717163,
      "learning_rate": 0.0001913583688221706,
      "loss": 1.0428,
      "step": 901
    },
    {
      "epoch": 0.13474753510606513,
      "grad_norm": 0.14307573437690735,
      "learning_rate": 0.00019133924546028942,
      "loss": 0.8198,
      "step": 902
    },
    {
      "epoch": 0.13489692261726918,
      "grad_norm": 0.27853766083717346,
      "learning_rate": 0.00019132010192012214,
      "loss": 0.9841,
      "step": 903
    },
    {
      "epoch": 0.13504631012847326,
      "grad_norm": 0.13049887120723724,
      "learning_rate": 0.00019130093820589791,
      "loss": 0.5867,
      "step": 904
    },
    {
      "epoch": 0.13519569763967731,
      "grad_norm": 0.1877085268497467,
      "learning_rate": 0.0001912817543218503,
      "loss": 1.0579,
      "step": 905
    },
    {
      "epoch": 0.1353450851508814,
      "grad_norm": 0.18851879239082336,
      "learning_rate": 0.00019126255027221735,
      "loss": 0.5173,
      "step": 906
    },
    {
      "epoch": 0.13549447266208545,
      "grad_norm": 0.1649446338415146,
      "learning_rate": 0.00019124332606124152,
      "loss": 0.8469,
      "step": 907
    },
    {
      "epoch": 0.13564386017328953,
      "grad_norm": 0.1373070925474167,
      "learning_rate": 0.00019122408169316976,
      "loss": 0.7754,
      "step": 908
    },
    {
      "epoch": 0.13579324768449358,
      "grad_norm": 0.13052485883235931,
      "learning_rate": 0.00019120481717225342,
      "loss": 0.9105,
      "step": 909
    },
    {
      "epoch": 0.13594263519569763,
      "grad_norm": 0.15370683372020721,
      "learning_rate": 0.00019118553250274832,
      "loss": 0.6646,
      "step": 910
    },
    {
      "epoch": 0.1360920227069017,
      "grad_norm": 0.1601782590150833,
      "learning_rate": 0.00019116622768891483,
      "loss": 0.7487,
      "step": 911
    },
    {
      "epoch": 0.13624141021810576,
      "grad_norm": 0.19794587790966034,
      "learning_rate": 0.00019114690273501765,
      "loss": 0.5437,
      "step": 912
    },
    {
      "epoch": 0.13639079772930984,
      "grad_norm": 0.12566497921943665,
      "learning_rate": 0.00019112755764532594,
      "loss": 0.8602,
      "step": 913
    },
    {
      "epoch": 0.1365401852405139,
      "grad_norm": 0.5696573257446289,
      "learning_rate": 0.00019110819242411337,
      "loss": 1.1799,
      "step": 914
    },
    {
      "epoch": 0.13668957275171795,
      "grad_norm": 0.11676638573408127,
      "learning_rate": 0.00019108880707565802,
      "loss": 0.6958,
      "step": 915
    },
    {
      "epoch": 0.13683896026292203,
      "grad_norm": 0.12003248929977417,
      "learning_rate": 0.00019106940160424244,
      "loss": 0.5425,
      "step": 916
    },
    {
      "epoch": 0.13698834777412608,
      "grad_norm": 0.265358030796051,
      "learning_rate": 0.00019104997601415352,
      "loss": 0.673,
      "step": 917
    },
    {
      "epoch": 0.13713773528533016,
      "grad_norm": 0.17708387970924377,
      "learning_rate": 0.0001910305303096828,
      "loss": 0.7885,
      "step": 918
    },
    {
      "epoch": 0.1372871227965342,
      "grad_norm": 0.17588114738464355,
      "learning_rate": 0.00019101106449512605,
      "loss": 0.8977,
      "step": 919
    },
    {
      "epoch": 0.13743651030773826,
      "grad_norm": 0.20153087377548218,
      "learning_rate": 0.0001909915785747836,
      "loss": 0.6529,
      "step": 920
    },
    {
      "epoch": 0.13758589781894234,
      "grad_norm": 0.14944609999656677,
      "learning_rate": 0.00019097207255296022,
      "loss": 0.7525,
      "step": 921
    },
    {
      "epoch": 0.1377352853301464,
      "grad_norm": 0.29275017976760864,
      "learning_rate": 0.00019095254643396512,
      "loss": 1.0195,
      "step": 922
    },
    {
      "epoch": 0.13788467284135048,
      "grad_norm": 0.26294517517089844,
      "learning_rate": 0.00019093300022211186,
      "loss": 1.2764,
      "step": 923
    },
    {
      "epoch": 0.13803406035255453,
      "grad_norm": 0.10997014492750168,
      "learning_rate": 0.0001909134339217186,
      "loss": 0.5097,
      "step": 924
    },
    {
      "epoch": 0.13818344786375858,
      "grad_norm": 0.25661003589630127,
      "learning_rate": 0.00019089384753710772,
      "loss": 1.1546,
      "step": 925
    },
    {
      "epoch": 0.13833283537496266,
      "grad_norm": 0.14332082867622375,
      "learning_rate": 0.00019087424107260627,
      "loss": 0.6839,
      "step": 926
    },
    {
      "epoch": 0.1384822228861667,
      "grad_norm": 0.29279524087905884,
      "learning_rate": 0.0001908546145325456,
      "loss": 0.8088,
      "step": 927
    },
    {
      "epoch": 0.1386316103973708,
      "grad_norm": 0.2230231910943985,
      "learning_rate": 0.00019083496792126153,
      "loss": 0.7403,
      "step": 928
    },
    {
      "epoch": 0.13878099790857484,
      "grad_norm": 0.20737332105636597,
      "learning_rate": 0.00019081530124309427,
      "loss": 1.1846,
      "step": 929
    },
    {
      "epoch": 0.1389303854197789,
      "grad_norm": 0.2875371277332306,
      "learning_rate": 0.00019079561450238854,
      "loss": 0.9049,
      "step": 930
    },
    {
      "epoch": 0.13907977293098298,
      "grad_norm": 0.12367036938667297,
      "learning_rate": 0.00019077590770349344,
      "loss": 0.8605,
      "step": 931
    },
    {
      "epoch": 0.13922916044218703,
      "grad_norm": 0.34429073333740234,
      "learning_rate": 0.00019075618085076247,
      "loss": 1.0233,
      "step": 932
    },
    {
      "epoch": 0.1393785479533911,
      "grad_norm": 0.15258093178272247,
      "learning_rate": 0.00019073643394855368,
      "loss": 0.8381,
      "step": 933
    },
    {
      "epoch": 0.13952793546459516,
      "grad_norm": 0.15117698907852173,
      "learning_rate": 0.00019071666700122946,
      "loss": 0.6532,
      "step": 934
    },
    {
      "epoch": 0.1396773229757992,
      "grad_norm": 0.11969941109418869,
      "learning_rate": 0.0001906968800131566,
      "loss": 0.7025,
      "step": 935
    },
    {
      "epoch": 0.1398267104870033,
      "grad_norm": 0.12973757088184357,
      "learning_rate": 0.00019067707298870638,
      "loss": 0.6276,
      "step": 936
    },
    {
      "epoch": 0.13997609799820734,
      "grad_norm": 0.24265740811824799,
      "learning_rate": 0.0001906572459322545,
      "loss": 0.8548,
      "step": 937
    },
    {
      "epoch": 0.14012548550941142,
      "grad_norm": 0.18727362155914307,
      "learning_rate": 0.00019063739884818103,
      "loss": 0.5687,
      "step": 938
    },
    {
      "epoch": 0.14027487302061548,
      "grad_norm": 0.13480857014656067,
      "learning_rate": 0.00019061753174087054,
      "loss": 0.9462,
      "step": 939
    },
    {
      "epoch": 0.14042426053181953,
      "grad_norm": 0.2221670150756836,
      "learning_rate": 0.000190597644614712,
      "loss": 0.6429,
      "step": 940
    },
    {
      "epoch": 0.1405736480430236,
      "grad_norm": 0.13278596103191376,
      "learning_rate": 0.00019057773747409879,
      "loss": 0.839,
      "step": 941
    },
    {
      "epoch": 0.14072303555422766,
      "grad_norm": 0.12343950569629669,
      "learning_rate": 0.00019055781032342864,
      "loss": 0.8807,
      "step": 942
    },
    {
      "epoch": 0.14087242306543174,
      "grad_norm": 0.1782051920890808,
      "learning_rate": 0.00019053786316710386,
      "loss": 1.0982,
      "step": 943
    },
    {
      "epoch": 0.1410218105766358,
      "grad_norm": 0.1475050151348114,
      "learning_rate": 0.00019051789600953102,
      "loss": 1.0571,
      "step": 944
    },
    {
      "epoch": 0.14117119808783984,
      "grad_norm": 0.12406841665506363,
      "learning_rate": 0.00019049790885512126,
      "loss": 0.641,
      "step": 945
    },
    {
      "epoch": 0.14132058559904392,
      "grad_norm": 0.13010482490062714,
      "learning_rate": 0.00019047790170829003,
      "loss": 0.7616,
      "step": 946
    },
    {
      "epoch": 0.14146997311024798,
      "grad_norm": 0.14790335297584534,
      "learning_rate": 0.00019045787457345722,
      "loss": 0.6587,
      "step": 947
    },
    {
      "epoch": 0.14161936062145206,
      "grad_norm": 0.11543125659227371,
      "learning_rate": 0.00019043782745504711,
      "loss": 0.7014,
      "step": 948
    },
    {
      "epoch": 0.1417687481326561,
      "grad_norm": 0.12451034784317017,
      "learning_rate": 0.00019041776035748847,
      "loss": 0.5931,
      "step": 949
    },
    {
      "epoch": 0.14191813564386016,
      "grad_norm": 0.1254526674747467,
      "learning_rate": 0.00019039767328521442,
      "loss": 0.7772,
      "step": 950
    },
    {
      "epoch": 0.14206752315506424,
      "grad_norm": 0.21236388385295868,
      "learning_rate": 0.00019037756624266252,
      "loss": 0.6836,
      "step": 951
    },
    {
      "epoch": 0.1422169106662683,
      "grad_norm": 0.1704844981431961,
      "learning_rate": 0.0001903574392342747,
      "loss": 0.5561,
      "step": 952
    },
    {
      "epoch": 0.14236629817747237,
      "grad_norm": 0.1470273733139038,
      "learning_rate": 0.0001903372922644974,
      "loss": 0.7572,
      "step": 953
    },
    {
      "epoch": 0.14251568568867642,
      "grad_norm": 0.1643514782190323,
      "learning_rate": 0.00019031712533778137,
      "loss": 0.8037,
      "step": 954
    },
    {
      "epoch": 0.14266507319988048,
      "grad_norm": 0.1775134652853012,
      "learning_rate": 0.0001902969384585818,
      "loss": 0.6325,
      "step": 955
    },
    {
      "epoch": 0.14281446071108456,
      "grad_norm": 0.16977538168430328,
      "learning_rate": 0.00019027673163135827,
      "loss": 0.4522,
      "step": 956
    },
    {
      "epoch": 0.1429638482222886,
      "grad_norm": 0.2153913825750351,
      "learning_rate": 0.00019025650486057484,
      "loss": 0.6189,
      "step": 957
    },
    {
      "epoch": 0.1431132357334927,
      "grad_norm": 0.12824203073978424,
      "learning_rate": 0.00019023625815069989,
      "loss": 0.5429,
      "step": 958
    },
    {
      "epoch": 0.14326262324469674,
      "grad_norm": 0.13086850941181183,
      "learning_rate": 0.00019021599150620625,
      "loss": 0.854,
      "step": 959
    },
    {
      "epoch": 0.1434120107559008,
      "grad_norm": 0.13822171092033386,
      "learning_rate": 0.00019019570493157114,
      "loss": 0.7447,
      "step": 960
    },
    {
      "epoch": 0.14356139826710487,
      "grad_norm": 0.13974325358867645,
      "learning_rate": 0.00019017539843127617,
      "loss": 0.9451,
      "step": 961
    },
    {
      "epoch": 0.14371078577830892,
      "grad_norm": 0.17371952533721924,
      "learning_rate": 0.0001901550720098074,
      "loss": 0.9114,
      "step": 962
    },
    {
      "epoch": 0.143860173289513,
      "grad_norm": 0.2538747191429138,
      "learning_rate": 0.00019013472567165523,
      "loss": 1.1961,
      "step": 963
    },
    {
      "epoch": 0.14400956080071706,
      "grad_norm": 0.110201895236969,
      "learning_rate": 0.00019011435942131448,
      "loss": 0.4677,
      "step": 964
    },
    {
      "epoch": 0.14415894831192114,
      "grad_norm": 0.11606165021657944,
      "learning_rate": 0.00019009397326328443,
      "loss": 0.7407,
      "step": 965
    },
    {
      "epoch": 0.1443083358231252,
      "grad_norm": 0.1762661188840866,
      "learning_rate": 0.00019007356720206865,
      "loss": 0.7177,
      "step": 966
    },
    {
      "epoch": 0.14445772333432924,
      "grad_norm": 0.16712108254432678,
      "learning_rate": 0.0001900531412421752,
      "loss": 0.858,
      "step": 967
    },
    {
      "epoch": 0.14460711084553332,
      "grad_norm": 0.17348961532115936,
      "learning_rate": 0.00019003269538811647,
      "loss": 1.1409,
      "step": 968
    },
    {
      "epoch": 0.14475649835673737,
      "grad_norm": 0.13761204481124878,
      "learning_rate": 0.0001900122296444093,
      "loss": 0.6666,
      "step": 969
    },
    {
      "epoch": 0.14490588586794145,
      "grad_norm": 0.20189805328845978,
      "learning_rate": 0.00018999174401557488,
      "loss": 0.736,
      "step": 970
    },
    {
      "epoch": 0.1450552733791455,
      "grad_norm": 0.15712390840053558,
      "learning_rate": 0.0001899712385061388,
      "loss": 0.7429,
      "step": 971
    },
    {
      "epoch": 0.14520466089034956,
      "grad_norm": 0.12312103062868118,
      "learning_rate": 0.00018995071312063105,
      "loss": 0.8887,
      "step": 972
    },
    {
      "epoch": 0.14535404840155364,
      "grad_norm": 0.16373762488365173,
      "learning_rate": 0.00018993016786358603,
      "loss": 1.0923,
      "step": 973
    },
    {
      "epoch": 0.1455034359127577,
      "grad_norm": 0.1286284178495407,
      "learning_rate": 0.00018990960273954254,
      "loss": 0.5527,
      "step": 974
    },
    {
      "epoch": 0.14565282342396177,
      "grad_norm": 0.1538635641336441,
      "learning_rate": 0.0001898890177530437,
      "loss": 0.8092,
      "step": 975
    },
    {
      "epoch": 0.14580221093516582,
      "grad_norm": 0.14671294391155243,
      "learning_rate": 0.00018986841290863704,
      "loss": 0.7165,
      "step": 976
    },
    {
      "epoch": 0.14595159844636987,
      "grad_norm": 0.1533249318599701,
      "learning_rate": 0.00018984778821087454,
      "loss": 0.6822,
      "step": 977
    },
    {
      "epoch": 0.14610098595757395,
      "grad_norm": 0.1579664945602417,
      "learning_rate": 0.0001898271436643125,
      "loss": 0.8128,
      "step": 978
    },
    {
      "epoch": 0.146250373468778,
      "grad_norm": 0.1557191014289856,
      "learning_rate": 0.00018980647927351166,
      "loss": 0.8699,
      "step": 979
    },
    {
      "epoch": 0.14639976097998209,
      "grad_norm": 0.1736837476491928,
      "learning_rate": 0.00018978579504303706,
      "loss": 1.2032,
      "step": 980
    },
    {
      "epoch": 0.14654914849118614,
      "grad_norm": 0.17181052267551422,
      "learning_rate": 0.00018976509097745826,
      "loss": 0.7804,
      "step": 981
    },
    {
      "epoch": 0.1466985360023902,
      "grad_norm": 0.20371553301811218,
      "learning_rate": 0.000189744367081349,
      "loss": 0.9971,
      "step": 982
    },
    {
      "epoch": 0.14684792351359427,
      "grad_norm": 0.11608695983886719,
      "learning_rate": 0.00018972362335928757,
      "loss": 0.7474,
      "step": 983
    },
    {
      "epoch": 0.14699731102479832,
      "grad_norm": 0.23304304480552673,
      "learning_rate": 0.00018970285981585662,
      "loss": 1.0192,
      "step": 984
    },
    {
      "epoch": 0.1471466985360024,
      "grad_norm": 0.18036411702632904,
      "learning_rate": 0.0001896820764556431,
      "loss": 1.1699,
      "step": 985
    },
    {
      "epoch": 0.14729608604720645,
      "grad_norm": 0.45688414573669434,
      "learning_rate": 0.00018966127328323842,
      "loss": 1.5615,
      "step": 986
    },
    {
      "epoch": 0.1474454735584105,
      "grad_norm": 0.11910564452409744,
      "learning_rate": 0.00018964045030323828,
      "loss": 0.5669,
      "step": 987
    },
    {
      "epoch": 0.14759486106961459,
      "grad_norm": 0.15722838044166565,
      "learning_rate": 0.00018961960752024288,
      "loss": 0.7162,
      "step": 988
    },
    {
      "epoch": 0.14774424858081864,
      "grad_norm": 0.13349586725234985,
      "learning_rate": 0.00018959874493885666,
      "loss": 0.8232,
      "step": 989
    },
    {
      "epoch": 0.14789363609202272,
      "grad_norm": 0.2732388973236084,
      "learning_rate": 0.0001895778625636885,
      "loss": 0.9254,
      "step": 990
    },
    {
      "epoch": 0.14804302360322677,
      "grad_norm": 0.12720026075839996,
      "learning_rate": 0.00018955696039935167,
      "loss": 0.6276,
      "step": 991
    },
    {
      "epoch": 0.14819241111443082,
      "grad_norm": 0.12968410551548004,
      "learning_rate": 0.00018953603845046378,
      "loss": 0.7688,
      "step": 992
    },
    {
      "epoch": 0.1483417986256349,
      "grad_norm": 0.2596571445465088,
      "learning_rate": 0.0001895150967216468,
      "loss": 1.0049,
      "step": 993
    },
    {
      "epoch": 0.14849118613683895,
      "grad_norm": 0.12322080880403519,
      "learning_rate": 0.00018949413521752713,
      "loss": 0.7845,
      "step": 994
    },
    {
      "epoch": 0.14864057364804303,
      "grad_norm": 0.12585557997226715,
      "learning_rate": 0.00018947315394273546,
      "loss": 0.5276,
      "step": 995
    },
    {
      "epoch": 0.14878996115924709,
      "grad_norm": 0.16490721702575684,
      "learning_rate": 0.00018945215290190693,
      "loss": 0.5467,
      "step": 996
    },
    {
      "epoch": 0.14893934867045114,
      "grad_norm": 0.20895147323608398,
      "learning_rate": 0.00018943113209968094,
      "loss": 0.6746,
      "step": 997
    },
    {
      "epoch": 0.14908873618165522,
      "grad_norm": 0.15641289949417114,
      "learning_rate": 0.00018941009154070136,
      "loss": 0.5641,
      "step": 998
    },
    {
      "epoch": 0.14923812369285927,
      "grad_norm": 0.12262630462646484,
      "learning_rate": 0.0001893890312296164,
      "loss": 0.5929,
      "step": 999
    },
    {
      "epoch": 0.14938751120406335,
      "grad_norm": 0.15614943206310272,
      "learning_rate": 0.00018936795117107855,
      "loss": 0.7035,
      "step": 1000
    },
    {
      "epoch": 0.1495368987152674,
      "grad_norm": 0.13168151676654816,
      "learning_rate": 0.00018934685136974482,
      "loss": 0.7546,
      "step": 1001
    },
    {
      "epoch": 0.14968628622647145,
      "grad_norm": 0.12817221879959106,
      "learning_rate": 0.0001893257318302764,
      "loss": 0.6542,
      "step": 1002
    },
    {
      "epoch": 0.14983567373767553,
      "grad_norm": 0.22590519487857819,
      "learning_rate": 0.000189304592557339,
      "loss": 0.9049,
      "step": 1003
    },
    {
      "epoch": 0.1499850612488796,
      "grad_norm": 0.1316809505224228,
      "learning_rate": 0.00018928343355560258,
      "loss": 0.8921,
      "step": 1004
    },
    {
      "epoch": 0.15013444876008367,
      "grad_norm": 0.16760273277759552,
      "learning_rate": 0.00018926225482974153,
      "loss": 0.5918,
      "step": 1005
    },
    {
      "epoch": 0.15028383627128772,
      "grad_norm": 0.1197691559791565,
      "learning_rate": 0.00018924105638443452,
      "loss": 0.6532,
      "step": 1006
    },
    {
      "epoch": 0.15043322378249177,
      "grad_norm": 0.1392420530319214,
      "learning_rate": 0.00018921983822436467,
      "loss": 0.9638,
      "step": 1007
    },
    {
      "epoch": 0.15058261129369585,
      "grad_norm": 0.17294703423976898,
      "learning_rate": 0.0001891986003542194,
      "loss": 0.7818,
      "step": 1008
    },
    {
      "epoch": 0.1507319988048999,
      "grad_norm": 0.2133634239435196,
      "learning_rate": 0.00018917734277869042,
      "loss": 0.8933,
      "step": 1009
    },
    {
      "epoch": 0.15088138631610398,
      "grad_norm": 0.13073426485061646,
      "learning_rate": 0.00018915606550247397,
      "loss": 0.7142,
      "step": 1010
    },
    {
      "epoch": 0.15103077382730803,
      "grad_norm": 0.2918628454208374,
      "learning_rate": 0.0001891347685302705,
      "loss": 0.9469,
      "step": 1011
    },
    {
      "epoch": 0.1511801613385121,
      "grad_norm": 0.12934882938861847,
      "learning_rate": 0.0001891134518667848,
      "loss": 0.8446,
      "step": 1012
    },
    {
      "epoch": 0.15132954884971617,
      "grad_norm": 0.12790237367153168,
      "learning_rate": 0.00018909211551672615,
      "loss": 0.5309,
      "step": 1013
    },
    {
      "epoch": 0.15147893636092022,
      "grad_norm": 0.2520146071910858,
      "learning_rate": 0.000189070759484808,
      "loss": 0.8729,
      "step": 1014
    },
    {
      "epoch": 0.1516283238721243,
      "grad_norm": 0.1819266974925995,
      "learning_rate": 0.00018904938377574827,
      "loss": 0.8364,
      "step": 1015
    },
    {
      "epoch": 0.15177771138332835,
      "grad_norm": 0.13670526444911957,
      "learning_rate": 0.0001890279883942692,
      "loss": 1.178,
      "step": 1016
    },
    {
      "epoch": 0.15192709889453243,
      "grad_norm": 0.15499313175678253,
      "learning_rate": 0.00018900657334509736,
      "loss": 0.6144,
      "step": 1017
    },
    {
      "epoch": 0.15207648640573648,
      "grad_norm": 0.13412916660308838,
      "learning_rate": 0.0001889851386329637,
      "loss": 0.7415,
      "step": 1018
    },
    {
      "epoch": 0.15222587391694054,
      "grad_norm": 0.17868481576442719,
      "learning_rate": 0.00018896368426260345,
      "loss": 1.0732,
      "step": 1019
    },
    {
      "epoch": 0.15237526142814461,
      "grad_norm": 0.197604700922966,
      "learning_rate": 0.00018894221023875622,
      "loss": 0.8012,
      "step": 1020
    },
    {
      "epoch": 0.15252464893934867,
      "grad_norm": 0.13587865233421326,
      "learning_rate": 0.000188920716566166,
      "loss": 0.5255,
      "step": 1021
    },
    {
      "epoch": 0.15267403645055275,
      "grad_norm": 0.13020993769168854,
      "learning_rate": 0.00018889920324958106,
      "loss": 0.5617,
      "step": 1022
    },
    {
      "epoch": 0.1528234239617568,
      "grad_norm": 0.19373324513435364,
      "learning_rate": 0.00018887767029375401,
      "loss": 1.085,
      "step": 1023
    },
    {
      "epoch": 0.15297281147296085,
      "grad_norm": 0.11815980821847916,
      "learning_rate": 0.00018885611770344185,
      "loss": 0.7142,
      "step": 1024
    },
    {
      "epoch": 0.15312219898416493,
      "grad_norm": 0.21390031278133392,
      "learning_rate": 0.00018883454548340587,
      "loss": 1.0385,
      "step": 1025
    },
    {
      "epoch": 0.15327158649536898,
      "grad_norm": 0.19770096242427826,
      "learning_rate": 0.00018881295363841174,
      "loss": 0.7931,
      "step": 1026
    },
    {
      "epoch": 0.15342097400657306,
      "grad_norm": 0.1958816796541214,
      "learning_rate": 0.00018879134217322942,
      "loss": 0.8291,
      "step": 1027
    },
    {
      "epoch": 0.15357036151777712,
      "grad_norm": 0.11775219440460205,
      "learning_rate": 0.00018876971109263324,
      "loss": 0.6206,
      "step": 1028
    },
    {
      "epoch": 0.15371974902898117,
      "grad_norm": 0.11675494909286499,
      "learning_rate": 0.00018874806040140185,
      "loss": 0.5483,
      "step": 1029
    },
    {
      "epoch": 0.15386913654018525,
      "grad_norm": 0.33525991439819336,
      "learning_rate": 0.00018872639010431822,
      "loss": 1.6754,
      "step": 1030
    },
    {
      "epoch": 0.1540185240513893,
      "grad_norm": 0.15912093222141266,
      "learning_rate": 0.00018870470020616967,
      "loss": 0.7152,
      "step": 1031
    },
    {
      "epoch": 0.15416791156259338,
      "grad_norm": 0.1797519475221634,
      "learning_rate": 0.0001886829907117478,
      "loss": 0.5595,
      "step": 1032
    },
    {
      "epoch": 0.15431729907379743,
      "grad_norm": 0.163869708776474,
      "learning_rate": 0.0001886612616258486,
      "loss": 0.6443,
      "step": 1033
    },
    {
      "epoch": 0.15446668658500148,
      "grad_norm": 0.21159547567367554,
      "learning_rate": 0.00018863951295327244,
      "loss": 0.8727,
      "step": 1034
    },
    {
      "epoch": 0.15461607409620556,
      "grad_norm": 0.17695502936840057,
      "learning_rate": 0.00018861774469882385,
      "loss": 1.269,
      "step": 1035
    },
    {
      "epoch": 0.15476546160740962,
      "grad_norm": 0.23800405859947205,
      "learning_rate": 0.00018859595686731187,
      "loss": 1.2047,
      "step": 1036
    },
    {
      "epoch": 0.1549148491186137,
      "grad_norm": 0.11121930181980133,
      "learning_rate": 0.0001885741494635497,
      "loss": 0.6537,
      "step": 1037
    },
    {
      "epoch": 0.15506423662981775,
      "grad_norm": 0.10217446833848953,
      "learning_rate": 0.00018855232249235498,
      "loss": 0.4512,
      "step": 1038
    },
    {
      "epoch": 0.1552136241410218,
      "grad_norm": 0.2477329820394516,
      "learning_rate": 0.0001885304759585496,
      "loss": 0.8929,
      "step": 1039
    },
    {
      "epoch": 0.15536301165222588,
      "grad_norm": 0.13516084849834442,
      "learning_rate": 0.00018850860986695985,
      "loss": 0.6883,
      "step": 1040
    },
    {
      "epoch": 0.15551239916342993,
      "grad_norm": 0.14974398910999298,
      "learning_rate": 0.0001884867242224163,
      "loss": 0.6658,
      "step": 1041
    },
    {
      "epoch": 0.155661786674634,
      "grad_norm": 0.12428376823663712,
      "learning_rate": 0.00018846481902975377,
      "loss": 0.8312,
      "step": 1042
    },
    {
      "epoch": 0.15581117418583806,
      "grad_norm": 0.10979831218719482,
      "learning_rate": 0.0001884428942938115,
      "loss": 0.6737,
      "step": 1043
    },
    {
      "epoch": 0.15596056169704212,
      "grad_norm": 0.18722845613956451,
      "learning_rate": 0.00018842095001943306,
      "loss": 1.0435,
      "step": 1044
    },
    {
      "epoch": 0.1561099492082462,
      "grad_norm": 0.1588689237833023,
      "learning_rate": 0.00018839898621146625,
      "loss": 0.792,
      "step": 1045
    },
    {
      "epoch": 0.15625933671945025,
      "grad_norm": 0.10412801802158356,
      "learning_rate": 0.00018837700287476316,
      "loss": 0.4995,
      "step": 1046
    },
    {
      "epoch": 0.15640872423065433,
      "grad_norm": 0.1294378787279129,
      "learning_rate": 0.00018835500001418036,
      "loss": 0.8798,
      "step": 1047
    },
    {
      "epoch": 0.15655811174185838,
      "grad_norm": 0.16280725598335266,
      "learning_rate": 0.00018833297763457858,
      "loss": 0.6571,
      "step": 1048
    },
    {
      "epoch": 0.15670749925306243,
      "grad_norm": 0.12636205554008484,
      "learning_rate": 0.0001883109357408229,
      "loss": 0.5629,
      "step": 1049
    },
    {
      "epoch": 0.1568568867642665,
      "grad_norm": 0.22952638566493988,
      "learning_rate": 0.00018828887433778278,
      "loss": 0.9414,
      "step": 1050
    },
    {
      "epoch": 0.15700627427547056,
      "grad_norm": 0.18174390494823456,
      "learning_rate": 0.00018826679343033186,
      "loss": 0.6405,
      "step": 1051
    },
    {
      "epoch": 0.15715566178667464,
      "grad_norm": 0.4626804292201996,
      "learning_rate": 0.00018824469302334822,
      "loss": 1.1709,
      "step": 1052
    },
    {
      "epoch": 0.1573050492978787,
      "grad_norm": 0.17580588161945343,
      "learning_rate": 0.00018822257312171416,
      "loss": 0.7962,
      "step": 1053
    },
    {
      "epoch": 0.15745443680908275,
      "grad_norm": 0.1333022564649582,
      "learning_rate": 0.0001882004337303163,
      "loss": 0.7548,
      "step": 1054
    },
    {
      "epoch": 0.15760382432028683,
      "grad_norm": 1.2796801328659058,
      "learning_rate": 0.00018817827485404564,
      "loss": 1.9406,
      "step": 1055
    },
    {
      "epoch": 0.15775321183149088,
      "grad_norm": 0.17332683503627777,
      "learning_rate": 0.0001881560964977974,
      "loss": 0.8372,
      "step": 1056
    },
    {
      "epoch": 0.15790259934269496,
      "grad_norm": 0.15442058444023132,
      "learning_rate": 0.0001881338986664711,
      "loss": 0.8606,
      "step": 1057
    },
    {
      "epoch": 0.158051986853899,
      "grad_norm": 0.14709503948688507,
      "learning_rate": 0.0001881116813649706,
      "loss": 0.8004,
      "step": 1058
    },
    {
      "epoch": 0.15820137436510306,
      "grad_norm": 0.18778932094573975,
      "learning_rate": 0.00018808944459820407,
      "loss": 0.9717,
      "step": 1059
    },
    {
      "epoch": 0.15835076187630714,
      "grad_norm": 0.1773632913827896,
      "learning_rate": 0.00018806718837108402,
      "loss": 0.981,
      "step": 1060
    },
    {
      "epoch": 0.1585001493875112,
      "grad_norm": 0.13269007205963135,
      "learning_rate": 0.00018804491268852708,
      "loss": 0.8844,
      "step": 1061
    },
    {
      "epoch": 0.15864953689871528,
      "grad_norm": 0.17859001457691193,
      "learning_rate": 0.00018802261755545443,
      "loss": 0.5105,
      "step": 1062
    },
    {
      "epoch": 0.15879892440991933,
      "grad_norm": 0.1438104659318924,
      "learning_rate": 0.0001880003029767913,
      "loss": 0.7643,
      "step": 1063
    },
    {
      "epoch": 0.15894831192112338,
      "grad_norm": 0.13282090425491333,
      "learning_rate": 0.0001879779689574674,
      "loss": 0.821,
      "step": 1064
    },
    {
      "epoch": 0.15909769943232746,
      "grad_norm": 0.24478429555892944,
      "learning_rate": 0.0001879556155024167,
      "loss": 0.6762,
      "step": 1065
    },
    {
      "epoch": 0.1592470869435315,
      "grad_norm": 0.11228256672620773,
      "learning_rate": 0.00018793324261657737,
      "loss": 0.8163,
      "step": 1066
    },
    {
      "epoch": 0.1593964744547356,
      "grad_norm": 0.10839138180017471,
      "learning_rate": 0.00018791085030489194,
      "loss": 0.6536,
      "step": 1067
    },
    {
      "epoch": 0.15954586196593964,
      "grad_norm": 0.14781688153743744,
      "learning_rate": 0.00018788843857230726,
      "loss": 0.7572,
      "step": 1068
    },
    {
      "epoch": 0.1596952494771437,
      "grad_norm": 0.18171058595180511,
      "learning_rate": 0.00018786600742377437,
      "loss": 0.9202,
      "step": 1069
    },
    {
      "epoch": 0.15984463698834778,
      "grad_norm": 0.11682315170764923,
      "learning_rate": 0.00018784355686424876,
      "loss": 0.7,
      "step": 1070
    },
    {
      "epoch": 0.15999402449955183,
      "grad_norm": 0.09597928076982498,
      "learning_rate": 0.00018782108689869006,
      "loss": 0.621,
      "step": 1071
    },
    {
      "epoch": 0.1601434120107559,
      "grad_norm": 0.15037846565246582,
      "learning_rate": 0.00018779859753206225,
      "loss": 0.7951,
      "step": 1072
    },
    {
      "epoch": 0.16029279952195996,
      "grad_norm": 0.11888540536165237,
      "learning_rate": 0.00018777608876933358,
      "loss": 0.7986,
      "step": 1073
    },
    {
      "epoch": 0.16044218703316404,
      "grad_norm": 0.4484766721725464,
      "learning_rate": 0.00018775356061547662,
      "loss": 1.231,
      "step": 1074
    },
    {
      "epoch": 0.1605915745443681,
      "grad_norm": 0.13222385942935944,
      "learning_rate": 0.00018773101307546815,
      "loss": 0.7957,
      "step": 1075
    },
    {
      "epoch": 0.16074096205557215,
      "grad_norm": 0.18500378727912903,
      "learning_rate": 0.00018770844615428932,
      "loss": 0.695,
      "step": 1076
    },
    {
      "epoch": 0.16089034956677623,
      "grad_norm": 0.10332745313644409,
      "learning_rate": 0.0001876858598569255,
      "loss": 0.556,
      "step": 1077
    },
    {
      "epoch": 0.16103973707798028,
      "grad_norm": 0.0973210334777832,
      "learning_rate": 0.00018766325418836637,
      "loss": 0.5246,
      "step": 1078
    },
    {
      "epoch": 0.16118912458918436,
      "grad_norm": 0.15679813921451569,
      "learning_rate": 0.00018764062915360588,
      "loss": 0.7182,
      "step": 1079
    },
    {
      "epoch": 0.1613385121003884,
      "grad_norm": 0.12232697755098343,
      "learning_rate": 0.00018761798475764224,
      "loss": 0.7503,
      "step": 1080
    },
    {
      "epoch": 0.16148789961159246,
      "grad_norm": 0.13506333529949188,
      "learning_rate": 0.00018759532100547799,
      "loss": 0.6708,
      "step": 1081
    },
    {
      "epoch": 0.16163728712279654,
      "grad_norm": 0.1789216846227646,
      "learning_rate": 0.00018757263790211988,
      "loss": 1.0662,
      "step": 1082
    },
    {
      "epoch": 0.1617866746340006,
      "grad_norm": 0.1154765784740448,
      "learning_rate": 0.00018754993545257894,
      "loss": 0.6931,
      "step": 1083
    },
    {
      "epoch": 0.16193606214520467,
      "grad_norm": 0.11015768349170685,
      "learning_rate": 0.0001875272136618706,
      "loss": 0.8244,
      "step": 1084
    },
    {
      "epoch": 0.16208544965640873,
      "grad_norm": 0.14176665246486664,
      "learning_rate": 0.00018750447253501436,
      "loss": 0.8434,
      "step": 1085
    },
    {
      "epoch": 0.16223483716761278,
      "grad_norm": 0.13398586213588715,
      "learning_rate": 0.00018748171207703417,
      "loss": 0.8307,
      "step": 1086
    },
    {
      "epoch": 0.16238422467881686,
      "grad_norm": 0.18871843814849854,
      "learning_rate": 0.00018745893229295813,
      "loss": 0.7193,
      "step": 1087
    },
    {
      "epoch": 0.1625336121900209,
      "grad_norm": 0.1976425051689148,
      "learning_rate": 0.00018743613318781868,
      "loss": 0.9341,
      "step": 1088
    },
    {
      "epoch": 0.162682999701225,
      "grad_norm": 0.12541085481643677,
      "learning_rate": 0.00018741331476665246,
      "loss": 0.7364,
      "step": 1089
    },
    {
      "epoch": 0.16283238721242904,
      "grad_norm": 0.3148552179336548,
      "learning_rate": 0.00018739047703450048,
      "loss": 1.0103,
      "step": 1090
    },
    {
      "epoch": 0.1629817747236331,
      "grad_norm": 0.24794772267341614,
      "learning_rate": 0.00018736761999640792,
      "loss": 0.7772,
      "step": 1091
    },
    {
      "epoch": 0.16313116223483717,
      "grad_norm": 0.12505854666233063,
      "learning_rate": 0.00018734474365742428,
      "loss": 0.7458,
      "step": 1092
    },
    {
      "epoch": 0.16328054974604123,
      "grad_norm": 0.13425281643867493,
      "learning_rate": 0.0001873218480226033,
      "loss": 0.6537,
      "step": 1093
    },
    {
      "epoch": 0.1634299372572453,
      "grad_norm": 0.29929280281066895,
      "learning_rate": 0.00018729893309700295,
      "loss": 1.2375,
      "step": 1094
    },
    {
      "epoch": 0.16357932476844936,
      "grad_norm": 0.13596223294734955,
      "learning_rate": 0.00018727599888568555,
      "loss": 0.956,
      "step": 1095
    },
    {
      "epoch": 0.1637287122796534,
      "grad_norm": 0.13480113446712494,
      "learning_rate": 0.0001872530453937176,
      "loss": 0.8181,
      "step": 1096
    },
    {
      "epoch": 0.1638780997908575,
      "grad_norm": 0.14178095757961273,
      "learning_rate": 0.00018723007262616998,
      "loss": 0.7811,
      "step": 1097
    },
    {
      "epoch": 0.16402748730206154,
      "grad_norm": 0.11537758260965347,
      "learning_rate": 0.0001872070805881176,
      "loss": 0.7824,
      "step": 1098
    },
    {
      "epoch": 0.16417687481326562,
      "grad_norm": 0.15772311389446259,
      "learning_rate": 0.00018718406928463986,
      "loss": 0.679,
      "step": 1099
    },
    {
      "epoch": 0.16432626232446967,
      "grad_norm": 0.22873377799987793,
      "learning_rate": 0.00018716103872082026,
      "loss": 0.4674,
      "step": 1100
    },
    {
      "epoch": 0.16447564983567373,
      "grad_norm": 0.14255951344966888,
      "learning_rate": 0.00018713798890174666,
      "loss": 0.7014,
      "step": 1101
    },
    {
      "epoch": 0.1646250373468778,
      "grad_norm": 0.15469695627689362,
      "learning_rate": 0.00018711491983251113,
      "loss": 0.8959,
      "step": 1102
    },
    {
      "epoch": 0.16477442485808186,
      "grad_norm": 0.10443451255559921,
      "learning_rate": 0.00018709183151820996,
      "loss": 0.548,
      "step": 1103
    },
    {
      "epoch": 0.16492381236928594,
      "grad_norm": 0.10996398329734802,
      "learning_rate": 0.00018706872396394376,
      "loss": 0.6857,
      "step": 1104
    },
    {
      "epoch": 0.16507319988049,
      "grad_norm": 0.15236075222492218,
      "learning_rate": 0.00018704559717481732,
      "loss": 0.796,
      "step": 1105
    },
    {
      "epoch": 0.16522258739169404,
      "grad_norm": 0.19824904203414917,
      "learning_rate": 0.00018702245115593974,
      "loss": 0.6438,
      "step": 1106
    },
    {
      "epoch": 0.16537197490289812,
      "grad_norm": 0.17726042866706848,
      "learning_rate": 0.00018699928591242438,
      "loss": 0.9564,
      "step": 1107
    },
    {
      "epoch": 0.16552136241410217,
      "grad_norm": 0.12704910337924957,
      "learning_rate": 0.0001869761014493887,
      "loss": 0.5796,
      "step": 1108
    },
    {
      "epoch": 0.16567074992530625,
      "grad_norm": 0.22029517590999603,
      "learning_rate": 0.00018695289777195462,
      "loss": 0.7932,
      "step": 1109
    },
    {
      "epoch": 0.1658201374365103,
      "grad_norm": 0.1251213401556015,
      "learning_rate": 0.00018692967488524812,
      "loss": 0.8287,
      "step": 1110
    },
    {
      "epoch": 0.16596952494771436,
      "grad_norm": 0.16409267485141754,
      "learning_rate": 0.00018690643279439958,
      "loss": 0.6792,
      "step": 1111
    },
    {
      "epoch": 0.16611891245891844,
      "grad_norm": 0.11526962369680405,
      "learning_rate": 0.0001868831715045435,
      "loss": 0.5228,
      "step": 1112
    },
    {
      "epoch": 0.1662682999701225,
      "grad_norm": 0.14684948325157166,
      "learning_rate": 0.00018685989102081867,
      "loss": 0.6461,
      "step": 1113
    },
    {
      "epoch": 0.16641768748132657,
      "grad_norm": 0.18166372179985046,
      "learning_rate": 0.00018683659134836813,
      "loss": 0.8579,
      "step": 1114
    },
    {
      "epoch": 0.16656707499253062,
      "grad_norm": 0.12297386676073074,
      "learning_rate": 0.00018681327249233913,
      "loss": 0.8628,
      "step": 1115
    },
    {
      "epoch": 0.16671646250373467,
      "grad_norm": 0.1687268614768982,
      "learning_rate": 0.00018678993445788323,
      "loss": 0.9859,
      "step": 1116
    },
    {
      "epoch": 0.16686585001493875,
      "grad_norm": 0.10087472945451736,
      "learning_rate": 0.00018676657725015606,
      "loss": 0.6073,
      "step": 1117
    },
    {
      "epoch": 0.1670152375261428,
      "grad_norm": 0.12146350741386414,
      "learning_rate": 0.00018674320087431768,
      "loss": 0.68,
      "step": 1118
    },
    {
      "epoch": 0.1671646250373469,
      "grad_norm": 0.13258694112300873,
      "learning_rate": 0.0001867198053355323,
      "loss": 0.8528,
      "step": 1119
    },
    {
      "epoch": 0.16731401254855094,
      "grad_norm": 0.15267039835453033,
      "learning_rate": 0.00018669639063896836,
      "loss": 0.7095,
      "step": 1120
    },
    {
      "epoch": 0.167463400059755,
      "grad_norm": 0.12942974269390106,
      "learning_rate": 0.00018667295678979852,
      "loss": 0.8282,
      "step": 1121
    },
    {
      "epoch": 0.16761278757095907,
      "grad_norm": 0.1521858274936676,
      "learning_rate": 0.0001866495037931997,
      "loss": 0.8557,
      "step": 1122
    },
    {
      "epoch": 0.16776217508216312,
      "grad_norm": 0.1155681312084198,
      "learning_rate": 0.000186626031654353,
      "loss": 0.6572,
      "step": 1123
    },
    {
      "epoch": 0.1679115625933672,
      "grad_norm": 0.13675054907798767,
      "learning_rate": 0.00018660254037844388,
      "loss": 0.7218,
      "step": 1124
    },
    {
      "epoch": 0.16806095010457126,
      "grad_norm": 0.13993573188781738,
      "learning_rate": 0.00018657902997066183,
      "loss": 0.6353,
      "step": 1125
    },
    {
      "epoch": 0.16821033761577533,
      "grad_norm": 0.16351041197776794,
      "learning_rate": 0.00018655550043620073,
      "loss": 1.027,
      "step": 1126
    },
    {
      "epoch": 0.1683597251269794,
      "grad_norm": 0.2799745500087738,
      "learning_rate": 0.00018653195178025864,
      "loss": 0.7067,
      "step": 1127
    },
    {
      "epoch": 0.16850911263818344,
      "grad_norm": 0.19105161726474762,
      "learning_rate": 0.0001865083840080378,
      "loss": 0.7683,
      "step": 1128
    },
    {
      "epoch": 0.16865850014938752,
      "grad_norm": 0.22695805132389069,
      "learning_rate": 0.00018648479712474468,
      "loss": 1.0468,
      "step": 1129
    },
    {
      "epoch": 0.16880788766059157,
      "grad_norm": 0.17338265478610992,
      "learning_rate": 0.00018646119113559006,
      "loss": 0.8898,
      "step": 1130
    },
    {
      "epoch": 0.16895727517179565,
      "grad_norm": 0.12882818281650543,
      "learning_rate": 0.00018643756604578885,
      "loss": 0.8652,
      "step": 1131
    },
    {
      "epoch": 0.1691066626829997,
      "grad_norm": 0.2222001552581787,
      "learning_rate": 0.00018641392186056016,
      "loss": 0.7363,
      "step": 1132
    },
    {
      "epoch": 0.16925605019420376,
      "grad_norm": 0.17244432866573334,
      "learning_rate": 0.00018639025858512741,
      "loss": 0.5881,
      "step": 1133
    },
    {
      "epoch": 0.16940543770540784,
      "grad_norm": 0.1207844614982605,
      "learning_rate": 0.0001863665762247182,
      "loss": 0.832,
      "step": 1134
    },
    {
      "epoch": 0.1695548252166119,
      "grad_norm": 0.20044654607772827,
      "learning_rate": 0.00018634287478456432,
      "loss": 0.9488,
      "step": 1135
    },
    {
      "epoch": 0.16970421272781597,
      "grad_norm": 0.13231255114078522,
      "learning_rate": 0.00018631915426990184,
      "loss": 0.5556,
      "step": 1136
    },
    {
      "epoch": 0.16985360023902002,
      "grad_norm": 0.444367378950119,
      "learning_rate": 0.00018629541468597092,
      "loss": 1.1058,
      "step": 1137
    },
    {
      "epoch": 0.17000298775022407,
      "grad_norm": 0.1951618492603302,
      "learning_rate": 0.00018627165603801605,
      "loss": 0.9483,
      "step": 1138
    },
    {
      "epoch": 0.17015237526142815,
      "grad_norm": 0.2736690044403076,
      "learning_rate": 0.00018624787833128588,
      "loss": 0.8482,
      "step": 1139
    },
    {
      "epoch": 0.1703017627726322,
      "grad_norm": 0.15237697958946228,
      "learning_rate": 0.0001862240815710333,
      "loss": 0.6764,
      "step": 1140
    },
    {
      "epoch": 0.17045115028383628,
      "grad_norm": 0.14255677163600922,
      "learning_rate": 0.0001862002657625154,
      "loss": 0.7176,
      "step": 1141
    },
    {
      "epoch": 0.17060053779504034,
      "grad_norm": 0.20565101504325867,
      "learning_rate": 0.0001861764309109934,
      "loss": 1.1075,
      "step": 1142
    },
    {
      "epoch": 0.1707499253062444,
      "grad_norm": 0.11338746547698975,
      "learning_rate": 0.0001861525770217329,
      "loss": 0.8231,
      "step": 1143
    },
    {
      "epoch": 0.17089931281744847,
      "grad_norm": 0.34042519330978394,
      "learning_rate": 0.00018612870410000354,
      "loss": 1.1565,
      "step": 1144
    },
    {
      "epoch": 0.17104870032865252,
      "grad_norm": 0.12565311789512634,
      "learning_rate": 0.00018610481215107925,
      "loss": 0.6115,
      "step": 1145
    },
    {
      "epoch": 0.1711980878398566,
      "grad_norm": 0.1262931525707245,
      "learning_rate": 0.00018608090118023808,
      "loss": 0.6307,
      "step": 1146
    },
    {
      "epoch": 0.17134747535106065,
      "grad_norm": 0.4232707917690277,
      "learning_rate": 0.00018605697119276242,
      "loss": 0.8655,
      "step": 1147
    },
    {
      "epoch": 0.1714968628622647,
      "grad_norm": 0.16684143245220184,
      "learning_rate": 0.00018603302219393874,
      "loss": 0.678,
      "step": 1148
    },
    {
      "epoch": 0.17164625037346878,
      "grad_norm": 0.17487527430057526,
      "learning_rate": 0.00018600905418905776,
      "loss": 0.4725,
      "step": 1149
    },
    {
      "epoch": 0.17179563788467284,
      "grad_norm": 0.1387338787317276,
      "learning_rate": 0.0001859850671834144,
      "loss": 0.8142,
      "step": 1150
    },
    {
      "epoch": 0.17194502539587692,
      "grad_norm": 0.15392735600471497,
      "learning_rate": 0.00018596106118230779,
      "loss": 0.7672,
      "step": 1151
    },
    {
      "epoch": 0.17209441290708097,
      "grad_norm": 0.251715749502182,
      "learning_rate": 0.0001859370361910412,
      "loss": 0.742,
      "step": 1152
    },
    {
      "epoch": 0.17224380041828502,
      "grad_norm": 0.18483877182006836,
      "learning_rate": 0.00018591299221492214,
      "loss": 0.8683,
      "step": 1153
    },
    {
      "epoch": 0.1723931879294891,
      "grad_norm": 0.14035139977931976,
      "learning_rate": 0.00018588892925926228,
      "loss": 0.6233,
      "step": 1154
    },
    {
      "epoch": 0.17254257544069315,
      "grad_norm": 0.43573567271232605,
      "learning_rate": 0.00018586484732937758,
      "loss": 0.9892,
      "step": 1155
    },
    {
      "epoch": 0.17269196295189723,
      "grad_norm": 0.11906327307224274,
      "learning_rate": 0.00018584074643058807,
      "loss": 0.7038,
      "step": 1156
    },
    {
      "epoch": 0.17284135046310128,
      "grad_norm": 0.19945280253887177,
      "learning_rate": 0.00018581662656821802,
      "loss": 0.6789,
      "step": 1157
    },
    {
      "epoch": 0.17299073797430534,
      "grad_norm": 0.25069642066955566,
      "learning_rate": 0.00018579248774759586,
      "loss": 0.9036,
      "step": 1158
    },
    {
      "epoch": 0.17314012548550942,
      "grad_norm": 0.1946730762720108,
      "learning_rate": 0.00018576832997405432,
      "loss": 0.9371,
      "step": 1159
    },
    {
      "epoch": 0.17328951299671347,
      "grad_norm": 0.16489769518375397,
      "learning_rate": 0.00018574415325293018,
      "loss": 0.5388,
      "step": 1160
    },
    {
      "epoch": 0.17343890050791755,
      "grad_norm": 0.13044323027133942,
      "learning_rate": 0.00018571995758956446,
      "loss": 0.6756,
      "step": 1161
    },
    {
      "epoch": 0.1735882880191216,
      "grad_norm": 0.13318844139575958,
      "learning_rate": 0.00018569574298930237,
      "loss": 0.9302,
      "step": 1162
    },
    {
      "epoch": 0.17373767553032565,
      "grad_norm": 0.1677417755126953,
      "learning_rate": 0.0001856715094574933,
      "loss": 0.9213,
      "step": 1163
    },
    {
      "epoch": 0.17388706304152973,
      "grad_norm": 0.11865703761577606,
      "learning_rate": 0.00018564725699949083,
      "loss": 0.8194,
      "step": 1164
    },
    {
      "epoch": 0.17403645055273378,
      "grad_norm": 0.16482964158058167,
      "learning_rate": 0.0001856229856206527,
      "loss": 0.7889,
      "step": 1165
    },
    {
      "epoch": 0.17418583806393786,
      "grad_norm": 0.13027256727218628,
      "learning_rate": 0.0001855986953263409,
      "loss": 0.7222,
      "step": 1166
    },
    {
      "epoch": 0.17433522557514192,
      "grad_norm": 0.15511631965637207,
      "learning_rate": 0.00018557438612192142,
      "loss": 1.2247,
      "step": 1167
    },
    {
      "epoch": 0.17448461308634597,
      "grad_norm": 0.16678637266159058,
      "learning_rate": 0.00018555005801276463,
      "loss": 0.5846,
      "step": 1168
    },
    {
      "epoch": 0.17463400059755005,
      "grad_norm": 0.15397848188877106,
      "learning_rate": 0.00018552571100424503,
      "loss": 0.7163,
      "step": 1169
    },
    {
      "epoch": 0.1747833881087541,
      "grad_norm": 0.3140554130077362,
      "learning_rate": 0.00018550134510174115,
      "loss": 1.1924,
      "step": 1170
    },
    {
      "epoch": 0.17493277561995818,
      "grad_norm": 0.14024198055267334,
      "learning_rate": 0.0001854769603106359,
      "loss": 0.9647,
      "step": 1171
    },
    {
      "epoch": 0.17508216313116223,
      "grad_norm": 0.2069367915391922,
      "learning_rate": 0.0001854525566363162,
      "loss": 0.9767,
      "step": 1172
    },
    {
      "epoch": 0.17523155064236628,
      "grad_norm": 0.13816118240356445,
      "learning_rate": 0.0001854281340841733,
      "loss": 0.542,
      "step": 1173
    },
    {
      "epoch": 0.17538093815357036,
      "grad_norm": 0.14822708070278168,
      "learning_rate": 0.00018540369265960242,
      "loss": 0.9574,
      "step": 1174
    },
    {
      "epoch": 0.17553032566477442,
      "grad_norm": 0.1753738522529602,
      "learning_rate": 0.00018537923236800315,
      "loss": 0.8379,
      "step": 1175
    },
    {
      "epoch": 0.1756797131759785,
      "grad_norm": 0.13829240202903748,
      "learning_rate": 0.00018535475321477906,
      "loss": 0.7342,
      "step": 1176
    },
    {
      "epoch": 0.17582910068718255,
      "grad_norm": 0.15101328492164612,
      "learning_rate": 0.00018533025520533805,
      "loss": 0.8258,
      "step": 1177
    },
    {
      "epoch": 0.1759784881983866,
      "grad_norm": 0.4592054784297943,
      "learning_rate": 0.00018530573834509215,
      "loss": 1.2556,
      "step": 1178
    },
    {
      "epoch": 0.17612787570959068,
      "grad_norm": 0.1719399243593216,
      "learning_rate": 0.00018528120263945744,
      "loss": 0.9381,
      "step": 1179
    },
    {
      "epoch": 0.17627726322079473,
      "grad_norm": 0.11263933032751083,
      "learning_rate": 0.0001852566480938543,
      "loss": 0.7793,
      "step": 1180
    },
    {
      "epoch": 0.1764266507319988,
      "grad_norm": 0.16535364091396332,
      "learning_rate": 0.00018523207471370716,
      "loss": 0.6528,
      "step": 1181
    },
    {
      "epoch": 0.17657603824320287,
      "grad_norm": 0.1505178064107895,
      "learning_rate": 0.00018520748250444474,
      "loss": 0.6552,
      "step": 1182
    },
    {
      "epoch": 0.17672542575440695,
      "grad_norm": 0.12604665756225586,
      "learning_rate": 0.00018518287147149983,
      "loss": 0.6862,
      "step": 1183
    },
    {
      "epoch": 0.176874813265611,
      "grad_norm": 0.12405974417924881,
      "learning_rate": 0.00018515824162030934,
      "loss": 0.7987,
      "step": 1184
    },
    {
      "epoch": 0.17702420077681505,
      "grad_norm": 0.15863847732543945,
      "learning_rate": 0.00018513359295631448,
      "loss": 0.7515,
      "step": 1185
    },
    {
      "epoch": 0.17717358828801913,
      "grad_norm": 0.16431525349617004,
      "learning_rate": 0.00018510892548496047,
      "loss": 0.6625,
      "step": 1186
    },
    {
      "epoch": 0.17732297579922318,
      "grad_norm": 0.2458639144897461,
      "learning_rate": 0.00018508423921169675,
      "loss": 1.3889,
      "step": 1187
    },
    {
      "epoch": 0.17747236331042726,
      "grad_norm": 0.1566503345966339,
      "learning_rate": 0.00018505953414197696,
      "loss": 0.8438,
      "step": 1188
    },
    {
      "epoch": 0.1776217508216313,
      "grad_norm": 0.12780135869979858,
      "learning_rate": 0.00018503481028125877,
      "loss": 0.9441,
      "step": 1189
    },
    {
      "epoch": 0.17777113833283537,
      "grad_norm": 0.2947339117527008,
      "learning_rate": 0.00018501006763500414,
      "loss": 1.2509,
      "step": 1190
    },
    {
      "epoch": 0.17792052584403945,
      "grad_norm": 0.23375345766544342,
      "learning_rate": 0.00018498530620867908,
      "loss": 0.6705,
      "step": 1191
    },
    {
      "epoch": 0.1780699133552435,
      "grad_norm": 0.2791805863380432,
      "learning_rate": 0.00018496052600775376,
      "loss": 0.9555,
      "step": 1192
    },
    {
      "epoch": 0.17821930086644758,
      "grad_norm": 0.14220461249351501,
      "learning_rate": 0.00018493572703770253,
      "loss": 0.8023,
      "step": 1193
    },
    {
      "epoch": 0.17836868837765163,
      "grad_norm": 0.10689699649810791,
      "learning_rate": 0.0001849109093040039,
      "loss": 0.3987,
      "step": 1194
    },
    {
      "epoch": 0.17851807588885568,
      "grad_norm": 0.11420352756977081,
      "learning_rate": 0.0001848860728121405,
      "loss": 0.6444,
      "step": 1195
    },
    {
      "epoch": 0.17866746340005976,
      "grad_norm": 0.1247362345457077,
      "learning_rate": 0.00018486121756759906,
      "loss": 0.8246,
      "step": 1196
    },
    {
      "epoch": 0.1788168509112638,
      "grad_norm": 0.15353257954120636,
      "learning_rate": 0.00018483634357587057,
      "loss": 0.5427,
      "step": 1197
    },
    {
      "epoch": 0.1789662384224679,
      "grad_norm": 0.11908576637506485,
      "learning_rate": 0.00018481145084245002,
      "loss": 0.671,
      "step": 1198
    },
    {
      "epoch": 0.17911562593367195,
      "grad_norm": 0.1491960883140564,
      "learning_rate": 0.0001847865393728366,
      "loss": 0.9276,
      "step": 1199
    },
    {
      "epoch": 0.179265013444876,
      "grad_norm": 0.16004832088947296,
      "learning_rate": 0.00018476160917253373,
      "loss": 0.9802,
      "step": 1200
    },
    {
      "epoch": 0.17941440095608008,
      "grad_norm": 0.20184099674224854,
      "learning_rate": 0.00018473666024704883,
      "loss": 0.8308,
      "step": 1201
    },
    {
      "epoch": 0.17956378846728413,
      "grad_norm": 0.16696742177009583,
      "learning_rate": 0.0001847116926018935,
      "loss": 1.1185,
      "step": 1202
    },
    {
      "epoch": 0.1797131759784882,
      "grad_norm": 0.1517256200313568,
      "learning_rate": 0.00018468670624258353,
      "loss": 0.7536,
      "step": 1203
    },
    {
      "epoch": 0.17986256348969226,
      "grad_norm": 0.16679219901561737,
      "learning_rate": 0.0001846617011746388,
      "loss": 0.6424,
      "step": 1204
    },
    {
      "epoch": 0.18001195100089631,
      "grad_norm": 0.15376229584217072,
      "learning_rate": 0.0001846366774035833,
      "loss": 0.9011,
      "step": 1205
    },
    {
      "epoch": 0.1801613385121004,
      "grad_norm": 0.13043151795864105,
      "learning_rate": 0.00018461163493494517,
      "loss": 0.9194,
      "step": 1206
    },
    {
      "epoch": 0.18031072602330445,
      "grad_norm": 0.15427030622959137,
      "learning_rate": 0.0001845865737742567,
      "loss": 0.8592,
      "step": 1207
    },
    {
      "epoch": 0.18046011353450853,
      "grad_norm": 0.4822232127189636,
      "learning_rate": 0.0001845614939270543,
      "loss": 1.6313,
      "step": 1208
    },
    {
      "epoch": 0.18060950104571258,
      "grad_norm": 0.15565815567970276,
      "learning_rate": 0.00018453639539887853,
      "loss": 0.7739,
      "step": 1209
    },
    {
      "epoch": 0.18075888855691663,
      "grad_norm": 0.11118531227111816,
      "learning_rate": 0.00018451127819527402,
      "loss": 0.6064,
      "step": 1210
    },
    {
      "epoch": 0.1809082760681207,
      "grad_norm": 0.14477503299713135,
      "learning_rate": 0.0001844861423217896,
      "loss": 0.9683,
      "step": 1211
    },
    {
      "epoch": 0.18105766357932476,
      "grad_norm": 0.1929170936346054,
      "learning_rate": 0.00018446098778397807,
      "loss": 0.6135,
      "step": 1212
    },
    {
      "epoch": 0.18120705109052884,
      "grad_norm": 0.13953596353530884,
      "learning_rate": 0.0001844358145873966,
      "loss": 1.0582,
      "step": 1213
    },
    {
      "epoch": 0.1813564386017329,
      "grad_norm": 0.1348966360092163,
      "learning_rate": 0.00018441062273760628,
      "loss": 0.4698,
      "step": 1214
    },
    {
      "epoch": 0.18150582611293695,
      "grad_norm": 0.1345159113407135,
      "learning_rate": 0.0001843854122401724,
      "loss": 0.848,
      "step": 1215
    },
    {
      "epoch": 0.18165521362414103,
      "grad_norm": 0.11462259292602539,
      "learning_rate": 0.00018436018310066435,
      "loss": 0.6769,
      "step": 1216
    },
    {
      "epoch": 0.18180460113534508,
      "grad_norm": 0.14895157516002655,
      "learning_rate": 0.0001843349353246557,
      "loss": 0.5784,
      "step": 1217
    },
    {
      "epoch": 0.18195398864654916,
      "grad_norm": 0.13859333097934723,
      "learning_rate": 0.000184309668917724,
      "loss": 0.903,
      "step": 1218
    },
    {
      "epoch": 0.1821033761577532,
      "grad_norm": 0.11198470741510391,
      "learning_rate": 0.00018428438388545107,
      "loss": 0.6247,
      "step": 1219
    },
    {
      "epoch": 0.18225276366895726,
      "grad_norm": 0.17095430195331573,
      "learning_rate": 0.0001842590802334227,
      "loss": 0.5969,
      "step": 1220
    },
    {
      "epoch": 0.18240215118016134,
      "grad_norm": 0.33381956815719604,
      "learning_rate": 0.00018423375796722895,
      "loss": 1.1353,
      "step": 1221
    },
    {
      "epoch": 0.1825515386913654,
      "grad_norm": 0.11893988400697708,
      "learning_rate": 0.00018420841709246383,
      "loss": 0.9319,
      "step": 1222
    },
    {
      "epoch": 0.18270092620256947,
      "grad_norm": 0.11547727882862091,
      "learning_rate": 0.00018418305761472566,
      "loss": 0.5685,
      "step": 1223
    },
    {
      "epoch": 0.18285031371377353,
      "grad_norm": 0.18776455521583557,
      "learning_rate": 0.0001841576795396166,
      "loss": 0.7411,
      "step": 1224
    },
    {
      "epoch": 0.18299970122497758,
      "grad_norm": 0.28690654039382935,
      "learning_rate": 0.00018413228287274315,
      "loss": 0.965,
      "step": 1225
    },
    {
      "epoch": 0.18314908873618166,
      "grad_norm": 0.12270329147577286,
      "learning_rate": 0.00018410686761971586,
      "loss": 0.6724,
      "step": 1226
    },
    {
      "epoch": 0.1832984762473857,
      "grad_norm": 0.19342897832393646,
      "learning_rate": 0.0001840814337861493,
      "loss": 0.8486,
      "step": 1227
    },
    {
      "epoch": 0.1834478637585898,
      "grad_norm": 0.12105520814657211,
      "learning_rate": 0.00018405598137766224,
      "loss": 0.9189,
      "step": 1228
    },
    {
      "epoch": 0.18359725126979384,
      "grad_norm": 0.14361943304538727,
      "learning_rate": 0.00018403051039987754,
      "loss": 0.4516,
      "step": 1229
    },
    {
      "epoch": 0.1837466387809979,
      "grad_norm": 0.23831139504909515,
      "learning_rate": 0.00018400502085842208,
      "loss": 1.2297,
      "step": 1230
    },
    {
      "epoch": 0.18389602629220198,
      "grad_norm": 0.19276589155197144,
      "learning_rate": 0.00018397951275892695,
      "loss": 0.6379,
      "step": 1231
    },
    {
      "epoch": 0.18404541380340603,
      "grad_norm": 0.32586970925331116,
      "learning_rate": 0.00018395398610702733,
      "loss": 1.0752,
      "step": 1232
    },
    {
      "epoch": 0.1841948013146101,
      "grad_norm": 0.1523769348859787,
      "learning_rate": 0.00018392844090836237,
      "loss": 1.0371,
      "step": 1233
    },
    {
      "epoch": 0.18434418882581416,
      "grad_norm": 0.1339939385652542,
      "learning_rate": 0.00018390287716857546,
      "loss": 0.5838,
      "step": 1234
    },
    {
      "epoch": 0.18449357633701824,
      "grad_norm": 0.17885640263557434,
      "learning_rate": 0.00018387729489331402,
      "loss": 1.0965,
      "step": 1235
    },
    {
      "epoch": 0.1846429638482223,
      "grad_norm": 0.2260708212852478,
      "learning_rate": 0.00018385169408822964,
      "loss": 1.0085,
      "step": 1236
    },
    {
      "epoch": 0.18479235135942634,
      "grad_norm": 0.20628871023654938,
      "learning_rate": 0.00018382607475897787,
      "loss": 0.4401,
      "step": 1237
    },
    {
      "epoch": 0.18494173887063042,
      "grad_norm": 0.17405936121940613,
      "learning_rate": 0.0001838004369112184,
      "loss": 0.5253,
      "step": 1238
    },
    {
      "epoch": 0.18509112638183448,
      "grad_norm": 0.11227359622716904,
      "learning_rate": 0.00018377478055061516,
      "loss": 0.6422,
      "step": 1239
    },
    {
      "epoch": 0.18524051389303856,
      "grad_norm": 0.18113653361797333,
      "learning_rate": 0.00018374910568283594,
      "loss": 0.5326,
      "step": 1240
    },
    {
      "epoch": 0.1853899014042426,
      "grad_norm": 0.12269013375043869,
      "learning_rate": 0.00018372341231355274,
      "loss": 0.6607,
      "step": 1241
    },
    {
      "epoch": 0.18553928891544666,
      "grad_norm": 0.31763309240341187,
      "learning_rate": 0.00018369770044844168,
      "loss": 0.8666,
      "step": 1242
    },
    {
      "epoch": 0.18568867642665074,
      "grad_norm": 0.17516912519931793,
      "learning_rate": 0.00018367197009318286,
      "loss": 1.0452,
      "step": 1243
    },
    {
      "epoch": 0.1858380639378548,
      "grad_norm": 0.11332941800355911,
      "learning_rate": 0.00018364622125346055,
      "loss": 0.4925,
      "step": 1244
    },
    {
      "epoch": 0.18598745144905887,
      "grad_norm": 0.1446741372346878,
      "learning_rate": 0.0001836204539349631,
      "loss": 0.6967,
      "step": 1245
    },
    {
      "epoch": 0.18613683896026292,
      "grad_norm": 0.177836611866951,
      "learning_rate": 0.0001835946681433829,
      "loss": 0.8671,
      "step": 1246
    },
    {
      "epoch": 0.18628622647146698,
      "grad_norm": 0.1338125616312027,
      "learning_rate": 0.00018356886388441645,
      "loss": 0.9787,
      "step": 1247
    },
    {
      "epoch": 0.18643561398267106,
      "grad_norm": 0.1504998356103897,
      "learning_rate": 0.00018354304116376425,
      "loss": 0.7123,
      "step": 1248
    },
    {
      "epoch": 0.1865850014938751,
      "grad_norm": 0.11992106586694717,
      "learning_rate": 0.00018351719998713106,
      "loss": 0.8016,
      "step": 1249
    },
    {
      "epoch": 0.1867343890050792,
      "grad_norm": 0.13362877070903778,
      "learning_rate": 0.0001834913403602255,
      "loss": 0.6951,
      "step": 1250
    },
    {
      "epoch": 0.18688377651628324,
      "grad_norm": 0.19289879500865936,
      "learning_rate": 0.00018346546228876047,
      "loss": 0.9931,
      "step": 1251
    },
    {
      "epoch": 0.1870331640274873,
      "grad_norm": 0.19565518200397491,
      "learning_rate": 0.00018343956577845276,
      "loss": 1.0326,
      "step": 1252
    },
    {
      "epoch": 0.18718255153869137,
      "grad_norm": 0.2017303854227066,
      "learning_rate": 0.00018341365083502335,
      "loss": 0.8296,
      "step": 1253
    },
    {
      "epoch": 0.18733193904989542,
      "grad_norm": 0.14418242871761322,
      "learning_rate": 0.00018338771746419726,
      "loss": 0.9156,
      "step": 1254
    },
    {
      "epoch": 0.1874813265610995,
      "grad_norm": 0.30200129747390747,
      "learning_rate": 0.00018336176567170363,
      "loss": 0.8995,
      "step": 1255
    },
    {
      "epoch": 0.18763071407230356,
      "grad_norm": 0.2137260138988495,
      "learning_rate": 0.00018333579546327556,
      "loss": 0.9876,
      "step": 1256
    },
    {
      "epoch": 0.1877801015835076,
      "grad_norm": 0.1436002552509308,
      "learning_rate": 0.0001833098068446503,
      "loss": 0.7803,
      "step": 1257
    },
    {
      "epoch": 0.1879294890947117,
      "grad_norm": 0.213914155960083,
      "learning_rate": 0.00018328379982156915,
      "loss": 0.8788,
      "step": 1258
    },
    {
      "epoch": 0.18807887660591574,
      "grad_norm": 0.18235763907432556,
      "learning_rate": 0.00018325777439977747,
      "loss": 0.6727,
      "step": 1259
    },
    {
      "epoch": 0.18822826411711982,
      "grad_norm": 0.184591144323349,
      "learning_rate": 0.00018323173058502472,
      "loss": 0.7196,
      "step": 1260
    },
    {
      "epoch": 0.18837765162832387,
      "grad_norm": 0.1435452103614807,
      "learning_rate": 0.00018320566838306431,
      "loss": 0.9443,
      "step": 1261
    },
    {
      "epoch": 0.18852703913952792,
      "grad_norm": 0.1862800419330597,
      "learning_rate": 0.00018317958779965387,
      "loss": 0.704,
      "step": 1262
    },
    {
      "epoch": 0.188676426650732,
      "grad_norm": 0.1972978413105011,
      "learning_rate": 0.000183153488840555,
      "loss": 0.7077,
      "step": 1263
    },
    {
      "epoch": 0.18882581416193606,
      "grad_norm": 0.13390903174877167,
      "learning_rate": 0.00018312737151153334,
      "loss": 0.658,
      "step": 1264
    },
    {
      "epoch": 0.18897520167314014,
      "grad_norm": 0.3871189057826996,
      "learning_rate": 0.00018310123581835868,
      "loss": 1.2051,
      "step": 1265
    },
    {
      "epoch": 0.1891245891843442,
      "grad_norm": 0.15519294142723083,
      "learning_rate": 0.00018307508176680472,
      "loss": 0.6735,
      "step": 1266
    },
    {
      "epoch": 0.18927397669554824,
      "grad_norm": 0.13292644917964935,
      "learning_rate": 0.0001830489093626494,
      "loss": 0.5655,
      "step": 1267
    },
    {
      "epoch": 0.18942336420675232,
      "grad_norm": 0.12040957808494568,
      "learning_rate": 0.00018302271861167456,
      "loss": 0.6574,
      "step": 1268
    },
    {
      "epoch": 0.18957275171795637,
      "grad_norm": 0.17542392015457153,
      "learning_rate": 0.00018299650951966614,
      "loss": 0.7045,
      "step": 1269
    },
    {
      "epoch": 0.18972213922916045,
      "grad_norm": 0.12977173924446106,
      "learning_rate": 0.0001829702820924142,
      "loss": 0.8453,
      "step": 1270
    },
    {
      "epoch": 0.1898715267403645,
      "grad_norm": 0.3114042282104492,
      "learning_rate": 0.00018294403633571275,
      "loss": 0.9561,
      "step": 1271
    },
    {
      "epoch": 0.19002091425156856,
      "grad_norm": 0.1601576954126358,
      "learning_rate": 0.00018291777225535994,
      "loss": 0.8407,
      "step": 1272
    },
    {
      "epoch": 0.19017030176277264,
      "grad_norm": 0.12280754745006561,
      "learning_rate": 0.00018289148985715784,
      "loss": 0.5095,
      "step": 1273
    },
    {
      "epoch": 0.1903196892739767,
      "grad_norm": 0.12627717852592468,
      "learning_rate": 0.00018286518914691272,
      "loss": 0.5383,
      "step": 1274
    },
    {
      "epoch": 0.19046907678518077,
      "grad_norm": 0.13007701933383942,
      "learning_rate": 0.0001828388701304348,
      "loss": 0.8532,
      "step": 1275
    },
    {
      "epoch": 0.19061846429638482,
      "grad_norm": 0.13100433349609375,
      "learning_rate": 0.00018281253281353838,
      "loss": 0.822,
      "step": 1276
    },
    {
      "epoch": 0.19076785180758887,
      "grad_norm": 0.1324629783630371,
      "learning_rate": 0.0001827861772020418,
      "loss": 0.7904,
      "step": 1277
    },
    {
      "epoch": 0.19091723931879295,
      "grad_norm": 0.13711746037006378,
      "learning_rate": 0.00018275980330176737,
      "loss": 0.498,
      "step": 1278
    },
    {
      "epoch": 0.191066626829997,
      "grad_norm": 0.12584207952022552,
      "learning_rate": 0.00018273341111854153,
      "loss": 1.0128,
      "step": 1279
    },
    {
      "epoch": 0.19121601434120108,
      "grad_norm": 0.20175109803676605,
      "learning_rate": 0.00018270700065819477,
      "loss": 0.6671,
      "step": 1280
    },
    {
      "epoch": 0.19136540185240514,
      "grad_norm": 0.17598234117031097,
      "learning_rate": 0.00018268057192656156,
      "loss": 1.0968,
      "step": 1281
    },
    {
      "epoch": 0.1915147893636092,
      "grad_norm": 0.13300736248493195,
      "learning_rate": 0.00018265412492948042,
      "loss": 0.7158,
      "step": 1282
    },
    {
      "epoch": 0.19166417687481327,
      "grad_norm": 0.1788729578256607,
      "learning_rate": 0.00018262765967279386,
      "loss": 0.6676,
      "step": 1283
    },
    {
      "epoch": 0.19181356438601732,
      "grad_norm": 0.23207572102546692,
      "learning_rate": 0.0001826011761623486,
      "loss": 0.9164,
      "step": 1284
    },
    {
      "epoch": 0.1919629518972214,
      "grad_norm": 0.1788550764322281,
      "learning_rate": 0.00018257467440399515,
      "loss": 1.1415,
      "step": 1285
    },
    {
      "epoch": 0.19211233940842545,
      "grad_norm": 0.10894573479890823,
      "learning_rate": 0.0001825481544035882,
      "loss": 0.716,
      "step": 1286
    },
    {
      "epoch": 0.1922617269196295,
      "grad_norm": 0.12275734543800354,
      "learning_rate": 0.00018252161616698646,
      "loss": 0.6821,
      "step": 1287
    },
    {
      "epoch": 0.19241111443083359,
      "grad_norm": 0.12295856326818466,
      "learning_rate": 0.00018249505970005262,
      "loss": 0.6685,
      "step": 1288
    },
    {
      "epoch": 0.19256050194203764,
      "grad_norm": 0.25377437472343445,
      "learning_rate": 0.00018246848500865347,
      "loss": 0.9663,
      "step": 1289
    },
    {
      "epoch": 0.19270988945324172,
      "grad_norm": 0.12609486281871796,
      "learning_rate": 0.00018244189209865974,
      "loss": 0.8103,
      "step": 1290
    },
    {
      "epoch": 0.19285927696444577,
      "grad_norm": 0.18513751029968262,
      "learning_rate": 0.0001824152809759462,
      "loss": 1.1248,
      "step": 1291
    },
    {
      "epoch": 0.19300866447564985,
      "grad_norm": 0.13006910681724548,
      "learning_rate": 0.00018238865164639173,
      "loss": 0.6911,
      "step": 1292
    },
    {
      "epoch": 0.1931580519868539,
      "grad_norm": 0.14160241186618805,
      "learning_rate": 0.00018236200411587915,
      "loss": 0.5689,
      "step": 1293
    },
    {
      "epoch": 0.19330743949805795,
      "grad_norm": 0.17542794346809387,
      "learning_rate": 0.0001823353383902953,
      "loss": 0.6207,
      "step": 1294
    },
    {
      "epoch": 0.19345682700926203,
      "grad_norm": 0.251878559589386,
      "learning_rate": 0.00018230865447553107,
      "loss": 0.9893,
      "step": 1295
    },
    {
      "epoch": 0.19360621452046609,
      "grad_norm": 0.14598213136196136,
      "learning_rate": 0.0001822819523774814,
      "loss": 0.823,
      "step": 1296
    },
    {
      "epoch": 0.19375560203167017,
      "grad_norm": 0.14267736673355103,
      "learning_rate": 0.0001822552321020451,
      "loss": 0.7624,
      "step": 1297
    },
    {
      "epoch": 0.19390498954287422,
      "grad_norm": 0.2022227942943573,
      "learning_rate": 0.00018222849365512523,
      "loss": 0.8711,
      "step": 1298
    },
    {
      "epoch": 0.19405437705407827,
      "grad_norm": 0.29249998927116394,
      "learning_rate": 0.0001822017370426287,
      "loss": 0.9054,
      "step": 1299
    },
    {
      "epoch": 0.19420376456528235,
      "grad_norm": 0.12678225338459015,
      "learning_rate": 0.0001821749622704664,
      "loss": 0.6689,
      "step": 1300
    },
    {
      "epoch": 0.1943531520764864,
      "grad_norm": 0.1714026778936386,
      "learning_rate": 0.00018214816934455333,
      "loss": 0.7199,
      "step": 1301
    },
    {
      "epoch": 0.19450253958769048,
      "grad_norm": 0.18611545860767365,
      "learning_rate": 0.00018212135827080857,
      "loss": 0.9119,
      "step": 1302
    },
    {
      "epoch": 0.19465192709889453,
      "grad_norm": 0.12778407335281372,
      "learning_rate": 0.00018209452905515496,
      "loss": 0.435,
      "step": 1303
    },
    {
      "epoch": 0.19480131461009859,
      "grad_norm": 0.3280845880508423,
      "learning_rate": 0.00018206768170351962,
      "loss": 0.9022,
      "step": 1304
    },
    {
      "epoch": 0.19495070212130267,
      "grad_norm": 0.13167747855186462,
      "learning_rate": 0.0001820408162218335,
      "loss": 0.8323,
      "step": 1305
    },
    {
      "epoch": 0.19510008963250672,
      "grad_norm": 0.14176693558692932,
      "learning_rate": 0.0001820139326160316,
      "loss": 0.5333,
      "step": 1306
    },
    {
      "epoch": 0.1952494771437108,
      "grad_norm": 0.20718522369861603,
      "learning_rate": 0.00018198703089205293,
      "loss": 0.8403,
      "step": 1307
    },
    {
      "epoch": 0.19539886465491485,
      "grad_norm": 0.15032266080379486,
      "learning_rate": 0.00018196011105584058,
      "loss": 1.1825,
      "step": 1308
    },
    {
      "epoch": 0.1955482521661189,
      "grad_norm": 0.13833296298980713,
      "learning_rate": 0.0001819331731133415,
      "loss": 1.0256,
      "step": 1309
    },
    {
      "epoch": 0.19569763967732298,
      "grad_norm": 0.15338090062141418,
      "learning_rate": 0.00018190621707050671,
      "loss": 0.7619,
      "step": 1310
    },
    {
      "epoch": 0.19584702718852703,
      "grad_norm": 0.15736734867095947,
      "learning_rate": 0.00018187924293329124,
      "loss": 0.9195,
      "step": 1311
    },
    {
      "epoch": 0.19599641469973111,
      "grad_norm": 0.1757577806711197,
      "learning_rate": 0.0001818522507076541,
      "loss": 0.8208,
      "step": 1312
    },
    {
      "epoch": 0.19614580221093517,
      "grad_norm": 0.16938672959804535,
      "learning_rate": 0.00018182524039955832,
      "loss": 0.6394,
      "step": 1313
    },
    {
      "epoch": 0.19629518972213922,
      "grad_norm": 0.11935129016637802,
      "learning_rate": 0.00018179821201497092,
      "loss": 0.8837,
      "step": 1314
    },
    {
      "epoch": 0.1964445772333433,
      "grad_norm": 0.1422419250011444,
      "learning_rate": 0.00018177116555986283,
      "loss": 0.8579,
      "step": 1315
    },
    {
      "epoch": 0.19659396474454735,
      "grad_norm": 0.16648200154304504,
      "learning_rate": 0.0001817441010402091,
      "loss": 0.4853,
      "step": 1316
    },
    {
      "epoch": 0.19674335225575143,
      "grad_norm": 0.16787053644657135,
      "learning_rate": 0.00018171701846198866,
      "loss": 0.5068,
      "step": 1317
    },
    {
      "epoch": 0.19689273976695548,
      "grad_norm": 0.13628660142421722,
      "learning_rate": 0.00018168991783118452,
      "loss": 0.7062,
      "step": 1318
    },
    {
      "epoch": 0.19704212727815953,
      "grad_norm": 0.11356477439403534,
      "learning_rate": 0.00018166279915378364,
      "loss": 0.5145,
      "step": 1319
    },
    {
      "epoch": 0.19719151478936361,
      "grad_norm": 0.1292952299118042,
      "learning_rate": 0.00018163566243577697,
      "loss": 0.7452,
      "step": 1320
    },
    {
      "epoch": 0.19734090230056767,
      "grad_norm": 0.17471875250339508,
      "learning_rate": 0.00018160850768315941,
      "loss": 0.9093,
      "step": 1321
    },
    {
      "epoch": 0.19749028981177175,
      "grad_norm": 0.1547754555940628,
      "learning_rate": 0.0001815813349019299,
      "loss": 0.6851,
      "step": 1322
    },
    {
      "epoch": 0.1976396773229758,
      "grad_norm": 0.11787121742963791,
      "learning_rate": 0.00018155414409809132,
      "loss": 0.8243,
      "step": 1323
    },
    {
      "epoch": 0.19778906483417985,
      "grad_norm": 0.13274557888507843,
      "learning_rate": 0.00018152693527765057,
      "loss": 0.7042,
      "step": 1324
    },
    {
      "epoch": 0.19793845234538393,
      "grad_norm": 0.1521092653274536,
      "learning_rate": 0.00018149970844661849,
      "loss": 0.5151,
      "step": 1325
    },
    {
      "epoch": 0.19808783985658798,
      "grad_norm": 0.13765467703342438,
      "learning_rate": 0.0001814724636110099,
      "loss": 0.8489,
      "step": 1326
    },
    {
      "epoch": 0.19823722736779206,
      "grad_norm": 0.26832038164138794,
      "learning_rate": 0.00018144520077684369,
      "loss": 0.7142,
      "step": 1327
    },
    {
      "epoch": 0.19838661487899611,
      "grad_norm": 0.10923701524734497,
      "learning_rate": 0.00018141791995014255,
      "loss": 0.6142,
      "step": 1328
    },
    {
      "epoch": 0.19853600239020017,
      "grad_norm": 0.1848011165857315,
      "learning_rate": 0.00018139062113693333,
      "loss": 0.8898,
      "step": 1329
    },
    {
      "epoch": 0.19868538990140425,
      "grad_norm": 0.12784364819526672,
      "learning_rate": 0.00018136330434324674,
      "loss": 0.8846,
      "step": 1330
    },
    {
      "epoch": 0.1988347774126083,
      "grad_norm": 0.13698415458202362,
      "learning_rate": 0.00018133596957511748,
      "loss": 0.7992,
      "step": 1331
    },
    {
      "epoch": 0.19898416492381238,
      "grad_norm": 0.16566555202007294,
      "learning_rate": 0.00018130861683858426,
      "loss": 0.8587,
      "step": 1332
    },
    {
      "epoch": 0.19913355243501643,
      "grad_norm": 0.12000430375337601,
      "learning_rate": 0.0001812812461396897,
      "loss": 0.815,
      "step": 1333
    },
    {
      "epoch": 0.19928293994622048,
      "grad_norm": 0.11932047456502914,
      "learning_rate": 0.00018125385748448048,
      "loss": 0.8236,
      "step": 1334
    },
    {
      "epoch": 0.19943232745742456,
      "grad_norm": 0.14856404066085815,
      "learning_rate": 0.00018122645087900708,
      "loss": 0.5428,
      "step": 1335
    },
    {
      "epoch": 0.19958171496862862,
      "grad_norm": 0.22421959042549133,
      "learning_rate": 0.00018119902632932416,
      "loss": 1.4216,
      "step": 1336
    },
    {
      "epoch": 0.1997311024798327,
      "grad_norm": 0.1441725790500641,
      "learning_rate": 0.00018117158384149023,
      "loss": 1.0382,
      "step": 1337
    },
    {
      "epoch": 0.19988048999103675,
      "grad_norm": 0.14890751242637634,
      "learning_rate": 0.0001811441234215677,
      "loss": 0.6485,
      "step": 1338
    },
    {
      "epoch": 0.2000298775022408,
      "grad_norm": 0.24321304261684418,
      "learning_rate": 0.00018111664507562304,
      "loss": 0.8343,
      "step": 1339
    },
    {
      "epoch": 0.20017926501344488,
      "grad_norm": 0.20097461342811584,
      "learning_rate": 0.0001810891488097267,
      "loss": 0.9692,
      "step": 1340
    },
    {
      "epoch": 0.20032865252464893,
      "grad_norm": 0.15573135018348694,
      "learning_rate": 0.00018106163462995297,
      "loss": 0.5672,
      "step": 1341
    },
    {
      "epoch": 0.200478040035853,
      "grad_norm": 0.1436656415462494,
      "learning_rate": 0.00018103410254238021,
      "loss": 0.7355,
      "step": 1342
    },
    {
      "epoch": 0.20062742754705706,
      "grad_norm": 0.28203514218330383,
      "learning_rate": 0.00018100655255309068,
      "loss": 1.5894,
      "step": 1343
    },
    {
      "epoch": 0.20077681505826112,
      "grad_norm": 0.15477240085601807,
      "learning_rate": 0.0001809789846681706,
      "loss": 0.9943,
      "step": 1344
    },
    {
      "epoch": 0.2009262025694652,
      "grad_norm": 0.20752249658107758,
      "learning_rate": 0.0001809513988937102,
      "loss": 0.6022,
      "step": 1345
    },
    {
      "epoch": 0.20107559008066925,
      "grad_norm": 0.12718671560287476,
      "learning_rate": 0.00018092379523580357,
      "loss": 0.7869,
      "step": 1346
    },
    {
      "epoch": 0.20122497759187333,
      "grad_norm": 0.14520499110221863,
      "learning_rate": 0.0001808961737005488,
      "loss": 0.8307,
      "step": 1347
    },
    {
      "epoch": 0.20137436510307738,
      "grad_norm": 0.18366779386997223,
      "learning_rate": 0.00018086853429404793,
      "loss": 1.0939,
      "step": 1348
    },
    {
      "epoch": 0.20152375261428146,
      "grad_norm": 0.15729284286499023,
      "learning_rate": 0.00018084087702240694,
      "loss": 0.7747,
      "step": 1349
    },
    {
      "epoch": 0.2016731401254855,
      "grad_norm": 0.23096966743469238,
      "learning_rate": 0.00018081320189173577,
      "loss": 0.8718,
      "step": 1350
    },
    {
      "epoch": 0.20182252763668956,
      "grad_norm": 0.17240802943706512,
      "learning_rate": 0.00018078550890814827,
      "loss": 0.7155,
      "step": 1351
    },
    {
      "epoch": 0.20197191514789364,
      "grad_norm": 0.1575528234243393,
      "learning_rate": 0.0001807577980777623,
      "loss": 0.7335,
      "step": 1352
    },
    {
      "epoch": 0.2021213026590977,
      "grad_norm": 0.16896086931228638,
      "learning_rate": 0.00018073006940669956,
      "loss": 1.0208,
      "step": 1353
    },
    {
      "epoch": 0.20227069017030178,
      "grad_norm": 0.2874300479888916,
      "learning_rate": 0.00018070232290108584,
      "loss": 1.075,
      "step": 1354
    },
    {
      "epoch": 0.20242007768150583,
      "grad_norm": 0.12777869403362274,
      "learning_rate": 0.0001806745585670507,
      "loss": 0.4453,
      "step": 1355
    },
    {
      "epoch": 0.20256946519270988,
      "grad_norm": 0.28166279196739197,
      "learning_rate": 0.00018064677641072775,
      "loss": 0.9604,
      "step": 1356
    },
    {
      "epoch": 0.20271885270391396,
      "grad_norm": 0.1567370742559433,
      "learning_rate": 0.0001806189764382545,
      "loss": 0.8008,
      "step": 1357
    },
    {
      "epoch": 0.202868240215118,
      "grad_norm": 0.12733782827854156,
      "learning_rate": 0.00018059115865577249,
      "loss": 0.8105,
      "step": 1358
    },
    {
      "epoch": 0.2030176277263221,
      "grad_norm": 0.1359173059463501,
      "learning_rate": 0.00018056332306942696,
      "loss": 0.8775,
      "step": 1359
    },
    {
      "epoch": 0.20316701523752614,
      "grad_norm": 0.10855043679475784,
      "learning_rate": 0.00018053546968536735,
      "loss": 0.6566,
      "step": 1360
    },
    {
      "epoch": 0.2033164027487302,
      "grad_norm": 0.1817154735326767,
      "learning_rate": 0.00018050759850974683,
      "loss": 0.4578,
      "step": 1361
    },
    {
      "epoch": 0.20346579025993428,
      "grad_norm": 0.13071177899837494,
      "learning_rate": 0.00018047970954872264,
      "loss": 0.6817,
      "step": 1362
    },
    {
      "epoch": 0.20361517777113833,
      "grad_norm": 0.26592084765434265,
      "learning_rate": 0.0001804518028084559,
      "loss": 0.9979,
      "step": 1363
    },
    {
      "epoch": 0.2037645652823424,
      "grad_norm": 0.17801976203918457,
      "learning_rate": 0.0001804238782951116,
      "loss": 0.5385,
      "step": 1364
    },
    {
      "epoch": 0.20391395279354646,
      "grad_norm": 0.13625523447990417,
      "learning_rate": 0.00018039593601485874,
      "loss": 0.7867,
      "step": 1365
    },
    {
      "epoch": 0.2040633403047505,
      "grad_norm": 0.12862275540828705,
      "learning_rate": 0.00018036797597387023,
      "loss": 0.8467,
      "step": 1366
    },
    {
      "epoch": 0.2042127278159546,
      "grad_norm": 0.709415078163147,
      "learning_rate": 0.00018033999817832286,
      "loss": 1.4571,
      "step": 1367
    },
    {
      "epoch": 0.20436211532715864,
      "grad_norm": 0.13892705738544464,
      "learning_rate": 0.00018031200263439736,
      "loss": 0.7435,
      "step": 1368
    },
    {
      "epoch": 0.20451150283836272,
      "grad_norm": 0.14498738944530487,
      "learning_rate": 0.0001802839893482784,
      "loss": 1.0691,
      "step": 1369
    },
    {
      "epoch": 0.20466089034956678,
      "grad_norm": 0.34637802839279175,
      "learning_rate": 0.00018025595832615459,
      "loss": 0.8286,
      "step": 1370
    },
    {
      "epoch": 0.20481027786077083,
      "grad_norm": 0.12984217703342438,
      "learning_rate": 0.00018022790957421836,
      "loss": 0.6684,
      "step": 1371
    },
    {
      "epoch": 0.2049596653719749,
      "grad_norm": 0.10101130604743958,
      "learning_rate": 0.00018019984309866619,
      "loss": 0.5913,
      "step": 1372
    },
    {
      "epoch": 0.20510905288317896,
      "grad_norm": 0.20061518251895905,
      "learning_rate": 0.00018017175890569834,
      "loss": 1.0633,
      "step": 1373
    },
    {
      "epoch": 0.20525844039438304,
      "grad_norm": 0.17312417924404144,
      "learning_rate": 0.00018014365700151912,
      "loss": 0.8474,
      "step": 1374
    },
    {
      "epoch": 0.2054078279055871,
      "grad_norm": 0.14086449146270752,
      "learning_rate": 0.00018011553739233666,
      "loss": 0.4827,
      "step": 1375
    },
    {
      "epoch": 0.20555721541679114,
      "grad_norm": 0.21013930439949036,
      "learning_rate": 0.000180087400084363,
      "loss": 0.8096,
      "step": 1376
    },
    {
      "epoch": 0.20570660292799522,
      "grad_norm": 0.2212519496679306,
      "learning_rate": 0.0001800592450838142,
      "loss": 1.0053,
      "step": 1377
    },
    {
      "epoch": 0.20585599043919928,
      "grad_norm": 0.15844549238681793,
      "learning_rate": 0.00018003107239691004,
      "loss": 0.7247,
      "step": 1378
    },
    {
      "epoch": 0.20600537795040336,
      "grad_norm": 0.12048251181840897,
      "learning_rate": 0.00018000288202987437,
      "loss": 0.6577,
      "step": 1379
    },
    {
      "epoch": 0.2061547654616074,
      "grad_norm": 0.15412500500679016,
      "learning_rate": 0.00017997467398893488,
      "loss": 0.8612,
      "step": 1380
    },
    {
      "epoch": 0.20630415297281146,
      "grad_norm": 0.2080097794532776,
      "learning_rate": 0.00017994644828032316,
      "loss": 1.1714,
      "step": 1381
    },
    {
      "epoch": 0.20645354048401554,
      "grad_norm": 0.10042672604322433,
      "learning_rate": 0.00017991820491027472,
      "loss": 0.4871,
      "step": 1382
    },
    {
      "epoch": 0.2066029279952196,
      "grad_norm": 0.12917503714561462,
      "learning_rate": 0.00017988994388502903,
      "loss": 0.9816,
      "step": 1383
    },
    {
      "epoch": 0.20675231550642367,
      "grad_norm": 0.1451895833015442,
      "learning_rate": 0.0001798616652108293,
      "loss": 0.7564,
      "step": 1384
    },
    {
      "epoch": 0.20690170301762772,
      "grad_norm": 0.12688519060611725,
      "learning_rate": 0.0001798333688939228,
      "loss": 0.9622,
      "step": 1385
    },
    {
      "epoch": 0.20705109052883178,
      "grad_norm": 0.11722414195537567,
      "learning_rate": 0.00017980505494056062,
      "loss": 0.7,
      "step": 1386
    },
    {
      "epoch": 0.20720047804003586,
      "grad_norm": 0.12333224713802338,
      "learning_rate": 0.00017977672335699776,
      "loss": 0.6952,
      "step": 1387
    },
    {
      "epoch": 0.2073498655512399,
      "grad_norm": 0.11864780634641647,
      "learning_rate": 0.00017974837414949307,
      "loss": 0.6831,
      "step": 1388
    },
    {
      "epoch": 0.207499253062444,
      "grad_norm": 0.16104672849178314,
      "learning_rate": 0.00017972000732430942,
      "loss": 0.8635,
      "step": 1389
    },
    {
      "epoch": 0.20764864057364804,
      "grad_norm": 0.13945555686950684,
      "learning_rate": 0.00017969162288771347,
      "loss": 0.9024,
      "step": 1390
    },
    {
      "epoch": 0.2077980280848521,
      "grad_norm": 0.11853618174791336,
      "learning_rate": 0.00017966322084597572,
      "loss": 0.7451,
      "step": 1391
    },
    {
      "epoch": 0.20794741559605617,
      "grad_norm": 0.12260474264621735,
      "learning_rate": 0.0001796348012053707,
      "loss": 0.5602,
      "step": 1392
    },
    {
      "epoch": 0.20809680310726023,
      "grad_norm": 0.2517350912094116,
      "learning_rate": 0.0001796063639721768,
      "loss": 0.6186,
      "step": 1393
    },
    {
      "epoch": 0.2082461906184643,
      "grad_norm": 0.2064390927553177,
      "learning_rate": 0.00017957790915267615,
      "loss": 0.8175,
      "step": 1394
    },
    {
      "epoch": 0.20839557812966836,
      "grad_norm": 0.24569760262966156,
      "learning_rate": 0.0001795494367531549,
      "loss": 0.8564,
      "step": 1395
    },
    {
      "epoch": 0.2085449656408724,
      "grad_norm": 0.2579876780509949,
      "learning_rate": 0.0001795209467799031,
      "loss": 1.3103,
      "step": 1396
    },
    {
      "epoch": 0.2086943531520765,
      "grad_norm": 0.24793849885463715,
      "learning_rate": 0.00017949243923921462,
      "loss": 0.8798,
      "step": 1397
    },
    {
      "epoch": 0.20884374066328054,
      "grad_norm": 0.32119929790496826,
      "learning_rate": 0.0001794639141373872,
      "loss": 0.9218,
      "step": 1398
    },
    {
      "epoch": 0.20899312817448462,
      "grad_norm": 0.17471039295196533,
      "learning_rate": 0.00017943537148072252,
      "loss": 0.5358,
      "step": 1399
    },
    {
      "epoch": 0.20914251568568867,
      "grad_norm": 0.1545010507106781,
      "learning_rate": 0.00017940681127552604,
      "loss": 0.7135,
      "step": 1400
    },
    {
      "epoch": 0.20929190319689275,
      "grad_norm": 0.15277154743671417,
      "learning_rate": 0.00017937823352810725,
      "loss": 0.8968,
      "step": 1401
    },
    {
      "epoch": 0.2094412907080968,
      "grad_norm": 0.14847959578037262,
      "learning_rate": 0.0001793496382447794,
      "loss": 0.6325,
      "step": 1402
    },
    {
      "epoch": 0.20959067821930086,
      "grad_norm": 0.14463500678539276,
      "learning_rate": 0.00017932102543185963,
      "loss": 0.4591,
      "step": 1403
    },
    {
      "epoch": 0.20974006573050494,
      "grad_norm": 0.12656496465206146,
      "learning_rate": 0.00017929239509566894,
      "loss": 0.8605,
      "step": 1404
    },
    {
      "epoch": 0.209889453241709,
      "grad_norm": 0.3494865298271179,
      "learning_rate": 0.00017926374724253226,
      "loss": 1.0462,
      "step": 1405
    },
    {
      "epoch": 0.21003884075291307,
      "grad_norm": 0.2256232500076294,
      "learning_rate": 0.00017923508187877834,
      "loss": 0.6951,
      "step": 1406
    },
    {
      "epoch": 0.21018822826411712,
      "grad_norm": 0.14871495962142944,
      "learning_rate": 0.00017920639901073976,
      "loss": 0.6265,
      "step": 1407
    },
    {
      "epoch": 0.21033761577532117,
      "grad_norm": 0.12966501712799072,
      "learning_rate": 0.00017917769864475314,
      "loss": 0.8843,
      "step": 1408
    },
    {
      "epoch": 0.21048700328652525,
      "grad_norm": 0.14218322932720184,
      "learning_rate": 0.00017914898078715875,
      "loss": 0.5568,
      "step": 1409
    },
    {
      "epoch": 0.2106363907977293,
      "grad_norm": 0.17389464378356934,
      "learning_rate": 0.00017912024544430088,
      "loss": 1.0362,
      "step": 1410
    },
    {
      "epoch": 0.21078577830893339,
      "grad_norm": 0.18730613589286804,
      "learning_rate": 0.00017909149262252755,
      "loss": 0.5673,
      "step": 1411
    },
    {
      "epoch": 0.21093516582013744,
      "grad_norm": 0.14220395684242249,
      "learning_rate": 0.0001790627223281908,
      "loss": 0.9562,
      "step": 1412
    },
    {
      "epoch": 0.2110845533313415,
      "grad_norm": 0.15215925872325897,
      "learning_rate": 0.00017903393456764635,
      "loss": 0.9981,
      "step": 1413
    },
    {
      "epoch": 0.21123394084254557,
      "grad_norm": 0.22159068286418915,
      "learning_rate": 0.00017900512934725397,
      "loss": 0.9884,
      "step": 1414
    },
    {
      "epoch": 0.21138332835374962,
      "grad_norm": 0.13190264999866486,
      "learning_rate": 0.0001789763066733771,
      "loss": 0.7929,
      "step": 1415
    },
    {
      "epoch": 0.2115327158649537,
      "grad_norm": 0.13530825078487396,
      "learning_rate": 0.0001789474665523832,
      "loss": 0.8792,
      "step": 1416
    },
    {
      "epoch": 0.21168210337615775,
      "grad_norm": 0.14375026524066925,
      "learning_rate": 0.00017891860899064342,
      "loss": 0.8539,
      "step": 1417
    },
    {
      "epoch": 0.2118314908873618,
      "grad_norm": 0.13115522265434265,
      "learning_rate": 0.00017888973399453296,
      "loss": 0.8168,
      "step": 1418
    },
    {
      "epoch": 0.2119808783985659,
      "grad_norm": 0.6125191450119019,
      "learning_rate": 0.0001788608415704307,
      "loss": 1.2228,
      "step": 1419
    },
    {
      "epoch": 0.21213026590976994,
      "grad_norm": 0.11920984089374542,
      "learning_rate": 0.00017883193172471944,
      "loss": 0.8112,
      "step": 1420
    },
    {
      "epoch": 0.21227965342097402,
      "grad_norm": 0.16258184611797333,
      "learning_rate": 0.00017880300446378584,
      "loss": 0.8269,
      "step": 1421
    },
    {
      "epoch": 0.21242904093217807,
      "grad_norm": 0.18811704218387604,
      "learning_rate": 0.00017877405979402038,
      "loss": 0.7807,
      "step": 1422
    },
    {
      "epoch": 0.21257842844338212,
      "grad_norm": 0.18149599432945251,
      "learning_rate": 0.00017874509772181738,
      "loss": 0.7866,
      "step": 1423
    },
    {
      "epoch": 0.2127278159545862,
      "grad_norm": 0.2281455546617508,
      "learning_rate": 0.00017871611825357502,
      "loss": 0.9836,
      "step": 1424
    },
    {
      "epoch": 0.21287720346579025,
      "grad_norm": 0.1633453071117401,
      "learning_rate": 0.00017868712139569535,
      "loss": 0.9219,
      "step": 1425
    },
    {
      "epoch": 0.21302659097699433,
      "grad_norm": 0.6165900230407715,
      "learning_rate": 0.00017865810715458427,
      "loss": 1.3481,
      "step": 1426
    },
    {
      "epoch": 0.2131759784881984,
      "grad_norm": 0.17836418747901917,
      "learning_rate": 0.0001786290755366514,
      "loss": 0.9347,
      "step": 1427
    },
    {
      "epoch": 0.21332536599940244,
      "grad_norm": 0.19245529174804688,
      "learning_rate": 0.00017860002654831032,
      "loss": 0.4793,
      "step": 1428
    },
    {
      "epoch": 0.21347475351060652,
      "grad_norm": 0.13874202966690063,
      "learning_rate": 0.00017857096019597844,
      "loss": 0.8586,
      "step": 1429
    },
    {
      "epoch": 0.21362414102181057,
      "grad_norm": 0.26488029956817627,
      "learning_rate": 0.00017854187648607694,
      "loss": 1.1623,
      "step": 1430
    },
    {
      "epoch": 0.21377352853301465,
      "grad_norm": 0.10857315361499786,
      "learning_rate": 0.00017851277542503085,
      "loss": 0.6721,
      "step": 1431
    },
    {
      "epoch": 0.2139229160442187,
      "grad_norm": 0.1778874397277832,
      "learning_rate": 0.00017848365701926913,
      "loss": 0.8528,
      "step": 1432
    },
    {
      "epoch": 0.21407230355542275,
      "grad_norm": 0.1809145212173462,
      "learning_rate": 0.00017845452127522444,
      "loss": 1.001,
      "step": 1433
    },
    {
      "epoch": 0.21422169106662683,
      "grad_norm": 0.17359715700149536,
      "learning_rate": 0.00017842536819933337,
      "loss": 1.0188,
      "step": 1434
    },
    {
      "epoch": 0.2143710785778309,
      "grad_norm": 0.13792802393436432,
      "learning_rate": 0.00017839619779803627,
      "loss": 0.9398,
      "step": 1435
    },
    {
      "epoch": 0.21452046608903497,
      "grad_norm": 0.1106923520565033,
      "learning_rate": 0.0001783670100777773,
      "loss": 0.6102,
      "step": 1436
    },
    {
      "epoch": 0.21466985360023902,
      "grad_norm": 0.1782097965478897,
      "learning_rate": 0.0001783378050450046,
      "loss": 0.4704,
      "step": 1437
    },
    {
      "epoch": 0.21481924111144307,
      "grad_norm": 0.1738913655281067,
      "learning_rate": 0.0001783085827061699,
      "loss": 0.7214,
      "step": 1438
    },
    {
      "epoch": 0.21496862862264715,
      "grad_norm": 0.15943187475204468,
      "learning_rate": 0.00017827934306772897,
      "loss": 0.962,
      "step": 1439
    },
    {
      "epoch": 0.2151180161338512,
      "grad_norm": 0.14920490980148315,
      "learning_rate": 0.00017825008613614127,
      "loss": 0.6464,
      "step": 1440
    },
    {
      "epoch": 0.21526740364505528,
      "grad_norm": 0.23284317553043365,
      "learning_rate": 0.0001782208119178701,
      "loss": 0.9425,
      "step": 1441
    },
    {
      "epoch": 0.21541679115625934,
      "grad_norm": 0.12878303229808807,
      "learning_rate": 0.00017819152041938265,
      "loss": 0.7147,
      "step": 1442
    },
    {
      "epoch": 0.2155661786674634,
      "grad_norm": 0.13469137251377106,
      "learning_rate": 0.00017816221164714983,
      "loss": 0.8396,
      "step": 1443
    },
    {
      "epoch": 0.21571556617866747,
      "grad_norm": 0.13271968066692352,
      "learning_rate": 0.00017813288560764647,
      "loss": 0.8708,
      "step": 1444
    },
    {
      "epoch": 0.21586495368987152,
      "grad_norm": 0.13078995048999786,
      "learning_rate": 0.00017810354230735108,
      "loss": 0.5511,
      "step": 1445
    },
    {
      "epoch": 0.2160143412010756,
      "grad_norm": 0.179561585187912,
      "learning_rate": 0.00017807418175274612,
      "loss": 0.5318,
      "step": 1446
    },
    {
      "epoch": 0.21616372871227965,
      "grad_norm": 0.16479915380477905,
      "learning_rate": 0.00017804480395031778,
      "loss": 0.9253,
      "step": 1447
    },
    {
      "epoch": 0.2163131162234837,
      "grad_norm": 0.15908388793468475,
      "learning_rate": 0.00017801540890655609,
      "loss": 0.7711,
      "step": 1448
    },
    {
      "epoch": 0.21646250373468778,
      "grad_norm": 0.16113927960395813,
      "learning_rate": 0.00017798599662795483,
      "loss": 0.643,
      "step": 1449
    },
    {
      "epoch": 0.21661189124589184,
      "grad_norm": 0.12595628201961517,
      "learning_rate": 0.00017795656712101172,
      "loss": 0.6977,
      "step": 1450
    },
    {
      "epoch": 0.21676127875709592,
      "grad_norm": 0.15012586116790771,
      "learning_rate": 0.00017792712039222815,
      "loss": 0.8391,
      "step": 1451
    },
    {
      "epoch": 0.21691066626829997,
      "grad_norm": 0.13315634429454803,
      "learning_rate": 0.00017789765644810935,
      "loss": 0.6284,
      "step": 1452
    },
    {
      "epoch": 0.21706005377950402,
      "grad_norm": 0.20887716114521027,
      "learning_rate": 0.00017786817529516445,
      "loss": 0.6472,
      "step": 1453
    },
    {
      "epoch": 0.2172094412907081,
      "grad_norm": 0.12334492802619934,
      "learning_rate": 0.00017783867693990624,
      "loss": 0.8016,
      "step": 1454
    },
    {
      "epoch": 0.21735882880191215,
      "grad_norm": 0.14215132594108582,
      "learning_rate": 0.0001778091613888514,
      "loss": 0.6711,
      "step": 1455
    },
    {
      "epoch": 0.21750821631311623,
      "grad_norm": 0.6180028319358826,
      "learning_rate": 0.0001777796286485204,
      "loss": 1.2348,
      "step": 1456
    },
    {
      "epoch": 0.21765760382432028,
      "grad_norm": 0.15732866525650024,
      "learning_rate": 0.00017775007872543745,
      "loss": 0.7951,
      "step": 1457
    },
    {
      "epoch": 0.21780699133552436,
      "grad_norm": 0.07052062451839447,
      "learning_rate": 0.0001777205116261306,
      "loss": 0.3056,
      "step": 1458
    },
    {
      "epoch": 0.21795637884672842,
      "grad_norm": 0.1404731720685959,
      "learning_rate": 0.00017769092735713172,
      "loss": 0.8616,
      "step": 1459
    },
    {
      "epoch": 0.21810576635793247,
      "grad_norm": 0.24289721250534058,
      "learning_rate": 0.0001776613259249764,
      "loss": 1.058,
      "step": 1460
    },
    {
      "epoch": 0.21825515386913655,
      "grad_norm": 0.3455839157104492,
      "learning_rate": 0.0001776317073362041,
      "loss": 0.9287,
      "step": 1461
    },
    {
      "epoch": 0.2184045413803406,
      "grad_norm": 0.181719109416008,
      "learning_rate": 0.00017760207159735805,
      "loss": 0.851,
      "step": 1462
    },
    {
      "epoch": 0.21855392889154468,
      "grad_norm": 0.29283466935157776,
      "learning_rate": 0.0001775724187149852,
      "loss": 0.9268,
      "step": 1463
    },
    {
      "epoch": 0.21870331640274873,
      "grad_norm": 0.12602414190769196,
      "learning_rate": 0.00017754274869563637,
      "loss": 0.8347,
      "step": 1464
    },
    {
      "epoch": 0.21885270391395278,
      "grad_norm": 0.13600635528564453,
      "learning_rate": 0.0001775130615458662,
      "loss": 0.9986,
      "step": 1465
    },
    {
      "epoch": 0.21900209142515686,
      "grad_norm": 0.15044277906417847,
      "learning_rate": 0.00017748335727223294,
      "loss": 0.3543,
      "step": 1466
    },
    {
      "epoch": 0.21915147893636092,
      "grad_norm": 0.1423201858997345,
      "learning_rate": 0.00017745363588129878,
      "loss": 0.9459,
      "step": 1467
    },
    {
      "epoch": 0.219300866447565,
      "grad_norm": 0.13680531084537506,
      "learning_rate": 0.00017742389737962966,
      "loss": 0.5642,
      "step": 1468
    },
    {
      "epoch": 0.21945025395876905,
      "grad_norm": 0.13396508991718292,
      "learning_rate": 0.00017739414177379528,
      "loss": 0.6551,
      "step": 1469
    },
    {
      "epoch": 0.2195996414699731,
      "grad_norm": 0.11900012195110321,
      "learning_rate": 0.0001773643690703691,
      "loss": 0.6664,
      "step": 1470
    },
    {
      "epoch": 0.21974902898117718,
      "grad_norm": 0.12804585695266724,
      "learning_rate": 0.00017733457927592846,
      "loss": 0.6117,
      "step": 1471
    },
    {
      "epoch": 0.21989841649238123,
      "grad_norm": 0.19028812646865845,
      "learning_rate": 0.00017730477239705428,
      "loss": 0.5632,
      "step": 1472
    },
    {
      "epoch": 0.2200478040035853,
      "grad_norm": 0.15631790459156036,
      "learning_rate": 0.00017727494844033145,
      "loss": 1.0133,
      "step": 1473
    },
    {
      "epoch": 0.22019719151478936,
      "grad_norm": 0.14937928318977356,
      "learning_rate": 0.00017724510741234858,
      "loss": 0.9413,
      "step": 1474
    },
    {
      "epoch": 0.22034657902599342,
      "grad_norm": 0.12716789543628693,
      "learning_rate": 0.00017721524931969796,
      "loss": 0.5547,
      "step": 1475
    },
    {
      "epoch": 0.2204959665371975,
      "grad_norm": 0.27234596014022827,
      "learning_rate": 0.0001771853741689757,
      "loss": 1.0554,
      "step": 1476
    },
    {
      "epoch": 0.22064535404840155,
      "grad_norm": 0.1304127275943756,
      "learning_rate": 0.00017715548196678177,
      "loss": 0.7407,
      "step": 1477
    },
    {
      "epoch": 0.22079474155960563,
      "grad_norm": 0.14339019358158112,
      "learning_rate": 0.0001771255727197198,
      "loss": 0.9733,
      "step": 1478
    },
    {
      "epoch": 0.22094412907080968,
      "grad_norm": 0.18441441655158997,
      "learning_rate": 0.0001770956464343972,
      "loss": 0.9436,
      "step": 1479
    },
    {
      "epoch": 0.22109351658201373,
      "grad_norm": 0.13946813344955444,
      "learning_rate": 0.00017706570311742516,
      "loss": 0.6699,
      "step": 1480
    },
    {
      "epoch": 0.2212429040932178,
      "grad_norm": 0.20996683835983276,
      "learning_rate": 0.00017703574277541865,
      "loss": 1.0231,
      "step": 1481
    },
    {
      "epoch": 0.22139229160442186,
      "grad_norm": 0.15931503474712372,
      "learning_rate": 0.0001770057654149964,
      "loss": 0.7794,
      "step": 1482
    },
    {
      "epoch": 0.22154167911562594,
      "grad_norm": 0.14282017946243286,
      "learning_rate": 0.00017697577104278084,
      "loss": 0.6515,
      "step": 1483
    },
    {
      "epoch": 0.22169106662683,
      "grad_norm": 0.2071201503276825,
      "learning_rate": 0.00017694575966539823,
      "loss": 0.8382,
      "step": 1484
    },
    {
      "epoch": 0.22184045413803405,
      "grad_norm": 0.16338187456130981,
      "learning_rate": 0.00017691573128947853,
      "loss": 0.9846,
      "step": 1485
    },
    {
      "epoch": 0.22198984164923813,
      "grad_norm": 0.13437145948410034,
      "learning_rate": 0.00017688568592165552,
      "loss": 0.8371,
      "step": 1486
    },
    {
      "epoch": 0.22213922916044218,
      "grad_norm": 0.10946273803710938,
      "learning_rate": 0.0001768556235685667,
      "loss": 0.6741,
      "step": 1487
    },
    {
      "epoch": 0.22228861667164626,
      "grad_norm": 0.19846905767917633,
      "learning_rate": 0.00017682554423685329,
      "loss": 0.6215,
      "step": 1488
    },
    {
      "epoch": 0.2224380041828503,
      "grad_norm": 0.12315358966588974,
      "learning_rate": 0.0001767954479331603,
      "loss": 0.6331,
      "step": 1489
    },
    {
      "epoch": 0.22258739169405437,
      "grad_norm": 0.30643758177757263,
      "learning_rate": 0.0001767653346641365,
      "loss": 0.8563,
      "step": 1490
    },
    {
      "epoch": 0.22273677920525844,
      "grad_norm": 0.1250000298023224,
      "learning_rate": 0.00017673520443643436,
      "loss": 0.8405,
      "step": 1491
    },
    {
      "epoch": 0.2228861667164625,
      "grad_norm": 0.12933503091335297,
      "learning_rate": 0.00017670505725671013,
      "loss": 0.7115,
      "step": 1492
    },
    {
      "epoch": 0.22303555422766658,
      "grad_norm": 0.2853139638900757,
      "learning_rate": 0.00017667489313162382,
      "loss": 0.8367,
      "step": 1493
    },
    {
      "epoch": 0.22318494173887063,
      "grad_norm": 0.13272625207901,
      "learning_rate": 0.00017664471206783915,
      "loss": 0.7379,
      "step": 1494
    },
    {
      "epoch": 0.22333432925007468,
      "grad_norm": 0.1108209565281868,
      "learning_rate": 0.0001766145140720236,
      "loss": 0.5442,
      "step": 1495
    },
    {
      "epoch": 0.22348371676127876,
      "grad_norm": 0.1489863097667694,
      "learning_rate": 0.00017658429915084835,
      "loss": 0.731,
      "step": 1496
    },
    {
      "epoch": 0.2236331042724828,
      "grad_norm": 0.146969735622406,
      "learning_rate": 0.0001765540673109884,
      "loss": 0.605,
      "step": 1497
    },
    {
      "epoch": 0.2237824917836869,
      "grad_norm": 0.2565854489803314,
      "learning_rate": 0.00017652381855912247,
      "loss": 1.1015,
      "step": 1498
    },
    {
      "epoch": 0.22393187929489095,
      "grad_norm": 0.1937399059534073,
      "learning_rate": 0.00017649355290193288,
      "loss": 1.0397,
      "step": 1499
    },
    {
      "epoch": 0.224081266806095,
      "grad_norm": 0.33174678683280945,
      "learning_rate": 0.0001764632703461059,
      "loss": 1.0074,
      "step": 1500
    },
    {
      "epoch": 0.22423065431729908,
      "grad_norm": 0.12607234716415405,
      "learning_rate": 0.00017643297089833134,
      "loss": 0.7882,
      "step": 1501
    },
    {
      "epoch": 0.22438004182850313,
      "grad_norm": 0.14097252488136292,
      "learning_rate": 0.00017640265456530293,
      "loss": 0.8961,
      "step": 1502
    },
    {
      "epoch": 0.2245294293397072,
      "grad_norm": 0.10994578897953033,
      "learning_rate": 0.00017637232135371794,
      "loss": 0.4649,
      "step": 1503
    },
    {
      "epoch": 0.22467881685091126,
      "grad_norm": 0.2514827847480774,
      "learning_rate": 0.0001763419712702775,
      "loss": 0.6585,
      "step": 1504
    },
    {
      "epoch": 0.2248282043621153,
      "grad_norm": 0.13599185645580292,
      "learning_rate": 0.0001763116043216864,
      "loss": 0.692,
      "step": 1505
    },
    {
      "epoch": 0.2249775918733194,
      "grad_norm": 0.12496299296617508,
      "learning_rate": 0.00017628122051465322,
      "loss": 0.802,
      "step": 1506
    },
    {
      "epoch": 0.22512697938452345,
      "grad_norm": 0.178312286734581,
      "learning_rate": 0.00017625081985589016,
      "loss": 0.7811,
      "step": 1507
    },
    {
      "epoch": 0.22527636689572753,
      "grad_norm": 0.12510164082050323,
      "learning_rate": 0.00017622040235211326,
      "loss": 0.4649,
      "step": 1508
    },
    {
      "epoch": 0.22542575440693158,
      "grad_norm": 0.14987626671791077,
      "learning_rate": 0.0001761899680100422,
      "loss": 0.4765,
      "step": 1509
    },
    {
      "epoch": 0.22557514191813566,
      "grad_norm": 0.554569661617279,
      "learning_rate": 0.00017615951683640045,
      "loss": 1.674,
      "step": 1510
    },
    {
      "epoch": 0.2257245294293397,
      "grad_norm": 0.16470187902450562,
      "learning_rate": 0.0001761290488379151,
      "loss": 0.5423,
      "step": 1511
    },
    {
      "epoch": 0.22587391694054376,
      "grad_norm": 0.13838253915309906,
      "learning_rate": 0.00017609856402131703,
      "loss": 0.5373,
      "step": 1512
    },
    {
      "epoch": 0.22602330445174784,
      "grad_norm": 0.13024069368839264,
      "learning_rate": 0.00017606806239334083,
      "loss": 0.8255,
      "step": 1513
    },
    {
      "epoch": 0.2261726919629519,
      "grad_norm": 0.13221733272075653,
      "learning_rate": 0.00017603754396072483,
      "loss": 0.7189,
      "step": 1514
    },
    {
      "epoch": 0.22632207947415597,
      "grad_norm": 0.15409418940544128,
      "learning_rate": 0.00017600700873021097,
      "loss": 1.0796,
      "step": 1515
    },
    {
      "epoch": 0.22647146698536003,
      "grad_norm": 0.16743522882461548,
      "learning_rate": 0.000175976456708545,
      "loss": 1.0363,
      "step": 1516
    },
    {
      "epoch": 0.22662085449656408,
      "grad_norm": 0.19694027304649353,
      "learning_rate": 0.0001759458879024763,
      "loss": 0.8512,
      "step": 1517
    },
    {
      "epoch": 0.22677024200776816,
      "grad_norm": 0.43795064091682434,
      "learning_rate": 0.0001759153023187581,
      "loss": 1.4031,
      "step": 1518
    },
    {
      "epoch": 0.2269196295189722,
      "grad_norm": 0.22090782225131989,
      "learning_rate": 0.00017588469996414715,
      "loss": 0.6461,
      "step": 1519
    },
    {
      "epoch": 0.2270690170301763,
      "grad_norm": 0.13713456690311432,
      "learning_rate": 0.00017585408084540405,
      "loss": 0.8236,
      "step": 1520
    },
    {
      "epoch": 0.22721840454138034,
      "grad_norm": 0.1438591182231903,
      "learning_rate": 0.000175823444969293,
      "loss": 1.1679,
      "step": 1521
    },
    {
      "epoch": 0.2273677920525844,
      "grad_norm": 0.12054278701543808,
      "learning_rate": 0.00017579279234258198,
      "loss": 0.6954,
      "step": 1522
    },
    {
      "epoch": 0.22751717956378847,
      "grad_norm": 0.12184090167284012,
      "learning_rate": 0.0001757621229720426,
      "loss": 0.7207,
      "step": 1523
    },
    {
      "epoch": 0.22766656707499253,
      "grad_norm": 0.36808204650878906,
      "learning_rate": 0.00017573143686445034,
      "loss": 0.8402,
      "step": 1524
    },
    {
      "epoch": 0.2278159545861966,
      "grad_norm": 0.20362381637096405,
      "learning_rate": 0.00017570073402658408,
      "loss": 0.7507,
      "step": 1525
    },
    {
      "epoch": 0.22796534209740066,
      "grad_norm": 0.32647380232810974,
      "learning_rate": 0.00017567001446522665,
      "loss": 0.6814,
      "step": 1526
    },
    {
      "epoch": 0.2281147296086047,
      "grad_norm": 0.18920038640499115,
      "learning_rate": 0.00017563927818716447,
      "loss": 0.503,
      "step": 1527
    },
    {
      "epoch": 0.2282641171198088,
      "grad_norm": 0.18404003977775574,
      "learning_rate": 0.0001756085251991877,
      "loss": 0.5798,
      "step": 1528
    },
    {
      "epoch": 0.22841350463101284,
      "grad_norm": 0.1709233671426773,
      "learning_rate": 0.0001755777555080901,
      "loss": 1.3392,
      "step": 1529
    },
    {
      "epoch": 0.22856289214221692,
      "grad_norm": 0.13622832298278809,
      "learning_rate": 0.00017554696912066924,
      "loss": 0.7546,
      "step": 1530
    },
    {
      "epoch": 0.22871227965342097,
      "grad_norm": 0.14805671572685242,
      "learning_rate": 0.00017551616604372629,
      "loss": 0.8027,
      "step": 1531
    },
    {
      "epoch": 0.22886166716462503,
      "grad_norm": 0.2947498559951782,
      "learning_rate": 0.00017548534628406616,
      "loss": 0.9071,
      "step": 1532
    },
    {
      "epoch": 0.2290110546758291,
      "grad_norm": 0.1275065690279007,
      "learning_rate": 0.0001754545098484974,
      "loss": 0.6312,
      "step": 1533
    },
    {
      "epoch": 0.22916044218703316,
      "grad_norm": 0.11439994722604752,
      "learning_rate": 0.00017542365674383227,
      "loss": 0.4962,
      "step": 1534
    },
    {
      "epoch": 0.22930982969823724,
      "grad_norm": 0.14503420889377594,
      "learning_rate": 0.00017539278697688672,
      "loss": 0.6967,
      "step": 1535
    },
    {
      "epoch": 0.2294592172094413,
      "grad_norm": 0.13170544803142548,
      "learning_rate": 0.00017536190055448037,
      "loss": 0.9048,
      "step": 1536
    },
    {
      "epoch": 0.22960860472064534,
      "grad_norm": 0.22682088613510132,
      "learning_rate": 0.0001753309974834365,
      "loss": 0.7506,
      "step": 1537
    },
    {
      "epoch": 0.22975799223184942,
      "grad_norm": 0.16265404224395752,
      "learning_rate": 0.00017530007777058213,
      "loss": 0.6615,
      "step": 1538
    },
    {
      "epoch": 0.22990737974305347,
      "grad_norm": 0.1368705928325653,
      "learning_rate": 0.0001752691414227479,
      "loss": 0.8368,
      "step": 1539
    },
    {
      "epoch": 0.23005676725425755,
      "grad_norm": 0.18166819214820862,
      "learning_rate": 0.0001752381884467681,
      "loss": 0.6836,
      "step": 1540
    },
    {
      "epoch": 0.2302061547654616,
      "grad_norm": 0.11365760862827301,
      "learning_rate": 0.00017520721884948075,
      "loss": 0.7073,
      "step": 1541
    },
    {
      "epoch": 0.23035554227666566,
      "grad_norm": 0.23853546380996704,
      "learning_rate": 0.00017517623263772758,
      "loss": 0.9235,
      "step": 1542
    },
    {
      "epoch": 0.23050492978786974,
      "grad_norm": 0.29412710666656494,
      "learning_rate": 0.00017514522981835383,
      "loss": 1.1033,
      "step": 1543
    },
    {
      "epoch": 0.2306543172990738,
      "grad_norm": 0.13339991867542267,
      "learning_rate": 0.00017511421039820863,
      "loss": 0.7038,
      "step": 1544
    },
    {
      "epoch": 0.23080370481027787,
      "grad_norm": 0.12489107251167297,
      "learning_rate": 0.00017508317438414458,
      "loss": 0.5625,
      "step": 1545
    },
    {
      "epoch": 0.23095309232148192,
      "grad_norm": 0.1896982640028,
      "learning_rate": 0.00017505212178301805,
      "loss": 0.7489,
      "step": 1546
    },
    {
      "epoch": 0.23110247983268598,
      "grad_norm": 0.13012000918388367,
      "learning_rate": 0.00017502105260168907,
      "loss": 0.615,
      "step": 1547
    },
    {
      "epoch": 0.23125186734389006,
      "grad_norm": 0.2946460247039795,
      "learning_rate": 0.00017498996684702132,
      "loss": 0.8367,
      "step": 1548
    },
    {
      "epoch": 0.2314012548550941,
      "grad_norm": 0.15457066893577576,
      "learning_rate": 0.00017495886452588205,
      "loss": 0.5542,
      "step": 1549
    },
    {
      "epoch": 0.2315506423662982,
      "grad_norm": 0.12016140669584274,
      "learning_rate": 0.00017492774564514235,
      "loss": 0.908,
      "step": 1550
    },
    {
      "epoch": 0.23170002987750224,
      "grad_norm": 0.17320853471755981,
      "learning_rate": 0.00017489661021167686,
      "loss": 0.816,
      "step": 1551
    },
    {
      "epoch": 0.2318494173887063,
      "grad_norm": 0.12414354085922241,
      "learning_rate": 0.00017486545823236385,
      "loss": 0.7893,
      "step": 1552
    },
    {
      "epoch": 0.23199880489991037,
      "grad_norm": 0.1295233517885208,
      "learning_rate": 0.00017483428971408534,
      "loss": 0.9902,
      "step": 1553
    },
    {
      "epoch": 0.23214819241111442,
      "grad_norm": 0.1532648801803589,
      "learning_rate": 0.00017480310466372686,
      "loss": 0.6839,
      "step": 1554
    },
    {
      "epoch": 0.2322975799223185,
      "grad_norm": 0.15763483941555023,
      "learning_rate": 0.00017477190308817778,
      "loss": 0.73,
      "step": 1555
    },
    {
      "epoch": 0.23244696743352256,
      "grad_norm": 0.22746244072914124,
      "learning_rate": 0.00017474068499433098,
      "loss": 0.9003,
      "step": 1556
    },
    {
      "epoch": 0.2325963549447266,
      "grad_norm": 0.11888179183006287,
      "learning_rate": 0.00017470945038908304,
      "loss": 0.4965,
      "step": 1557
    },
    {
      "epoch": 0.2327457424559307,
      "grad_norm": 0.11085602641105652,
      "learning_rate": 0.00017467819927933416,
      "loss": 0.4841,
      "step": 1558
    },
    {
      "epoch": 0.23289512996713474,
      "grad_norm": 0.145162433385849,
      "learning_rate": 0.0001746469316719882,
      "loss": 0.9839,
      "step": 1559
    },
    {
      "epoch": 0.23304451747833882,
      "grad_norm": 0.12735643982887268,
      "learning_rate": 0.00017461564757395272,
      "loss": 0.601,
      "step": 1560
    },
    {
      "epoch": 0.23319390498954287,
      "grad_norm": 0.1690744012594223,
      "learning_rate": 0.00017458434699213883,
      "loss": 0.909,
      "step": 1561
    },
    {
      "epoch": 0.23334329250074692,
      "grad_norm": 0.13597418367862701,
      "learning_rate": 0.00017455302993346134,
      "loss": 0.8371,
      "step": 1562
    },
    {
      "epoch": 0.233492680011951,
      "grad_norm": 0.1151067465543747,
      "learning_rate": 0.00017452169640483865,
      "loss": 0.7987,
      "step": 1563
    },
    {
      "epoch": 0.23364206752315506,
      "grad_norm": 0.1729961335659027,
      "learning_rate": 0.00017449034641319288,
      "loss": 0.434,
      "step": 1564
    },
    {
      "epoch": 0.23379145503435914,
      "grad_norm": 0.1317516565322876,
      "learning_rate": 0.00017445897996544972,
      "loss": 0.9376,
      "step": 1565
    },
    {
      "epoch": 0.2339408425455632,
      "grad_norm": 0.3169648349285126,
      "learning_rate": 0.00017442759706853855,
      "loss": 0.7284,
      "step": 1566
    },
    {
      "epoch": 0.23409023005676727,
      "grad_norm": 0.15828759968280792,
      "learning_rate": 0.0001743961977293923,
      "loss": 0.8469,
      "step": 1567
    },
    {
      "epoch": 0.23423961756797132,
      "grad_norm": 0.15548141300678253,
      "learning_rate": 0.00017436478195494756,
      "loss": 0.6545,
      "step": 1568
    },
    {
      "epoch": 0.23438900507917537,
      "grad_norm": 0.11150869727134705,
      "learning_rate": 0.00017433334975214463,
      "loss": 0.6005,
      "step": 1569
    },
    {
      "epoch": 0.23453839259037945,
      "grad_norm": 0.1405535489320755,
      "learning_rate": 0.00017430190112792737,
      "loss": 0.825,
      "step": 1570
    },
    {
      "epoch": 0.2346877801015835,
      "grad_norm": 0.1619478166103363,
      "learning_rate": 0.00017427043608924325,
      "loss": 0.5656,
      "step": 1571
    },
    {
      "epoch": 0.23483716761278758,
      "grad_norm": 0.1910037249326706,
      "learning_rate": 0.00017423895464304342,
      "loss": 0.9477,
      "step": 1572
    },
    {
      "epoch": 0.23498655512399164,
      "grad_norm": 0.13986697793006897,
      "learning_rate": 0.00017420745679628264,
      "loss": 0.5093,
      "step": 1573
    },
    {
      "epoch": 0.2351359426351957,
      "grad_norm": 0.13165700435638428,
      "learning_rate": 0.00017417594255591927,
      "loss": 0.7117,
      "step": 1574
    },
    {
      "epoch": 0.23528533014639977,
      "grad_norm": 0.14331944286823273,
      "learning_rate": 0.00017414441192891529,
      "loss": 0.8661,
      "step": 1575
    },
    {
      "epoch": 0.23543471765760382,
      "grad_norm": 0.12297346442937851,
      "learning_rate": 0.00017411286492223632,
      "loss": 0.9035,
      "step": 1576
    },
    {
      "epoch": 0.2355841051688079,
      "grad_norm": 0.11895941197872162,
      "learning_rate": 0.00017408130154285162,
      "loss": 0.6656,
      "step": 1577
    },
    {
      "epoch": 0.23573349268001195,
      "grad_norm": 0.1288233995437622,
      "learning_rate": 0.000174049721797734,
      "loss": 0.8691,
      "step": 1578
    },
    {
      "epoch": 0.235882880191216,
      "grad_norm": 0.10877169668674469,
      "learning_rate": 0.00017401812569385998,
      "loss": 0.6643,
      "step": 1579
    },
    {
      "epoch": 0.23603226770242008,
      "grad_norm": 0.11538077890872955,
      "learning_rate": 0.00017398651323820958,
      "loss": 0.7094,
      "step": 1580
    },
    {
      "epoch": 0.23618165521362414,
      "grad_norm": 0.12808218598365784,
      "learning_rate": 0.00017395488443776652,
      "loss": 0.5041,
      "step": 1581
    },
    {
      "epoch": 0.23633104272482822,
      "grad_norm": 0.26270732283592224,
      "learning_rate": 0.00017392323929951812,
      "loss": 0.8308,
      "step": 1582
    },
    {
      "epoch": 0.23648043023603227,
      "grad_norm": 0.14876054227352142,
      "learning_rate": 0.0001738915778304553,
      "loss": 0.5326,
      "step": 1583
    },
    {
      "epoch": 0.23662981774723632,
      "grad_norm": 0.11852624267339706,
      "learning_rate": 0.0001738599000375725,
      "loss": 0.6705,
      "step": 1584
    },
    {
      "epoch": 0.2367792052584404,
      "grad_norm": 0.21757078170776367,
      "learning_rate": 0.00017382820592786791,
      "loss": 0.8034,
      "step": 1585
    },
    {
      "epoch": 0.23692859276964445,
      "grad_norm": 0.20093433558940887,
      "learning_rate": 0.00017379649550834327,
      "loss": 0.5919,
      "step": 1586
    },
    {
      "epoch": 0.23707798028084853,
      "grad_norm": 0.1466897428035736,
      "learning_rate": 0.0001737647687860039,
      "loss": 0.8727,
      "step": 1587
    },
    {
      "epoch": 0.23722736779205258,
      "grad_norm": 0.13090848922729492,
      "learning_rate": 0.00017373302576785874,
      "loss": 0.8407,
      "step": 1588
    },
    {
      "epoch": 0.23737675530325664,
      "grad_norm": 0.265153706073761,
      "learning_rate": 0.0001737012664609203,
      "loss": 0.6411,
      "step": 1589
    },
    {
      "epoch": 0.23752614281446072,
      "grad_norm": 0.1178729236125946,
      "learning_rate": 0.00017366949087220472,
      "loss": 0.7404,
      "step": 1590
    },
    {
      "epoch": 0.23767553032566477,
      "grad_norm": 0.1343647837638855,
      "learning_rate": 0.00017363769900873173,
      "loss": 1.0259,
      "step": 1591
    },
    {
      "epoch": 0.23782491783686885,
      "grad_norm": 0.2027769237756729,
      "learning_rate": 0.0001736058908775247,
      "loss": 0.5823,
      "step": 1592
    },
    {
      "epoch": 0.2379743053480729,
      "grad_norm": 0.12886838614940643,
      "learning_rate": 0.00017357406648561052,
      "loss": 0.7036,
      "step": 1593
    },
    {
      "epoch": 0.23812369285927695,
      "grad_norm": 0.11439285427331924,
      "learning_rate": 0.0001735422258400197,
      "loss": 0.6114,
      "step": 1594
    },
    {
      "epoch": 0.23827308037048103,
      "grad_norm": 0.12173706293106079,
      "learning_rate": 0.00017351036894778634,
      "loss": 0.5466,
      "step": 1595
    },
    {
      "epoch": 0.23842246788168509,
      "grad_norm": 0.15928342938423157,
      "learning_rate": 0.0001734784958159481,
      "loss": 1.0324,
      "step": 1596
    },
    {
      "epoch": 0.23857185539288916,
      "grad_norm": 0.24861174821853638,
      "learning_rate": 0.00017344660645154635,
      "loss": 0.954,
      "step": 1597
    },
    {
      "epoch": 0.23872124290409322,
      "grad_norm": 0.24405913054943085,
      "learning_rate": 0.00017341470086162586,
      "loss": 0.984,
      "step": 1598
    },
    {
      "epoch": 0.23887063041529727,
      "grad_norm": 0.12469593435525894,
      "learning_rate": 0.00017338277905323515,
      "loss": 0.8056,
      "step": 1599
    },
    {
      "epoch": 0.23902001792650135,
      "grad_norm": 0.13238497078418732,
      "learning_rate": 0.0001733508410334262,
      "loss": 0.9383,
      "step": 1600
    },
    {
      "epoch": 0.2391694054377054,
      "grad_norm": 0.1149744838476181,
      "learning_rate": 0.00017331888680925466,
      "loss": 0.7664,
      "step": 1601
    },
    {
      "epoch": 0.23931879294890948,
      "grad_norm": 0.25224781036376953,
      "learning_rate": 0.0001732869163877797,
      "loss": 0.8615,
      "step": 1602
    },
    {
      "epoch": 0.23946818046011353,
      "grad_norm": 0.12752684950828552,
      "learning_rate": 0.00017325492977606413,
      "loss": 0.6047,
      "step": 1603
    },
    {
      "epoch": 0.23961756797131759,
      "grad_norm": 0.12648506462574005,
      "learning_rate": 0.00017322292698117425,
      "loss": 0.6071,
      "step": 1604
    },
    {
      "epoch": 0.23976695548252167,
      "grad_norm": 0.13702405989170074,
      "learning_rate": 0.00017319090801018003,
      "loss": 0.878,
      "step": 1605
    },
    {
      "epoch": 0.23991634299372572,
      "grad_norm": 0.15288682281970978,
      "learning_rate": 0.00017315887287015492,
      "loss": 0.807,
      "step": 1606
    },
    {
      "epoch": 0.2400657305049298,
      "grad_norm": 0.2027759850025177,
      "learning_rate": 0.00017312682156817602,
      "loss": 0.5606,
      "step": 1607
    },
    {
      "epoch": 0.24021511801613385,
      "grad_norm": 0.21917293965816498,
      "learning_rate": 0.000173094754111324,
      "loss": 0.9878,
      "step": 1608
    },
    {
      "epoch": 0.2403645055273379,
      "grad_norm": 0.24028894305229187,
      "learning_rate": 0.000173062670506683,
      "loss": 1.0361,
      "step": 1609
    },
    {
      "epoch": 0.24051389303854198,
      "grad_norm": 0.3546265959739685,
      "learning_rate": 0.00017303057076134085,
      "loss": 1.2186,
      "step": 1610
    },
    {
      "epoch": 0.24066328054974603,
      "grad_norm": 0.12783119082450867,
      "learning_rate": 0.0001729984548823889,
      "loss": 0.8723,
      "step": 1611
    },
    {
      "epoch": 0.2408126680609501,
      "grad_norm": 0.11363591253757477,
      "learning_rate": 0.000172966322876922,
      "loss": 0.572,
      "step": 1612
    },
    {
      "epoch": 0.24096205557215417,
      "grad_norm": 0.15105083584785461,
      "learning_rate": 0.00017293417475203863,
      "loss": 0.7239,
      "step": 1613
    },
    {
      "epoch": 0.24111144308335822,
      "grad_norm": 0.2824059724807739,
      "learning_rate": 0.00017290201051484085,
      "loss": 1.0547,
      "step": 1614
    },
    {
      "epoch": 0.2412608305945623,
      "grad_norm": 0.21242327988147736,
      "learning_rate": 0.00017286983017243424,
      "loss": 0.6498,
      "step": 1615
    },
    {
      "epoch": 0.24141021810576635,
      "grad_norm": 0.1261444389820099,
      "learning_rate": 0.00017283763373192798,
      "loss": 0.8672,
      "step": 1616
    },
    {
      "epoch": 0.24155960561697043,
      "grad_norm": 0.26087701320648193,
      "learning_rate": 0.00017280542120043472,
      "loss": 0.6234,
      "step": 1617
    },
    {
      "epoch": 0.24170899312817448,
      "grad_norm": 0.1511380523443222,
      "learning_rate": 0.00017277319258507073,
      "loss": 0.6773,
      "step": 1618
    },
    {
      "epoch": 0.24185838063937856,
      "grad_norm": 0.14391735196113586,
      "learning_rate": 0.00017274094789295586,
      "loss": 0.8325,
      "step": 1619
    },
    {
      "epoch": 0.24200776815058261,
      "grad_norm": 0.17719805240631104,
      "learning_rate": 0.0001727086871312134,
      "loss": 0.7249,
      "step": 1620
    },
    {
      "epoch": 0.24215715566178667,
      "grad_norm": 0.1089552789926529,
      "learning_rate": 0.00017267641030697034,
      "loss": 0.6459,
      "step": 1621
    },
    {
      "epoch": 0.24230654317299075,
      "grad_norm": 0.15426376461982727,
      "learning_rate": 0.00017264411742735707,
      "loss": 0.8286,
      "step": 1622
    },
    {
      "epoch": 0.2424559306841948,
      "grad_norm": 0.1446702480316162,
      "learning_rate": 0.00017261180849950766,
      "loss": 0.5768,
      "step": 1623
    },
    {
      "epoch": 0.24260531819539888,
      "grad_norm": 0.23412354290485382,
      "learning_rate": 0.00017257948353055963,
      "loss": 0.5719,
      "step": 1624
    },
    {
      "epoch": 0.24275470570660293,
      "grad_norm": 0.1535494327545166,
      "learning_rate": 0.00017254714252765412,
      "loss": 0.7175,
      "step": 1625
    },
    {
      "epoch": 0.24290409321780698,
      "grad_norm": 0.1816725730895996,
      "learning_rate": 0.0001725147854979357,
      "loss": 0.8062,
      "step": 1626
    },
    {
      "epoch": 0.24305348072901106,
      "grad_norm": 0.2537088692188263,
      "learning_rate": 0.00017248241244855264,
      "loss": 0.8809,
      "step": 1627
    },
    {
      "epoch": 0.24320286824021511,
      "grad_norm": 0.12580998241901398,
      "learning_rate": 0.00017245002338665656,
      "loss": 0.5737,
      "step": 1628
    },
    {
      "epoch": 0.2433522557514192,
      "grad_norm": 0.18232382833957672,
      "learning_rate": 0.0001724176183194028,
      "loss": 0.5088,
      "step": 1629
    },
    {
      "epoch": 0.24350164326262325,
      "grad_norm": 0.14372123777866364,
      "learning_rate": 0.00017238519725395007,
      "loss": 0.8043,
      "step": 1630
    },
    {
      "epoch": 0.2436510307738273,
      "grad_norm": 0.1633867621421814,
      "learning_rate": 0.00017235276019746077,
      "loss": 0.8469,
      "step": 1631
    },
    {
      "epoch": 0.24380041828503138,
      "grad_norm": 0.23584675788879395,
      "learning_rate": 0.00017232030715710076,
      "loss": 0.9823,
      "step": 1632
    },
    {
      "epoch": 0.24394980579623543,
      "grad_norm": 0.13388711214065552,
      "learning_rate": 0.00017228783814003936,
      "loss": 0.7298,
      "step": 1633
    },
    {
      "epoch": 0.2440991933074395,
      "grad_norm": 0.14213724434375763,
      "learning_rate": 0.00017225535315344955,
      "loss": 0.7396,
      "step": 1634
    },
    {
      "epoch": 0.24424858081864356,
      "grad_norm": 0.15166963636875153,
      "learning_rate": 0.0001722228522045078,
      "loss": 0.735,
      "step": 1635
    },
    {
      "epoch": 0.24439796832984761,
      "grad_norm": 0.1204567551612854,
      "learning_rate": 0.00017219033530039397,
      "loss": 0.6299,
      "step": 1636
    },
    {
      "epoch": 0.2445473558410517,
      "grad_norm": 0.11346624791622162,
      "learning_rate": 0.0001721578024482917,
      "loss": 0.6771,
      "step": 1637
    },
    {
      "epoch": 0.24469674335225575,
      "grad_norm": 0.16759368777275085,
      "learning_rate": 0.00017212525365538792,
      "loss": 0.7302,
      "step": 1638
    },
    {
      "epoch": 0.24484613086345983,
      "grad_norm": 0.12218065559864044,
      "learning_rate": 0.0001720926889288732,
      "loss": 0.7969,
      "step": 1639
    },
    {
      "epoch": 0.24499551837466388,
      "grad_norm": 0.11079906672239304,
      "learning_rate": 0.00017206010827594163,
      "loss": 0.5828,
      "step": 1640
    },
    {
      "epoch": 0.24514490588586793,
      "grad_norm": 0.1696266084909439,
      "learning_rate": 0.00017202751170379075,
      "loss": 1.0499,
      "step": 1641
    },
    {
      "epoch": 0.245294293397072,
      "grad_norm": 0.2520163357257843,
      "learning_rate": 0.0001719948992196217,
      "loss": 1.0662,
      "step": 1642
    },
    {
      "epoch": 0.24544368090827606,
      "grad_norm": 0.1606278270483017,
      "learning_rate": 0.00017196227083063906,
      "loss": 0.6013,
      "step": 1643
    },
    {
      "epoch": 0.24559306841948014,
      "grad_norm": 0.11815240234136581,
      "learning_rate": 0.00017192962654405096,
      "loss": 0.5613,
      "step": 1644
    },
    {
      "epoch": 0.2457424559306842,
      "grad_norm": 0.1555798351764679,
      "learning_rate": 0.00017189696636706904,
      "loss": 0.9341,
      "step": 1645
    },
    {
      "epoch": 0.24589184344188825,
      "grad_norm": 0.1412414312362671,
      "learning_rate": 0.00017186429030690848,
      "loss": 0.8947,
      "step": 1646
    },
    {
      "epoch": 0.24604123095309233,
      "grad_norm": 0.1420399695634842,
      "learning_rate": 0.00017183159837078792,
      "loss": 0.7541,
      "step": 1647
    },
    {
      "epoch": 0.24619061846429638,
      "grad_norm": 0.16363643109798431,
      "learning_rate": 0.00017179889056592954,
      "loss": 0.7413,
      "step": 1648
    },
    {
      "epoch": 0.24634000597550046,
      "grad_norm": 0.27612796425819397,
      "learning_rate": 0.000171766166899559,
      "loss": 1.1101,
      "step": 1649
    },
    {
      "epoch": 0.2464893934867045,
      "grad_norm": 0.169744610786438,
      "learning_rate": 0.00017173342737890544,
      "loss": 0.7638,
      "step": 1650
    },
    {
      "epoch": 0.24663878099790856,
      "grad_norm": 0.2400025874376297,
      "learning_rate": 0.0001717006720112016,
      "loss": 0.6642,
      "step": 1651
    },
    {
      "epoch": 0.24678816850911264,
      "grad_norm": 0.12115002423524857,
      "learning_rate": 0.00017166790080368357,
      "loss": 0.5818,
      "step": 1652
    },
    {
      "epoch": 0.2469375560203167,
      "grad_norm": 0.11767467111349106,
      "learning_rate": 0.00017163511376359116,
      "loss": 0.7801,
      "step": 1653
    },
    {
      "epoch": 0.24708694353152078,
      "grad_norm": 0.18294836580753326,
      "learning_rate": 0.00017160231089816748,
      "loss": 0.9749,
      "step": 1654
    },
    {
      "epoch": 0.24723633104272483,
      "grad_norm": 0.16762547194957733,
      "learning_rate": 0.00017156949221465916,
      "loss": 0.5872,
      "step": 1655
    },
    {
      "epoch": 0.24738571855392888,
      "grad_norm": 0.28492650389671326,
      "learning_rate": 0.00017153665772031643,
      "loss": 0.9859,
      "step": 1656
    },
    {
      "epoch": 0.24753510606513296,
      "grad_norm": 0.516017735004425,
      "learning_rate": 0.00017150380742239293,
      "loss": 1.2148,
      "step": 1657
    },
    {
      "epoch": 0.247684493576337,
      "grad_norm": 0.1820579469203949,
      "learning_rate": 0.0001714709413281458,
      "loss": 0.8255,
      "step": 1658
    },
    {
      "epoch": 0.2478338810875411,
      "grad_norm": 0.1763305366039276,
      "learning_rate": 0.0001714380594448357,
      "loss": 0.7998,
      "step": 1659
    },
    {
      "epoch": 0.24798326859874514,
      "grad_norm": 0.1392301470041275,
      "learning_rate": 0.00017140516177972676,
      "loss": 0.8738,
      "step": 1660
    },
    {
      "epoch": 0.2481326561099492,
      "grad_norm": 0.2311377078294754,
      "learning_rate": 0.00017137224834008657,
      "loss": 0.9255,
      "step": 1661
    },
    {
      "epoch": 0.24828204362115328,
      "grad_norm": 0.14632488787174225,
      "learning_rate": 0.00017133931913318625,
      "loss": 0.8224,
      "step": 1662
    },
    {
      "epoch": 0.24843143113235733,
      "grad_norm": 0.1478954702615738,
      "learning_rate": 0.00017130637416630035,
      "loss": 0.6986,
      "step": 1663
    },
    {
      "epoch": 0.2485808186435614,
      "grad_norm": 0.161132350564003,
      "learning_rate": 0.00017127341344670696,
      "loss": 0.7352,
      "step": 1664
    },
    {
      "epoch": 0.24873020615476546,
      "grad_norm": 0.13480094075202942,
      "learning_rate": 0.00017124043698168764,
      "loss": 0.7767,
      "step": 1665
    },
    {
      "epoch": 0.2488795936659695,
      "grad_norm": 0.25371015071868896,
      "learning_rate": 0.00017120744477852745,
      "loss": 1.0407,
      "step": 1666
    },
    {
      "epoch": 0.2490289811771736,
      "grad_norm": 0.13566429913043976,
      "learning_rate": 0.00017117443684451478,
      "loss": 0.7924,
      "step": 1667
    },
    {
      "epoch": 0.24917836868837764,
      "grad_norm": 0.15394295752048492,
      "learning_rate": 0.00017114141318694167,
      "loss": 0.6343,
      "step": 1668
    },
    {
      "epoch": 0.24932775619958172,
      "grad_norm": 0.15498654544353485,
      "learning_rate": 0.0001711083738131036,
      "loss": 0.5479,
      "step": 1669
    },
    {
      "epoch": 0.24947714371078578,
      "grad_norm": 0.11515049636363983,
      "learning_rate": 0.00017107531873029942,
      "loss": 0.7509,
      "step": 1670
    },
    {
      "epoch": 0.24962653122198983,
      "grad_norm": 0.12256692349910736,
      "learning_rate": 0.00017104224794583158,
      "loss": 0.4964,
      "step": 1671
    },
    {
      "epoch": 0.2497759187331939,
      "grad_norm": 0.5561460852622986,
      "learning_rate": 0.0001710091614670059,
      "loss": 1.3467,
      "step": 1672
    },
    {
      "epoch": 0.24992530624439796,
      "grad_norm": 0.1814536154270172,
      "learning_rate": 0.00017097605930113175,
      "loss": 0.7138,
      "step": 1673
    },
    {
      "epoch": 0.25007469375560204,
      "grad_norm": 0.19458015263080597,
      "learning_rate": 0.00017094294145552188,
      "loss": 0.8228,
      "step": 1674
    },
    {
      "epoch": 0.2502240812668061,
      "grad_norm": 0.24046920239925385,
      "learning_rate": 0.00017090980793749258,
      "loss": 0.6464,
      "step": 1675
    },
    {
      "epoch": 0.25037346877801014,
      "grad_norm": 0.1473151296377182,
      "learning_rate": 0.00017087665875436354,
      "loss": 0.4695,
      "step": 1676
    },
    {
      "epoch": 0.2505228562892142,
      "grad_norm": 0.1309395283460617,
      "learning_rate": 0.00017084349391345796,
      "loss": 0.5628,
      "step": 1677
    },
    {
      "epoch": 0.2506722438004183,
      "grad_norm": 0.1466517597436905,
      "learning_rate": 0.00017081031342210245,
      "loss": 0.8019,
      "step": 1678
    },
    {
      "epoch": 0.25082163131162233,
      "grad_norm": 0.12185301631689072,
      "learning_rate": 0.00017077711728762714,
      "loss": 0.7219,
      "step": 1679
    },
    {
      "epoch": 0.2509710188228264,
      "grad_norm": 0.17365945875644684,
      "learning_rate": 0.0001707439055173656,
      "loss": 0.7096,
      "step": 1680
    },
    {
      "epoch": 0.2511204063340305,
      "grad_norm": 0.1582822948694229,
      "learning_rate": 0.00017071067811865476,
      "loss": 0.7828,
      "step": 1681
    },
    {
      "epoch": 0.2512697938452345,
      "grad_norm": 0.1119188591837883,
      "learning_rate": 0.00017067743509883515,
      "loss": 0.612,
      "step": 1682
    },
    {
      "epoch": 0.2514191813564386,
      "grad_norm": 0.11489804089069366,
      "learning_rate": 0.00017064417646525066,
      "loss": 0.653,
      "step": 1683
    },
    {
      "epoch": 0.2515685688676427,
      "grad_norm": 0.11803659796714783,
      "learning_rate": 0.00017061090222524863,
      "loss": 0.7372,
      "step": 1684
    },
    {
      "epoch": 0.25171795637884675,
      "grad_norm": 0.1338784098625183,
      "learning_rate": 0.00017057761238617984,
      "loss": 0.5076,
      "step": 1685
    },
    {
      "epoch": 0.2518673438900508,
      "grad_norm": 0.148620143532753,
      "learning_rate": 0.00017054430695539864,
      "loss": 0.8958,
      "step": 1686
    },
    {
      "epoch": 0.25201673140125486,
      "grad_norm": 0.13378937542438507,
      "learning_rate": 0.00017051098594026267,
      "loss": 0.5958,
      "step": 1687
    },
    {
      "epoch": 0.25216611891245894,
      "grad_norm": 0.24176190793514252,
      "learning_rate": 0.00017047764934813303,
      "loss": 1.1407,
      "step": 1688
    },
    {
      "epoch": 0.25231550642366296,
      "grad_norm": 0.18258847296237946,
      "learning_rate": 0.00017044429718637437,
      "loss": 0.7464,
      "step": 1689
    },
    {
      "epoch": 0.25246489393486704,
      "grad_norm": 0.1730254888534546,
      "learning_rate": 0.00017041092946235467,
      "loss": 0.5463,
      "step": 1690
    },
    {
      "epoch": 0.2526142814460711,
      "grad_norm": 0.09865910559892654,
      "learning_rate": 0.0001703775461834454,
      "loss": 0.6086,
      "step": 1691
    },
    {
      "epoch": 0.25276366895727514,
      "grad_norm": 0.16275614500045776,
      "learning_rate": 0.00017034414735702145,
      "loss": 0.8216,
      "step": 1692
    },
    {
      "epoch": 0.2529130564684792,
      "grad_norm": 0.22617553174495697,
      "learning_rate": 0.00017031073299046117,
      "loss": 1.282,
      "step": 1693
    },
    {
      "epoch": 0.2530624439796833,
      "grad_norm": 0.13999703526496887,
      "learning_rate": 0.0001702773030911463,
      "loss": 0.5571,
      "step": 1694
    },
    {
      "epoch": 0.2532118314908874,
      "grad_norm": 0.22081904113292694,
      "learning_rate": 0.00017024385766646204,
      "loss": 0.716,
      "step": 1695
    },
    {
      "epoch": 0.2533612190020914,
      "grad_norm": 0.1647699475288391,
      "learning_rate": 0.00017021039672379703,
      "loss": 0.7478,
      "step": 1696
    },
    {
      "epoch": 0.2535106065132955,
      "grad_norm": 1.2335481643676758,
      "learning_rate": 0.0001701769202705433,
      "loss": 1.2852,
      "step": 1697
    },
    {
      "epoch": 0.25365999402449957,
      "grad_norm": 0.13648250699043274,
      "learning_rate": 0.00017014342831409634,
      "loss": 0.7554,
      "step": 1698
    },
    {
      "epoch": 0.2538093815357036,
      "grad_norm": 0.17244575917720795,
      "learning_rate": 0.00017010992086185505,
      "loss": 0.839,
      "step": 1699
    },
    {
      "epoch": 0.2539587690469077,
      "grad_norm": 0.1780940443277359,
      "learning_rate": 0.00017007639792122173,
      "loss": 0.3983,
      "step": 1700
    },
    {
      "epoch": 0.25410815655811175,
      "grad_norm": 0.10594518482685089,
      "learning_rate": 0.00017004285949960218,
      "loss": 0.5268,
      "step": 1701
    },
    {
      "epoch": 0.25425754406931583,
      "grad_norm": 0.24656297266483307,
      "learning_rate": 0.00017000930560440554,
      "loss": 1.0687,
      "step": 1702
    },
    {
      "epoch": 0.25440693158051986,
      "grad_norm": 0.11236447095870972,
      "learning_rate": 0.00016997573624304445,
      "loss": 0.7733,
      "step": 1703
    },
    {
      "epoch": 0.25455631909172394,
      "grad_norm": 0.19340038299560547,
      "learning_rate": 0.0001699421514229348,
      "loss": 0.6995,
      "step": 1704
    },
    {
      "epoch": 0.254705706602928,
      "grad_norm": 0.16833987832069397,
      "learning_rate": 0.00016990855115149608,
      "loss": 0.4307,
      "step": 1705
    },
    {
      "epoch": 0.25485509411413204,
      "grad_norm": 0.14893771708011627,
      "learning_rate": 0.00016987493543615115,
      "loss": 0.6488,
      "step": 1706
    },
    {
      "epoch": 0.2550044816253361,
      "grad_norm": 0.2579140067100525,
      "learning_rate": 0.0001698413042843262,
      "loss": 0.7579,
      "step": 1707
    },
    {
      "epoch": 0.2551538691365402,
      "grad_norm": 0.12917333841323853,
      "learning_rate": 0.0001698076577034509,
      "loss": 0.4014,
      "step": 1708
    },
    {
      "epoch": 0.2553032566477442,
      "grad_norm": 0.12658651173114777,
      "learning_rate": 0.00016977399570095836,
      "loss": 0.8899,
      "step": 1709
    },
    {
      "epoch": 0.2554526441589483,
      "grad_norm": 0.18011939525604248,
      "learning_rate": 0.00016974031828428495,
      "loss": 0.4706,
      "step": 1710
    },
    {
      "epoch": 0.2556020316701524,
      "grad_norm": 0.1294494867324829,
      "learning_rate": 0.00016970662546087061,
      "loss": 0.7174,
      "step": 1711
    },
    {
      "epoch": 0.25575141918135647,
      "grad_norm": 0.11419305950403214,
      "learning_rate": 0.00016967291723815863,
      "loss": 0.7966,
      "step": 1712
    },
    {
      "epoch": 0.2559008066925605,
      "grad_norm": 0.13392961025238037,
      "learning_rate": 0.0001696391936235957,
      "loss": 0.8852,
      "step": 1713
    },
    {
      "epoch": 0.25605019420376457,
      "grad_norm": 0.10392757505178452,
      "learning_rate": 0.00016960545462463183,
      "loss": 0.5903,
      "step": 1714
    },
    {
      "epoch": 0.25619958171496865,
      "grad_norm": 0.1128273457288742,
      "learning_rate": 0.00016957170024872053,
      "loss": 0.7823,
      "step": 1715
    },
    {
      "epoch": 0.2563489692261727,
      "grad_norm": 0.16612160205841064,
      "learning_rate": 0.0001695379305033187,
      "loss": 0.6937,
      "step": 1716
    },
    {
      "epoch": 0.25649835673737675,
      "grad_norm": 0.17383776605129242,
      "learning_rate": 0.00016950414539588658,
      "loss": 0.7404,
      "step": 1717
    },
    {
      "epoch": 0.25664774424858083,
      "grad_norm": 0.19787214696407318,
      "learning_rate": 0.00016947034493388786,
      "loss": 0.5815,
      "step": 1718
    },
    {
      "epoch": 0.25679713175978486,
      "grad_norm": 0.25284725427627563,
      "learning_rate": 0.00016943652912478963,
      "loss": 0.8694,
      "step": 1719
    },
    {
      "epoch": 0.25694651927098894,
      "grad_norm": 0.15624594688415527,
      "learning_rate": 0.00016940269797606228,
      "loss": 0.6359,
      "step": 1720
    },
    {
      "epoch": 0.257095906782193,
      "grad_norm": 0.13568609952926636,
      "learning_rate": 0.00016936885149517966,
      "loss": 0.6661,
      "step": 1721
    },
    {
      "epoch": 0.2572452942933971,
      "grad_norm": 0.1571207344532013,
      "learning_rate": 0.00016933498968961898,
      "loss": 0.5693,
      "step": 1722
    },
    {
      "epoch": 0.2573946818046011,
      "grad_norm": 0.18845240771770477,
      "learning_rate": 0.0001693011125668609,
      "loss": 0.8907,
      "step": 1723
    },
    {
      "epoch": 0.2575440693158052,
      "grad_norm": 0.16922013461589813,
      "learning_rate": 0.00016926722013438936,
      "loss": 0.7377,
      "step": 1724
    },
    {
      "epoch": 0.2576934568270093,
      "grad_norm": 0.15387395024299622,
      "learning_rate": 0.00016923331239969175,
      "loss": 0.9347,
      "step": 1725
    },
    {
      "epoch": 0.2578428443382133,
      "grad_norm": 0.19560673832893372,
      "learning_rate": 0.00016919938937025886,
      "loss": 0.5263,
      "step": 1726
    },
    {
      "epoch": 0.2579922318494174,
      "grad_norm": 1.0470530986785889,
      "learning_rate": 0.00016916545105358477,
      "loss": 1.2196,
      "step": 1727
    },
    {
      "epoch": 0.25814161936062147,
      "grad_norm": 0.104112908244133,
      "learning_rate": 0.00016913149745716703,
      "loss": 0.5982,
      "step": 1728
    },
    {
      "epoch": 0.2582910068718255,
      "grad_norm": 0.22203882038593292,
      "learning_rate": 0.00016909752858850649,
      "loss": 0.9925,
      "step": 1729
    },
    {
      "epoch": 0.25844039438302957,
      "grad_norm": 0.13334202766418457,
      "learning_rate": 0.00016906354445510747,
      "loss": 0.8159,
      "step": 1730
    },
    {
      "epoch": 0.25858978189423365,
      "grad_norm": 0.1822672188282013,
      "learning_rate": 0.00016902954506447756,
      "loss": 0.5895,
      "step": 1731
    },
    {
      "epoch": 0.25873916940543773,
      "grad_norm": 0.13168515264987946,
      "learning_rate": 0.0001689955304241278,
      "loss": 0.8679,
      "step": 1732
    },
    {
      "epoch": 0.25888855691664175,
      "grad_norm": 0.1478189080953598,
      "learning_rate": 0.00016896150054157246,
      "loss": 1.0138,
      "step": 1733
    },
    {
      "epoch": 0.25903794442784583,
      "grad_norm": 0.12031956017017365,
      "learning_rate": 0.0001689274554243294,
      "loss": 0.6412,
      "step": 1734
    },
    {
      "epoch": 0.2591873319390499,
      "grad_norm": 0.13320299983024597,
      "learning_rate": 0.00016889339507991972,
      "loss": 0.8683,
      "step": 1735
    },
    {
      "epoch": 0.25933671945025394,
      "grad_norm": 0.2738458812236786,
      "learning_rate": 0.00016885931951586783,
      "loss": 1.3964,
      "step": 1736
    },
    {
      "epoch": 0.259486106961458,
      "grad_norm": 0.18454548716545105,
      "learning_rate": 0.00016882522873970157,
      "loss": 0.6525,
      "step": 1737
    },
    {
      "epoch": 0.2596354944726621,
      "grad_norm": 0.4490959644317627,
      "learning_rate": 0.00016879112275895215,
      "loss": 0.8293,
      "step": 1738
    },
    {
      "epoch": 0.2597848819838661,
      "grad_norm": 0.4163506329059601,
      "learning_rate": 0.0001687570015811541,
      "loss": 0.86,
      "step": 1739
    },
    {
      "epoch": 0.2599342694950702,
      "grad_norm": 0.14413362741470337,
      "learning_rate": 0.00016872286521384537,
      "loss": 0.5609,
      "step": 1740
    },
    {
      "epoch": 0.2600836570062743,
      "grad_norm": 0.1181647926568985,
      "learning_rate": 0.0001686887136645672,
      "loss": 0.7817,
      "step": 1741
    },
    {
      "epoch": 0.26023304451747836,
      "grad_norm": 0.13126704096794128,
      "learning_rate": 0.0001686545469408642,
      "loss": 0.7343,
      "step": 1742
    },
    {
      "epoch": 0.2603824320286824,
      "grad_norm": 0.11598172783851624,
      "learning_rate": 0.00016862036505028435,
      "loss": 0.774,
      "step": 1743
    },
    {
      "epoch": 0.26053181953988647,
      "grad_norm": 0.11938168853521347,
      "learning_rate": 0.000168586168000379,
      "loss": 0.6485,
      "step": 1744
    },
    {
      "epoch": 0.26068120705109055,
      "grad_norm": 0.15194714069366455,
      "learning_rate": 0.00016855195579870276,
      "loss": 0.7801,
      "step": 1745
    },
    {
      "epoch": 0.26083059456229457,
      "grad_norm": 0.11620033532381058,
      "learning_rate": 0.00016851772845281367,
      "loss": 0.6402,
      "step": 1746
    },
    {
      "epoch": 0.26097998207349865,
      "grad_norm": 0.11883977055549622,
      "learning_rate": 0.0001684834859702731,
      "loss": 0.8832,
      "step": 1747
    },
    {
      "epoch": 0.26112936958470273,
      "grad_norm": 0.17073407769203186,
      "learning_rate": 0.00016844922835864575,
      "loss": 0.7806,
      "step": 1748
    },
    {
      "epoch": 0.26127875709590676,
      "grad_norm": 0.15572674572467804,
      "learning_rate": 0.0001684149556254997,
      "loss": 0.7371,
      "step": 1749
    },
    {
      "epoch": 0.26142814460711084,
      "grad_norm": 0.11433901637792587,
      "learning_rate": 0.00016838066777840629,
      "loss": 0.5395,
      "step": 1750
    },
    {
      "epoch": 0.2615775321183149,
      "grad_norm": 0.23120561242103577,
      "learning_rate": 0.0001683463648249403,
      "loss": 0.861,
      "step": 1751
    },
    {
      "epoch": 0.261726919629519,
      "grad_norm": 0.21362490952014923,
      "learning_rate": 0.00016831204677267975,
      "loss": 0.6806,
      "step": 1752
    },
    {
      "epoch": 0.261876307140723,
      "grad_norm": 0.16032354533672333,
      "learning_rate": 0.0001682777136292061,
      "loss": 0.5942,
      "step": 1753
    },
    {
      "epoch": 0.2620256946519271,
      "grad_norm": 0.11645185202360153,
      "learning_rate": 0.00016824336540210402,
      "loss": 0.6547,
      "step": 1754
    },
    {
      "epoch": 0.2621750821631312,
      "grad_norm": 0.15311047434806824,
      "learning_rate": 0.0001682090020989616,
      "loss": 0.6924,
      "step": 1755
    },
    {
      "epoch": 0.2623244696743352,
      "grad_norm": 0.11865057051181793,
      "learning_rate": 0.0001681746237273702,
      "loss": 0.5077,
      "step": 1756
    },
    {
      "epoch": 0.2624738571855393,
      "grad_norm": 0.16047270596027374,
      "learning_rate": 0.00016814023029492465,
      "loss": 0.9401,
      "step": 1757
    },
    {
      "epoch": 0.26262324469674336,
      "grad_norm": 0.30298060178756714,
      "learning_rate": 0.00016810582180922293,
      "loss": 0.9419,
      "step": 1758
    },
    {
      "epoch": 0.26277263220794744,
      "grad_norm": 0.17041756212711334,
      "learning_rate": 0.0001680713982778664,
      "loss": 0.6029,
      "step": 1759
    },
    {
      "epoch": 0.26292201971915147,
      "grad_norm": 0.12116200476884842,
      "learning_rate": 0.00016803695970845985,
      "loss": 0.5324,
      "step": 1760
    },
    {
      "epoch": 0.26307140723035555,
      "grad_norm": 0.2546985149383545,
      "learning_rate": 0.00016800250610861123,
      "loss": 0.8727,
      "step": 1761
    },
    {
      "epoch": 0.2632207947415596,
      "grad_norm": 0.16526702046394348,
      "learning_rate": 0.0001679680374859319,
      "loss": 0.8323,
      "step": 1762
    },
    {
      "epoch": 0.26337018225276365,
      "grad_norm": 0.13681448996067047,
      "learning_rate": 0.00016793355384803652,
      "loss": 0.5819,
      "step": 1763
    },
    {
      "epoch": 0.26351956976396773,
      "grad_norm": 0.1449958086013794,
      "learning_rate": 0.0001678990552025431,
      "loss": 0.8017,
      "step": 1764
    },
    {
      "epoch": 0.2636689572751718,
      "grad_norm": 3.745661973953247,
      "learning_rate": 0.00016786454155707289,
      "loss": 2.6157,
      "step": 1765
    },
    {
      "epoch": 0.26381834478637584,
      "grad_norm": 0.1667809933423996,
      "learning_rate": 0.00016783001291925055,
      "loss": 0.5333,
      "step": 1766
    },
    {
      "epoch": 0.2639677322975799,
      "grad_norm": 0.1692322939634323,
      "learning_rate": 0.000167795469296704,
      "loss": 0.8491,
      "step": 1767
    },
    {
      "epoch": 0.264117119808784,
      "grad_norm": 0.18635401129722595,
      "learning_rate": 0.00016776091069706442,
      "loss": 0.7551,
      "step": 1768
    },
    {
      "epoch": 0.2642665073199881,
      "grad_norm": 0.1936548948287964,
      "learning_rate": 0.00016772633712796643,
      "loss": 0.6007,
      "step": 1769
    },
    {
      "epoch": 0.2644158948311921,
      "grad_norm": 0.22752203047275543,
      "learning_rate": 0.00016769174859704783,
      "loss": 1.2796,
      "step": 1770
    },
    {
      "epoch": 0.2645652823423962,
      "grad_norm": 0.22011956572532654,
      "learning_rate": 0.00016765714511194978,
      "loss": 0.7837,
      "step": 1771
    },
    {
      "epoch": 0.26471466985360026,
      "grad_norm": 0.12708061933517456,
      "learning_rate": 0.00016762252668031674,
      "loss": 0.8151,
      "step": 1772
    },
    {
      "epoch": 0.2648640573648043,
      "grad_norm": 0.16485874354839325,
      "learning_rate": 0.00016758789330979646,
      "loss": 0.7342,
      "step": 1773
    },
    {
      "epoch": 0.26501344487600836,
      "grad_norm": 0.15620820224285126,
      "learning_rate": 0.00016755324500804,
      "loss": 0.7541,
      "step": 1774
    },
    {
      "epoch": 0.26516283238721244,
      "grad_norm": 0.13292229175567627,
      "learning_rate": 0.00016751858178270176,
      "loss": 0.7372,
      "step": 1775
    },
    {
      "epoch": 0.26531221989841647,
      "grad_norm": 0.1720595508813858,
      "learning_rate": 0.00016748390364143938,
      "loss": 0.6299,
      "step": 1776
    },
    {
      "epoch": 0.26546160740962055,
      "grad_norm": 0.15689946711063385,
      "learning_rate": 0.00016744921059191377,
      "loss": 0.8878,
      "step": 1777
    },
    {
      "epoch": 0.26561099492082463,
      "grad_norm": 0.3296888470649719,
      "learning_rate": 0.00016741450264178917,
      "loss": 0.8915,
      "step": 1778
    },
    {
      "epoch": 0.2657603824320287,
      "grad_norm": 0.2351844161748886,
      "learning_rate": 0.00016737977979873318,
      "loss": 1.1235,
      "step": 1779
    },
    {
      "epoch": 0.26590976994323273,
      "grad_norm": 0.19016700983047485,
      "learning_rate": 0.00016734504207041663,
      "loss": 0.5777,
      "step": 1780
    },
    {
      "epoch": 0.2660591574544368,
      "grad_norm": 0.15265557169914246,
      "learning_rate": 0.00016731028946451353,
      "loss": 0.5994,
      "step": 1781
    },
    {
      "epoch": 0.2662085449656409,
      "grad_norm": 0.17204977571964264,
      "learning_rate": 0.00016727552198870135,
      "loss": 0.6589,
      "step": 1782
    },
    {
      "epoch": 0.2663579324768449,
      "grad_norm": 0.12247525155544281,
      "learning_rate": 0.00016724073965066082,
      "loss": 0.8026,
      "step": 1783
    },
    {
      "epoch": 0.266507319988049,
      "grad_norm": 0.3057596981525421,
      "learning_rate": 0.00016720594245807582,
      "loss": 0.8267,
      "step": 1784
    },
    {
      "epoch": 0.2666567074992531,
      "grad_norm": 0.1400333046913147,
      "learning_rate": 0.00016717113041863369,
      "loss": 0.6098,
      "step": 1785
    },
    {
      "epoch": 0.2668060950104571,
      "grad_norm": 0.13758665323257446,
      "learning_rate": 0.00016713630354002484,
      "loss": 0.8526,
      "step": 1786
    },
    {
      "epoch": 0.2669554825216612,
      "grad_norm": 0.12495380640029907,
      "learning_rate": 0.00016710146182994322,
      "loss": 0.6674,
      "step": 1787
    },
    {
      "epoch": 0.26710487003286526,
      "grad_norm": 0.12957152724266052,
      "learning_rate": 0.00016706660529608583,
      "loss": 0.8127,
      "step": 1788
    },
    {
      "epoch": 0.26725425754406934,
      "grad_norm": 0.18763193488121033,
      "learning_rate": 0.00016703173394615303,
      "loss": 0.9102,
      "step": 1789
    },
    {
      "epoch": 0.26740364505527336,
      "grad_norm": 0.17572201788425446,
      "learning_rate": 0.0001669968477878485,
      "loss": 0.8989,
      "step": 1790
    },
    {
      "epoch": 0.26755303256647744,
      "grad_norm": 0.16582119464874268,
      "learning_rate": 0.00016696194682887914,
      "loss": 1.0639,
      "step": 1791
    },
    {
      "epoch": 0.2677024200776815,
      "grad_norm": 0.42142021656036377,
      "learning_rate": 0.00016692703107695507,
      "loss": 1.1036,
      "step": 1792
    },
    {
      "epoch": 0.26785180758888555,
      "grad_norm": 0.21679432690143585,
      "learning_rate": 0.0001668921005397898,
      "loss": 0.5188,
      "step": 1793
    },
    {
      "epoch": 0.26800119510008963,
      "grad_norm": 0.16721928119659424,
      "learning_rate": 0.00016685715522509994,
      "loss": 0.8261,
      "step": 1794
    },
    {
      "epoch": 0.2681505826112937,
      "grad_norm": 0.1881019026041031,
      "learning_rate": 0.00016682219514060556,
      "loss": 0.9753,
      "step": 1795
    },
    {
      "epoch": 0.26829997012249773,
      "grad_norm": 0.18447300791740417,
      "learning_rate": 0.0001667872202940299,
      "loss": 0.8063,
      "step": 1796
    },
    {
      "epoch": 0.2684493576337018,
      "grad_norm": 0.13168549537658691,
      "learning_rate": 0.00016675223069309935,
      "loss": 0.7722,
      "step": 1797
    },
    {
      "epoch": 0.2685987451449059,
      "grad_norm": 0.3020504117012024,
      "learning_rate": 0.0001667172263455438,
      "loss": 0.8602,
      "step": 1798
    },
    {
      "epoch": 0.26874813265611,
      "grad_norm": 0.13748160004615784,
      "learning_rate": 0.00016668220725909617,
      "loss": 0.9058,
      "step": 1799
    },
    {
      "epoch": 0.268897520167314,
      "grad_norm": 0.19238565862178802,
      "learning_rate": 0.00016664717344149277,
      "loss": 0.6365,
      "step": 1800
    },
    {
      "epoch": 0.2690469076785181,
      "grad_norm": 0.2118740677833557,
      "learning_rate": 0.00016661212490047314,
      "loss": 0.7105,
      "step": 1801
    },
    {
      "epoch": 0.26919629518972216,
      "grad_norm": 0.17175479233264923,
      "learning_rate": 0.00016657706164378,
      "loss": 0.6657,
      "step": 1802
    },
    {
      "epoch": 0.2693456827009262,
      "grad_norm": 0.1766873598098755,
      "learning_rate": 0.00016654198367915948,
      "loss": 0.9525,
      "step": 1803
    },
    {
      "epoch": 0.26949507021213026,
      "grad_norm": 0.2608683705329895,
      "learning_rate": 0.00016650689101436073,
      "loss": 1.3181,
      "step": 1804
    },
    {
      "epoch": 0.26964445772333434,
      "grad_norm": 0.18048213422298431,
      "learning_rate": 0.00016647178365713637,
      "loss": 0.6132,
      "step": 1805
    },
    {
      "epoch": 0.26979384523453837,
      "grad_norm": 0.17701251804828644,
      "learning_rate": 0.00016643666161524217,
      "loss": 0.7133,
      "step": 1806
    },
    {
      "epoch": 0.26994323274574245,
      "grad_norm": 0.25974783301353455,
      "learning_rate": 0.00016640152489643705,
      "loss": 1.0835,
      "step": 1807
    },
    {
      "epoch": 0.2700926202569465,
      "grad_norm": 0.1478545069694519,
      "learning_rate": 0.00016636637350848338,
      "loss": 0.8298,
      "step": 1808
    },
    {
      "epoch": 0.2702420077681506,
      "grad_norm": 0.12416455149650574,
      "learning_rate": 0.00016633120745914663,
      "loss": 0.775,
      "step": 1809
    },
    {
      "epoch": 0.27039139527935463,
      "grad_norm": 0.14755703508853912,
      "learning_rate": 0.00016629602675619548,
      "loss": 0.7025,
      "step": 1810
    },
    {
      "epoch": 0.2705407827905587,
      "grad_norm": 0.2348576784133911,
      "learning_rate": 0.00016626083140740193,
      "loss": 0.5877,
      "step": 1811
    },
    {
      "epoch": 0.2706901703017628,
      "grad_norm": 0.19373933970928192,
      "learning_rate": 0.0001662256214205413,
      "loss": 0.6254,
      "step": 1812
    },
    {
      "epoch": 0.2708395578129668,
      "grad_norm": 0.2443433403968811,
      "learning_rate": 0.00016619039680339183,
      "loss": 1.0805,
      "step": 1813
    },
    {
      "epoch": 0.2709889453241709,
      "grad_norm": 0.1242481917142868,
      "learning_rate": 0.00016615515756373533,
      "loss": 0.8073,
      "step": 1814
    },
    {
      "epoch": 0.271138332835375,
      "grad_norm": 0.14438317716121674,
      "learning_rate": 0.00016611990370935672,
      "loss": 0.8175,
      "step": 1815
    },
    {
      "epoch": 0.27128772034657905,
      "grad_norm": 0.24103055894374847,
      "learning_rate": 0.00016608463524804407,
      "loss": 0.6098,
      "step": 1816
    },
    {
      "epoch": 0.2714371078577831,
      "grad_norm": 0.11732879281044006,
      "learning_rate": 0.00016604935218758878,
      "loss": 0.5935,
      "step": 1817
    },
    {
      "epoch": 0.27158649536898716,
      "grad_norm": 0.13892412185668945,
      "learning_rate": 0.0001660140545357854,
      "loss": 0.6891,
      "step": 1818
    },
    {
      "epoch": 0.27173588288019124,
      "grad_norm": 0.1574752926826477,
      "learning_rate": 0.0001659787423004318,
      "loss": 0.5805,
      "step": 1819
    },
    {
      "epoch": 0.27188527039139526,
      "grad_norm": 0.1563543677330017,
      "learning_rate": 0.00016594341548932894,
      "loss": 0.7904,
      "step": 1820
    },
    {
      "epoch": 0.27203465790259934,
      "grad_norm": 0.2136799395084381,
      "learning_rate": 0.0001659080741102811,
      "loss": 0.7477,
      "step": 1821
    },
    {
      "epoch": 0.2721840454138034,
      "grad_norm": 0.11503917723894119,
      "learning_rate": 0.0001658727181710958,
      "loss": 0.8353,
      "step": 1822
    },
    {
      "epoch": 0.27233343292500745,
      "grad_norm": 0.10798639804124832,
      "learning_rate": 0.00016583734767958367,
      "loss": 0.7049,
      "step": 1823
    },
    {
      "epoch": 0.2724828204362115,
      "grad_norm": 0.11826090514659882,
      "learning_rate": 0.0001658019626435586,
      "loss": 0.8643,
      "step": 1824
    },
    {
      "epoch": 0.2726322079474156,
      "grad_norm": 0.19389161467552185,
      "learning_rate": 0.00016576656307083778,
      "loss": 0.585,
      "step": 1825
    },
    {
      "epoch": 0.2727815954586197,
      "grad_norm": 0.13553759455680847,
      "learning_rate": 0.00016573114896924147,
      "loss": 0.7292,
      "step": 1826
    },
    {
      "epoch": 0.2729309829698237,
      "grad_norm": 0.1668972373008728,
      "learning_rate": 0.00016569572034659324,
      "loss": 0.6368,
      "step": 1827
    },
    {
      "epoch": 0.2730803704810278,
      "grad_norm": 0.22873035073280334,
      "learning_rate": 0.0001656602772107198,
      "loss": 0.8858,
      "step": 1828
    },
    {
      "epoch": 0.27322975799223187,
      "grad_norm": 0.13842542469501495,
      "learning_rate": 0.0001656248195694511,
      "loss": 0.6738,
      "step": 1829
    },
    {
      "epoch": 0.2733791455034359,
      "grad_norm": 0.14224855601787567,
      "learning_rate": 0.00016558934743062035,
      "loss": 0.6905,
      "step": 1830
    },
    {
      "epoch": 0.27352853301464,
      "grad_norm": 0.28141728043556213,
      "learning_rate": 0.00016555386080206383,
      "loss": 0.6765,
      "step": 1831
    },
    {
      "epoch": 0.27367792052584405,
      "grad_norm": 0.16277846693992615,
      "learning_rate": 0.00016551835969162118,
      "loss": 0.8857,
      "step": 1832
    },
    {
      "epoch": 0.2738273080370481,
      "grad_norm": 0.1706254631280899,
      "learning_rate": 0.0001654828441071351,
      "loss": 0.6825,
      "step": 1833
    },
    {
      "epoch": 0.27397669554825216,
      "grad_norm": 0.15771432220935822,
      "learning_rate": 0.00016544731405645154,
      "loss": 0.7739,
      "step": 1834
    },
    {
      "epoch": 0.27412608305945624,
      "grad_norm": 0.15347278118133545,
      "learning_rate": 0.00016541176954741973,
      "loss": 0.719,
      "step": 1835
    },
    {
      "epoch": 0.2742754705706603,
      "grad_norm": 0.2687349319458008,
      "learning_rate": 0.00016537621058789194,
      "loss": 0.7605,
      "step": 1836
    },
    {
      "epoch": 0.27442485808186434,
      "grad_norm": 0.1504988670349121,
      "learning_rate": 0.0001653406371857237,
      "loss": 0.933,
      "step": 1837
    },
    {
      "epoch": 0.2745742455930684,
      "grad_norm": 0.11636245250701904,
      "learning_rate": 0.00016530504934877377,
      "loss": 0.4423,
      "step": 1838
    },
    {
      "epoch": 0.2747236331042725,
      "grad_norm": 0.13434423506259918,
      "learning_rate": 0.00016526944708490407,
      "loss": 1.0305,
      "step": 1839
    },
    {
      "epoch": 0.2748730206154765,
      "grad_norm": 0.16145826876163483,
      "learning_rate": 0.0001652338304019797,
      "loss": 0.4641,
      "step": 1840
    },
    {
      "epoch": 0.2750224081266806,
      "grad_norm": 0.29437121748924255,
      "learning_rate": 0.00016519819930786893,
      "loss": 0.5978,
      "step": 1841
    },
    {
      "epoch": 0.2751717956378847,
      "grad_norm": 0.1476840227842331,
      "learning_rate": 0.00016516255381044323,
      "loss": 0.8374,
      "step": 1842
    },
    {
      "epoch": 0.2753211831490887,
      "grad_norm": 0.1587522178888321,
      "learning_rate": 0.0001651268939175773,
      "loss": 0.701,
      "step": 1843
    },
    {
      "epoch": 0.2754705706602928,
      "grad_norm": 0.18530099093914032,
      "learning_rate": 0.00016509121963714896,
      "loss": 0.5378,
      "step": 1844
    },
    {
      "epoch": 0.27561995817149687,
      "grad_norm": 0.17483104765415192,
      "learning_rate": 0.00016505553097703916,
      "loss": 0.7805,
      "step": 1845
    },
    {
      "epoch": 0.27576934568270095,
      "grad_norm": 0.1965104341506958,
      "learning_rate": 0.00016501982794513219,
      "loss": 0.9346,
      "step": 1846
    },
    {
      "epoch": 0.275918733193905,
      "grad_norm": 0.18628351390361786,
      "learning_rate": 0.00016498411054931533,
      "loss": 0.5019,
      "step": 1847
    },
    {
      "epoch": 0.27606812070510905,
      "grad_norm": 0.139807790517807,
      "learning_rate": 0.00016494837879747916,
      "loss": 0.8478,
      "step": 1848
    },
    {
      "epoch": 0.27621750821631313,
      "grad_norm": 0.14454811811447144,
      "learning_rate": 0.0001649126326975174,
      "loss": 0.7929,
      "step": 1849
    },
    {
      "epoch": 0.27636689572751716,
      "grad_norm": 0.21324700117111206,
      "learning_rate": 0.00016487687225732694,
      "loss": 0.8371,
      "step": 1850
    },
    {
      "epoch": 0.27651628323872124,
      "grad_norm": 0.24831005930900574,
      "learning_rate": 0.00016484109748480776,
      "loss": 0.9214,
      "step": 1851
    },
    {
      "epoch": 0.2766656707499253,
      "grad_norm": 0.11538607627153397,
      "learning_rate": 0.00016480530838786312,
      "loss": 0.6869,
      "step": 1852
    },
    {
      "epoch": 0.27681505826112934,
      "grad_norm": 0.166727215051651,
      "learning_rate": 0.0001647695049743994,
      "loss": 1.0586,
      "step": 1853
    },
    {
      "epoch": 0.2769644457723334,
      "grad_norm": 0.15276376903057098,
      "learning_rate": 0.00016473368725232614,
      "loss": 0.7167,
      "step": 1854
    },
    {
      "epoch": 0.2771138332835375,
      "grad_norm": 0.191617950797081,
      "learning_rate": 0.00016469785522955604,
      "loss": 0.6044,
      "step": 1855
    },
    {
      "epoch": 0.2772632207947416,
      "grad_norm": 0.20602108538150787,
      "learning_rate": 0.0001646620089140049,
      "loss": 0.768,
      "step": 1856
    },
    {
      "epoch": 0.2774126083059456,
      "grad_norm": 0.19726546108722687,
      "learning_rate": 0.0001646261483135919,
      "loss": 0.6565,
      "step": 1857
    },
    {
      "epoch": 0.2775619958171497,
      "grad_norm": 0.16957935690879822,
      "learning_rate": 0.00016459027343623906,
      "loss": 0.5894,
      "step": 1858
    },
    {
      "epoch": 0.27771138332835377,
      "grad_norm": 0.15182845294475555,
      "learning_rate": 0.00016455438428987176,
      "loss": 1.0849,
      "step": 1859
    },
    {
      "epoch": 0.2778607708395578,
      "grad_norm": 0.11782501637935638,
      "learning_rate": 0.00016451848088241847,
      "loss": 0.8136,
      "step": 1860
    },
    {
      "epoch": 0.27801015835076187,
      "grad_norm": 0.17523236572742462,
      "learning_rate": 0.00016448256322181085,
      "loss": 0.8055,
      "step": 1861
    },
    {
      "epoch": 0.27815954586196595,
      "grad_norm": 0.1287536919116974,
      "learning_rate": 0.00016444663131598365,
      "loss": 0.687,
      "step": 1862
    },
    {
      "epoch": 0.27830893337317,
      "grad_norm": 0.21639418601989746,
      "learning_rate": 0.00016441068517287482,
      "loss": 1.1522,
      "step": 1863
    },
    {
      "epoch": 0.27845832088437406,
      "grad_norm": 0.13961181044578552,
      "learning_rate": 0.00016437472480042544,
      "loss": 0.8324,
      "step": 1864
    },
    {
      "epoch": 0.27860770839557814,
      "grad_norm": 0.2107275277376175,
      "learning_rate": 0.00016433875020657968,
      "loss": 0.9217,
      "step": 1865
    },
    {
      "epoch": 0.2787570959067822,
      "grad_norm": 0.1293714940547943,
      "learning_rate": 0.00016430276139928494,
      "loss": 0.7742,
      "step": 1866
    },
    {
      "epoch": 0.27890648341798624,
      "grad_norm": 0.1160321906208992,
      "learning_rate": 0.00016426675838649172,
      "loss": 0.8035,
      "step": 1867
    },
    {
      "epoch": 0.2790558709291903,
      "grad_norm": 0.12264332175254822,
      "learning_rate": 0.00016423074117615362,
      "loss": 0.7399,
      "step": 1868
    },
    {
      "epoch": 0.2792052584403944,
      "grad_norm": 0.14644618332386017,
      "learning_rate": 0.00016419470977622742,
      "loss": 0.7063,
      "step": 1869
    },
    {
      "epoch": 0.2793546459515984,
      "grad_norm": 0.15369118750095367,
      "learning_rate": 0.00016415866419467308,
      "loss": 0.9317,
      "step": 1870
    },
    {
      "epoch": 0.2795040334628025,
      "grad_norm": 0.176469087600708,
      "learning_rate": 0.00016412260443945357,
      "loss": 0.9826,
      "step": 1871
    },
    {
      "epoch": 0.2796534209740066,
      "grad_norm": 0.12978896498680115,
      "learning_rate": 0.00016408653051853505,
      "loss": 0.9042,
      "step": 1872
    },
    {
      "epoch": 0.27980280848521066,
      "grad_norm": 0.1965731978416443,
      "learning_rate": 0.00016405044243988695,
      "loss": 0.6585,
      "step": 1873
    },
    {
      "epoch": 0.2799521959964147,
      "grad_norm": 0.17228776216506958,
      "learning_rate": 0.00016401434021148155,
      "loss": 0.8369,
      "step": 1874
    },
    {
      "epoch": 0.28010158350761877,
      "grad_norm": 0.2215079665184021,
      "learning_rate": 0.00016397822384129444,
      "loss": 1.0549,
      "step": 1875
    },
    {
      "epoch": 0.28025097101882285,
      "grad_norm": 0.15248584747314453,
      "learning_rate": 0.00016394209333730437,
      "loss": 0.7295,
      "step": 1876
    },
    {
      "epoch": 0.28040035853002687,
      "grad_norm": 0.11419791728258133,
      "learning_rate": 0.00016390594870749307,
      "loss": 0.5573,
      "step": 1877
    },
    {
      "epoch": 0.28054974604123095,
      "grad_norm": 0.1532645970582962,
      "learning_rate": 0.0001638697899598455,
      "loss": 0.6374,
      "step": 1878
    },
    {
      "epoch": 0.28069913355243503,
      "grad_norm": 0.27644631266593933,
      "learning_rate": 0.00016383361710234966,
      "loss": 1.0993,
      "step": 1879
    },
    {
      "epoch": 0.28084852106363906,
      "grad_norm": 0.5468740463256836,
      "learning_rate": 0.00016379743014299675,
      "loss": 1.2103,
      "step": 1880
    },
    {
      "epoch": 0.28099790857484314,
      "grad_norm": 0.2976016700267792,
      "learning_rate": 0.000163761229089781,
      "loss": 0.8506,
      "step": 1881
    },
    {
      "epoch": 0.2811472960860472,
      "grad_norm": 0.3081592917442322,
      "learning_rate": 0.00016372501395069984,
      "loss": 0.8399,
      "step": 1882
    },
    {
      "epoch": 0.2812966835972513,
      "grad_norm": 0.5960981249809265,
      "learning_rate": 0.00016368878473375372,
      "loss": 1.0627,
      "step": 1883
    },
    {
      "epoch": 0.2814460711084553,
      "grad_norm": 0.16949574649333954,
      "learning_rate": 0.0001636525414469463,
      "loss": 0.6284,
      "step": 1884
    },
    {
      "epoch": 0.2815954586196594,
      "grad_norm": 0.14496923983097076,
      "learning_rate": 0.00016361628409828426,
      "loss": 0.5671,
      "step": 1885
    },
    {
      "epoch": 0.2817448461308635,
      "grad_norm": 0.1284521222114563,
      "learning_rate": 0.00016358001269577743,
      "loss": 0.7364,
      "step": 1886
    },
    {
      "epoch": 0.2818942336420675,
      "grad_norm": 0.10659526288509369,
      "learning_rate": 0.00016354372724743877,
      "loss": 0.6766,
      "step": 1887
    },
    {
      "epoch": 0.2820436211532716,
      "grad_norm": 0.13739757239818573,
      "learning_rate": 0.00016350742776128423,
      "loss": 0.8011,
      "step": 1888
    },
    {
      "epoch": 0.28219300866447566,
      "grad_norm": 0.14763128757476807,
      "learning_rate": 0.00016347111424533302,
      "loss": 0.8036,
      "step": 1889
    },
    {
      "epoch": 0.2823423961756797,
      "grad_norm": 0.14890700578689575,
      "learning_rate": 0.00016343478670760732,
      "loss": 0.9068,
      "step": 1890
    },
    {
      "epoch": 0.28249178368688377,
      "grad_norm": 0.12007980793714523,
      "learning_rate": 0.0001633984451561325,
      "loss": 0.4907,
      "step": 1891
    },
    {
      "epoch": 0.28264117119808785,
      "grad_norm": 0.258691668510437,
      "learning_rate": 0.00016336208959893698,
      "loss": 0.7515,
      "step": 1892
    },
    {
      "epoch": 0.28279055870929193,
      "grad_norm": 0.21225577592849731,
      "learning_rate": 0.00016332572004405224,
      "loss": 0.6392,
      "step": 1893
    },
    {
      "epoch": 0.28293994622049595,
      "grad_norm": 0.15706035494804382,
      "learning_rate": 0.00016328933649951293,
      "loss": 0.7764,
      "step": 1894
    },
    {
      "epoch": 0.28308933373170003,
      "grad_norm": 0.8196190595626831,
      "learning_rate": 0.0001632529389733567,
      "loss": 1.6892,
      "step": 1895
    },
    {
      "epoch": 0.2832387212429041,
      "grad_norm": 0.1771058291196823,
      "learning_rate": 0.00016321652747362445,
      "loss": 1.2452,
      "step": 1896
    },
    {
      "epoch": 0.28338810875410814,
      "grad_norm": 0.17324721813201904,
      "learning_rate": 0.00016318010200835992,
      "loss": 0.3363,
      "step": 1897
    },
    {
      "epoch": 0.2835374962653122,
      "grad_norm": 0.14077776670455933,
      "learning_rate": 0.00016314366258561016,
      "loss": 0.495,
      "step": 1898
    },
    {
      "epoch": 0.2836868837765163,
      "grad_norm": 0.1295582801103592,
      "learning_rate": 0.00016310720921342518,
      "loss": 0.7699,
      "step": 1899
    },
    {
      "epoch": 0.2838362712877203,
      "grad_norm": 0.18893714249134064,
      "learning_rate": 0.00016307074189985814,
      "loss": 0.9228,
      "step": 1900
    },
    {
      "epoch": 0.2839856587989244,
      "grad_norm": 0.18363916873931885,
      "learning_rate": 0.00016303426065296522,
      "loss": 0.557,
      "step": 1901
    },
    {
      "epoch": 0.2841350463101285,
      "grad_norm": 0.156977578997612,
      "learning_rate": 0.0001629977654808057,
      "loss": 0.7707,
      "step": 1902
    },
    {
      "epoch": 0.28428443382133256,
      "grad_norm": 0.23877717554569244,
      "learning_rate": 0.00016296125639144192,
      "loss": 0.8239,
      "step": 1903
    },
    {
      "epoch": 0.2844338213325366,
      "grad_norm": 0.13115395605564117,
      "learning_rate": 0.0001629247333929394,
      "loss": 0.7643,
      "step": 1904
    },
    {
      "epoch": 0.28458320884374066,
      "grad_norm": 0.12260690331459045,
      "learning_rate": 0.00016288819649336656,
      "loss": 0.6755,
      "step": 1905
    },
    {
      "epoch": 0.28473259635494474,
      "grad_norm": 0.1967897266149521,
      "learning_rate": 0.00016285164570079504,
      "loss": 0.5642,
      "step": 1906
    },
    {
      "epoch": 0.28488198386614877,
      "grad_norm": 0.263201504945755,
      "learning_rate": 0.00016281508102329947,
      "loss": 0.9733,
      "step": 1907
    },
    {
      "epoch": 0.28503137137735285,
      "grad_norm": 0.18227405846118927,
      "learning_rate": 0.00016277850246895753,
      "loss": 0.8968,
      "step": 1908
    },
    {
      "epoch": 0.28518075888855693,
      "grad_norm": 0.3462936282157898,
      "learning_rate": 0.00016274191004585009,
      "loss": 0.9113,
      "step": 1909
    },
    {
      "epoch": 0.28533014639976095,
      "grad_norm": 0.13814112544059753,
      "learning_rate": 0.0001627053037620609,
      "loss": 0.8253,
      "step": 1910
    },
    {
      "epoch": 0.28547953391096503,
      "grad_norm": 0.27378007769584656,
      "learning_rate": 0.0001626686836256769,
      "loss": 0.7296,
      "step": 1911
    },
    {
      "epoch": 0.2856289214221691,
      "grad_norm": 0.16056250035762787,
      "learning_rate": 0.00016263204964478807,
      "loss": 0.9768,
      "step": 1912
    },
    {
      "epoch": 0.2857783089333732,
      "grad_norm": 0.28953564167022705,
      "learning_rate": 0.00016259540182748746,
      "loss": 0.5664,
      "step": 1913
    },
    {
      "epoch": 0.2859276964445772,
      "grad_norm": 0.22288711369037628,
      "learning_rate": 0.00016255874018187113,
      "loss": 0.9053,
      "step": 1914
    },
    {
      "epoch": 0.2860770839557813,
      "grad_norm": 0.153893381357193,
      "learning_rate": 0.00016252206471603818,
      "loss": 0.8007,
      "step": 1915
    },
    {
      "epoch": 0.2862264714669854,
      "grad_norm": 0.2964473068714142,
      "learning_rate": 0.00016248537543809085,
      "loss": 0.6401,
      "step": 1916
    },
    {
      "epoch": 0.2863758589781894,
      "grad_norm": 0.16791103780269623,
      "learning_rate": 0.00016244867235613438,
      "loss": 0.8178,
      "step": 1917
    },
    {
      "epoch": 0.2865252464893935,
      "grad_norm": 0.3182014524936676,
      "learning_rate": 0.00016241195547827704,
      "loss": 0.96,
      "step": 1918
    },
    {
      "epoch": 0.28667463400059756,
      "grad_norm": 0.15290340781211853,
      "learning_rate": 0.00016237522481263018,
      "loss": 0.9855,
      "step": 1919
    },
    {
      "epoch": 0.2868240215118016,
      "grad_norm": 0.36779022216796875,
      "learning_rate": 0.00016233848036730818,
      "loss": 1.0061,
      "step": 1920
    },
    {
      "epoch": 0.28697340902300567,
      "grad_norm": 0.2556033432483673,
      "learning_rate": 0.0001623017221504285,
      "loss": 0.7636,
      "step": 1921
    },
    {
      "epoch": 0.28712279653420975,
      "grad_norm": 0.19862814247608185,
      "learning_rate": 0.00016226495017011155,
      "loss": 0.8552,
      "step": 1922
    },
    {
      "epoch": 0.2872721840454138,
      "grad_norm": 0.18107233941555023,
      "learning_rate": 0.00016222816443448092,
      "loss": 0.907,
      "step": 1923
    },
    {
      "epoch": 0.28742157155661785,
      "grad_norm": 0.33394572138786316,
      "learning_rate": 0.0001621913649516631,
      "loss": 1.0993,
      "step": 1924
    },
    {
      "epoch": 0.28757095906782193,
      "grad_norm": 0.18873590230941772,
      "learning_rate": 0.00016215455172978772,
      "loss": 1.1411,
      "step": 1925
    },
    {
      "epoch": 0.287720346579026,
      "grad_norm": 0.17854920029640198,
      "learning_rate": 0.00016211772477698737,
      "loss": 1.1384,
      "step": 1926
    },
    {
      "epoch": 0.28786973409023003,
      "grad_norm": 0.1506814956665039,
      "learning_rate": 0.00016208088410139772,
      "loss": 1.0847,
      "step": 1927
    },
    {
      "epoch": 0.2880191216014341,
      "grad_norm": 0.11812624335289001,
      "learning_rate": 0.0001620440297111575,
      "loss": 0.7856,
      "step": 1928
    },
    {
      "epoch": 0.2881685091126382,
      "grad_norm": 0.36629819869995117,
      "learning_rate": 0.00016200716161440833,
      "loss": 1.1171,
      "step": 1929
    },
    {
      "epoch": 0.2883178966238423,
      "grad_norm": 0.12690721452236176,
      "learning_rate": 0.00016197027981929506,
      "loss": 0.8426,
      "step": 1930
    },
    {
      "epoch": 0.2884672841350463,
      "grad_norm": 0.17128652334213257,
      "learning_rate": 0.00016193338433396538,
      "loss": 0.7219,
      "step": 1931
    },
    {
      "epoch": 0.2886166716462504,
      "grad_norm": 0.19333554804325104,
      "learning_rate": 0.00016189647516657018,
      "loss": 0.4354,
      "step": 1932
    },
    {
      "epoch": 0.28876605915745446,
      "grad_norm": 0.1483379602432251,
      "learning_rate": 0.0001618595523252632,
      "loss": 0.5917,
      "step": 1933
    },
    {
      "epoch": 0.2889154466686585,
      "grad_norm": 0.18341517448425293,
      "learning_rate": 0.0001618226158182013,
      "loss": 0.5832,
      "step": 1934
    },
    {
      "epoch": 0.28906483417986256,
      "grad_norm": 0.13665004074573517,
      "learning_rate": 0.00016178566565354437,
      "loss": 1.0354,
      "step": 1935
    },
    {
      "epoch": 0.28921422169106664,
      "grad_norm": 0.19725897908210754,
      "learning_rate": 0.00016174870183945523,
      "loss": 0.8194,
      "step": 1936
    },
    {
      "epoch": 0.28936360920227067,
      "grad_norm": 0.1692163348197937,
      "learning_rate": 0.0001617117243840998,
      "loss": 0.9514,
      "step": 1937
    },
    {
      "epoch": 0.28951299671347475,
      "grad_norm": 0.22694113850593567,
      "learning_rate": 0.00016167473329564705,
      "loss": 0.6955,
      "step": 1938
    },
    {
      "epoch": 0.2896623842246788,
      "grad_norm": 0.16074326634407043,
      "learning_rate": 0.00016163772858226877,
      "loss": 0.8205,
      "step": 1939
    },
    {
      "epoch": 0.2898117717358829,
      "grad_norm": 0.1290196180343628,
      "learning_rate": 0.00016160071025213998,
      "loss": 0.76,
      "step": 1940
    },
    {
      "epoch": 0.28996115924708693,
      "grad_norm": 0.12540550529956818,
      "learning_rate": 0.0001615636783134386,
      "loss": 0.6737,
      "step": 1941
    },
    {
      "epoch": 0.290110546758291,
      "grad_norm": 0.1340838223695755,
      "learning_rate": 0.00016152663277434556,
      "loss": 0.9725,
      "step": 1942
    },
    {
      "epoch": 0.2902599342694951,
      "grad_norm": 0.14680004119873047,
      "learning_rate": 0.0001614895736430448,
      "loss": 0.7528,
      "step": 1943
    },
    {
      "epoch": 0.2904093217806991,
      "grad_norm": 0.2111632525920868,
      "learning_rate": 0.0001614525009277233,
      "loss": 0.9577,
      "step": 1944
    },
    {
      "epoch": 0.2905587092919032,
      "grad_norm": 0.16464073956012726,
      "learning_rate": 0.00016141541463657094,
      "loss": 0.7268,
      "step": 1945
    },
    {
      "epoch": 0.2907080968031073,
      "grad_norm": 0.13918276131153107,
      "learning_rate": 0.00016137831477778077,
      "loss": 0.7131,
      "step": 1946
    },
    {
      "epoch": 0.2908574843143113,
      "grad_norm": 0.18370576202869415,
      "learning_rate": 0.00016134120135954865,
      "loss": 0.5412,
      "step": 1947
    },
    {
      "epoch": 0.2910068718255154,
      "grad_norm": 0.17753422260284424,
      "learning_rate": 0.00016130407439007355,
      "loss": 0.9692,
      "step": 1948
    },
    {
      "epoch": 0.29115625933671946,
      "grad_norm": 0.22409065067768097,
      "learning_rate": 0.00016126693387755742,
      "loss": 0.7073,
      "step": 1949
    },
    {
      "epoch": 0.29130564684792354,
      "grad_norm": 0.11738231033086777,
      "learning_rate": 0.0001612297798302052,
      "loss": 0.6564,
      "step": 1950
    },
    {
      "epoch": 0.29145503435912756,
      "grad_norm": 0.13526974618434906,
      "learning_rate": 0.0001611926122562248,
      "loss": 0.7621,
      "step": 1951
    },
    {
      "epoch": 0.29160442187033164,
      "grad_norm": 0.14819179475307465,
      "learning_rate": 0.00016115543116382707,
      "loss": 0.7947,
      "step": 1952
    },
    {
      "epoch": 0.2917538093815357,
      "grad_norm": 0.1239335685968399,
      "learning_rate": 0.00016111823656122596,
      "loss": 0.7881,
      "step": 1953
    },
    {
      "epoch": 0.29190319689273975,
      "grad_norm": 0.1371903270483017,
      "learning_rate": 0.00016108102845663832,
      "loss": 0.7273,
      "step": 1954
    },
    {
      "epoch": 0.2920525844039438,
      "grad_norm": 0.2668689489364624,
      "learning_rate": 0.00016104380685828403,
      "loss": 0.8244,
      "step": 1955
    },
    {
      "epoch": 0.2922019719151479,
      "grad_norm": 0.27288052439689636,
      "learning_rate": 0.00016100657177438592,
      "loss": 1.1713,
      "step": 1956
    },
    {
      "epoch": 0.29235135942635193,
      "grad_norm": 0.12943297624588013,
      "learning_rate": 0.0001609693232131698,
      "loss": 0.5734,
      "step": 1957
    },
    {
      "epoch": 0.292500746937556,
      "grad_norm": 0.1556374877691269,
      "learning_rate": 0.0001609320611828645,
      "loss": 0.7606,
      "step": 1958
    },
    {
      "epoch": 0.2926501344487601,
      "grad_norm": 0.1746862679719925,
      "learning_rate": 0.00016089478569170174,
      "loss": 0.7793,
      "step": 1959
    },
    {
      "epoch": 0.29279952195996417,
      "grad_norm": 0.18083463609218597,
      "learning_rate": 0.0001608574967479163,
      "loss": 0.5813,
      "step": 1960
    },
    {
      "epoch": 0.2929489094711682,
      "grad_norm": 0.25068581104278564,
      "learning_rate": 0.00016082019435974593,
      "loss": 1.1797,
      "step": 1961
    },
    {
      "epoch": 0.2930982969823723,
      "grad_norm": 0.1725584864616394,
      "learning_rate": 0.00016078287853543125,
      "loss": 0.7816,
      "step": 1962
    },
    {
      "epoch": 0.29324768449357635,
      "grad_norm": 0.19390223920345306,
      "learning_rate": 0.00016074554928321598,
      "loss": 0.5733,
      "step": 1963
    },
    {
      "epoch": 0.2933970720047804,
      "grad_norm": 0.1887221783399582,
      "learning_rate": 0.00016070820661134668,
      "loss": 0.78,
      "step": 1964
    },
    {
      "epoch": 0.29354645951598446,
      "grad_norm": 0.1666865348815918,
      "learning_rate": 0.000160670850528073,
      "loss": 0.7833,
      "step": 1965
    },
    {
      "epoch": 0.29369584702718854,
      "grad_norm": 0.13192425668239594,
      "learning_rate": 0.00016063348104164744,
      "loss": 0.706,
      "step": 1966
    },
    {
      "epoch": 0.29384523453839256,
      "grad_norm": 0.17119993269443512,
      "learning_rate": 0.00016059609816032556,
      "loss": 0.7467,
      "step": 1967
    },
    {
      "epoch": 0.29399462204959664,
      "grad_norm": 0.12513847649097443,
      "learning_rate": 0.00016055870189236578,
      "loss": 0.6887,
      "step": 1968
    },
    {
      "epoch": 0.2941440095608007,
      "grad_norm": 0.13904425501823425,
      "learning_rate": 0.00016052129224602957,
      "loss": 0.661,
      "step": 1969
    },
    {
      "epoch": 0.2942933970720048,
      "grad_norm": 0.12954014539718628,
      "learning_rate": 0.00016048386922958127,
      "loss": 0.774,
      "step": 1970
    },
    {
      "epoch": 0.2944427845832088,
      "grad_norm": 0.17040160298347473,
      "learning_rate": 0.00016044643285128827,
      "loss": 0.8955,
      "step": 1971
    },
    {
      "epoch": 0.2945921720944129,
      "grad_norm": 0.3505501449108124,
      "learning_rate": 0.00016040898311942082,
      "loss": 0.6914,
      "step": 1972
    },
    {
      "epoch": 0.294741559605617,
      "grad_norm": 0.14326271414756775,
      "learning_rate": 0.00016037152004225215,
      "loss": 0.6905,
      "step": 1973
    },
    {
      "epoch": 0.294890947116821,
      "grad_norm": 0.12812425196170807,
      "learning_rate": 0.0001603340436280585,
      "loss": 0.7456,
      "step": 1974
    },
    {
      "epoch": 0.2950403346280251,
      "grad_norm": 0.20053675770759583,
      "learning_rate": 0.00016029655388511896,
      "loss": 0.6051,
      "step": 1975
    },
    {
      "epoch": 0.29518972213922917,
      "grad_norm": 0.1312316656112671,
      "learning_rate": 0.00016025905082171562,
      "loss": 0.6035,
      "step": 1976
    },
    {
      "epoch": 0.29533910965043325,
      "grad_norm": 0.14165259897708893,
      "learning_rate": 0.00016022153444613348,
      "loss": 0.8824,
      "step": 1977
    },
    {
      "epoch": 0.2954884971616373,
      "grad_norm": 0.17321546375751495,
      "learning_rate": 0.00016018400476666055,
      "loss": 0.8771,
      "step": 1978
    },
    {
      "epoch": 0.29563788467284136,
      "grad_norm": 0.11130014806985855,
      "learning_rate": 0.00016014646179158767,
      "loss": 0.6402,
      "step": 1979
    },
    {
      "epoch": 0.29578727218404544,
      "grad_norm": 0.22360561788082123,
      "learning_rate": 0.00016010890552920875,
      "loss": 0.4773,
      "step": 1980
    },
    {
      "epoch": 0.29593665969524946,
      "grad_norm": 0.1546655297279358,
      "learning_rate": 0.00016007133598782048,
      "loss": 1.1104,
      "step": 1981
    },
    {
      "epoch": 0.29608604720645354,
      "grad_norm": 0.3024921417236328,
      "learning_rate": 0.00016003375317572263,
      "loss": 0.6703,
      "step": 1982
    },
    {
      "epoch": 0.2962354347176576,
      "grad_norm": 0.4379715919494629,
      "learning_rate": 0.00015999615710121784,
      "loss": 1.0485,
      "step": 1983
    },
    {
      "epoch": 0.29638482222886164,
      "grad_norm": 0.1305762678384781,
      "learning_rate": 0.00015995854777261161,
      "loss": 0.9361,
      "step": 1984
    },
    {
      "epoch": 0.2965342097400657,
      "grad_norm": 0.12884147465229034,
      "learning_rate": 0.0001599209251982125,
      "loss": 0.9236,
      "step": 1985
    },
    {
      "epoch": 0.2966835972512698,
      "grad_norm": 0.15657733380794525,
      "learning_rate": 0.00015988328938633191,
      "loss": 0.9226,
      "step": 1986
    },
    {
      "epoch": 0.2968329847624739,
      "grad_norm": 0.15643414855003357,
      "learning_rate": 0.00015984564034528422,
      "loss": 0.7931,
      "step": 1987
    },
    {
      "epoch": 0.2969823722736779,
      "grad_norm": 0.15294334292411804,
      "learning_rate": 0.00015980797808338664,
      "loss": 0.9234,
      "step": 1988
    },
    {
      "epoch": 0.297131759784882,
      "grad_norm": 0.17562635242938995,
      "learning_rate": 0.00015977030260895943,
      "loss": 1.0773,
      "step": 1989
    },
    {
      "epoch": 0.29728114729608607,
      "grad_norm": 0.15384536981582642,
      "learning_rate": 0.00015973261393032563,
      "loss": 0.7348,
      "step": 1990
    },
    {
      "epoch": 0.2974305348072901,
      "grad_norm": 0.1446111500263214,
      "learning_rate": 0.00015969491205581132,
      "loss": 0.6682,
      "step": 1991
    },
    {
      "epoch": 0.29757992231849417,
      "grad_norm": 0.12944035232067108,
      "learning_rate": 0.0001596571969937454,
      "loss": 0.8242,
      "step": 1992
    },
    {
      "epoch": 0.29772930982969825,
      "grad_norm": 0.15892425179481506,
      "learning_rate": 0.00015961946875245973,
      "loss": 0.6055,
      "step": 1993
    },
    {
      "epoch": 0.2978786973409023,
      "grad_norm": 0.13054129481315613,
      "learning_rate": 0.0001595817273402891,
      "loss": 0.5977,
      "step": 1994
    },
    {
      "epoch": 0.29802808485210636,
      "grad_norm": 0.2998169958591461,
      "learning_rate": 0.0001595439727655712,
      "loss": 1.0095,
      "step": 1995
    },
    {
      "epoch": 0.29817747236331044,
      "grad_norm": 0.11575822532176971,
      "learning_rate": 0.00015950620503664658,
      "loss": 0.831,
      "step": 1996
    },
    {
      "epoch": 0.2983268598745145,
      "grad_norm": 0.176680326461792,
      "learning_rate": 0.0001594684241618587,
      "loss": 0.8144,
      "step": 1997
    },
    {
      "epoch": 0.29847624738571854,
      "grad_norm": 0.2667008936405182,
      "learning_rate": 0.00015943063014955402,
      "loss": 1.1887,
      "step": 1998
    },
    {
      "epoch": 0.2986256348969226,
      "grad_norm": 0.2525520324707031,
      "learning_rate": 0.0001593928230080818,
      "loss": 0.7398,
      "step": 1999
    },
    {
      "epoch": 0.2987750224081267,
      "grad_norm": 0.20424726605415344,
      "learning_rate": 0.00015935500274579426,
      "loss": 0.9998,
      "step": 2000
    },
    {
      "epoch": 0.2989244099193307,
      "grad_norm": 0.24848030507564545,
      "learning_rate": 0.00015931716937104645,
      "loss": 0.9428,
      "step": 2001
    },
    {
      "epoch": 0.2990737974305348,
      "grad_norm": 0.1705709546804428,
      "learning_rate": 0.00015927932289219642,
      "loss": 0.7016,
      "step": 2002
    },
    {
      "epoch": 0.2992231849417389,
      "grad_norm": 0.20898346602916718,
      "learning_rate": 0.000159241463317605,
      "loss": 1.2988,
      "step": 2003
    },
    {
      "epoch": 0.2993725724529429,
      "grad_norm": 0.1826796531677246,
      "learning_rate": 0.00015920359065563604,
      "loss": 0.8184,
      "step": 2004
    },
    {
      "epoch": 0.299521959964147,
      "grad_norm": 0.12231456488370895,
      "learning_rate": 0.00015916570491465612,
      "loss": 0.8552,
      "step": 2005
    },
    {
      "epoch": 0.29967134747535107,
      "grad_norm": 0.10954627394676208,
      "learning_rate": 0.0001591278061030349,
      "loss": 0.5853,
      "step": 2006
    },
    {
      "epoch": 0.29982073498655515,
      "grad_norm": 0.17874775826931,
      "learning_rate": 0.00015908989422914473,
      "loss": 0.7906,
      "step": 2007
    },
    {
      "epoch": 0.2999701224977592,
      "grad_norm": 0.12277006357908249,
      "learning_rate": 0.00015905196930136097,
      "loss": 0.794,
      "step": 2008
    },
    {
      "epoch": 0.30011951000896325,
      "grad_norm": 0.12225458770990372,
      "learning_rate": 0.0001590140313280619,
      "loss": 0.6106,
      "step": 2009
    },
    {
      "epoch": 0.30026889752016733,
      "grad_norm": 0.16903723776340485,
      "learning_rate": 0.0001589760803176286,
      "loss": 0.6282,
      "step": 2010
    },
    {
      "epoch": 0.30041828503137136,
      "grad_norm": 0.20449140667915344,
      "learning_rate": 0.00015893811627844498,
      "loss": 0.7326,
      "step": 2011
    },
    {
      "epoch": 0.30056767254257544,
      "grad_norm": 0.26871007680892944,
      "learning_rate": 0.00015890013921889795,
      "loss": 0.8725,
      "step": 2012
    },
    {
      "epoch": 0.3007170600537795,
      "grad_norm": 0.11706727743148804,
      "learning_rate": 0.0001588621491473772,
      "loss": 0.4953,
      "step": 2013
    },
    {
      "epoch": 0.30086644756498354,
      "grad_norm": 0.14688648283481598,
      "learning_rate": 0.00015882414607227546,
      "loss": 0.4496,
      "step": 2014
    },
    {
      "epoch": 0.3010158350761876,
      "grad_norm": 0.126808762550354,
      "learning_rate": 0.00015878613000198808,
      "loss": 0.5907,
      "step": 2015
    },
    {
      "epoch": 0.3011652225873917,
      "grad_norm": 0.22014611959457397,
      "learning_rate": 0.00015874810094491343,
      "loss": 0.5632,
      "step": 2016
    },
    {
      "epoch": 0.3013146100985958,
      "grad_norm": 0.13114488124847412,
      "learning_rate": 0.00015871005890945279,
      "loss": 0.6928,
      "step": 2017
    },
    {
      "epoch": 0.3014639976097998,
      "grad_norm": 0.12499664723873138,
      "learning_rate": 0.00015867200390401023,
      "loss": 0.5916,
      "step": 2018
    },
    {
      "epoch": 0.3016133851210039,
      "grad_norm": 0.13385756313800812,
      "learning_rate": 0.00015863393593699269,
      "loss": 0.7382,
      "step": 2019
    },
    {
      "epoch": 0.30176277263220797,
      "grad_norm": 0.14371304214000702,
      "learning_rate": 0.00015859585501681,
      "loss": 1.0358,
      "step": 2020
    },
    {
      "epoch": 0.301912160143412,
      "grad_norm": 0.16064706444740295,
      "learning_rate": 0.00015855776115187474,
      "loss": 0.4691,
      "step": 2021
    },
    {
      "epoch": 0.30206154765461607,
      "grad_norm": 0.17672288417816162,
      "learning_rate": 0.00015851965435060262,
      "loss": 0.6921,
      "step": 2022
    },
    {
      "epoch": 0.30221093516582015,
      "grad_norm": 0.15828637778759003,
      "learning_rate": 0.00015848153462141192,
      "loss": 0.7354,
      "step": 2023
    },
    {
      "epoch": 0.3023603226770242,
      "grad_norm": 0.29658231139183044,
      "learning_rate": 0.00015844340197272393,
      "loss": 0.8949,
      "step": 2024
    },
    {
      "epoch": 0.30250971018822825,
      "grad_norm": 0.11581136286258698,
      "learning_rate": 0.00015840525641296275,
      "loss": 0.6845,
      "step": 2025
    },
    {
      "epoch": 0.30265909769943233,
      "grad_norm": 0.15277673304080963,
      "learning_rate": 0.00015836709795055532,
      "loss": 0.7643,
      "step": 2026
    },
    {
      "epoch": 0.3028084852106364,
      "grad_norm": 0.20397962629795074,
      "learning_rate": 0.00015832892659393147,
      "loss": 0.9567,
      "step": 2027
    },
    {
      "epoch": 0.30295787272184044,
      "grad_norm": 0.18316620588302612,
      "learning_rate": 0.0001582907423515239,
      "loss": 0.5847,
      "step": 2028
    },
    {
      "epoch": 0.3031072602330445,
      "grad_norm": 0.17659157514572144,
      "learning_rate": 0.00015825254523176802,
      "loss": 0.5976,
      "step": 2029
    },
    {
      "epoch": 0.3032566477442486,
      "grad_norm": 0.17253398895263672,
      "learning_rate": 0.00015821433524310224,
      "loss": 0.8699,
      "step": 2030
    },
    {
      "epoch": 0.3034060352554526,
      "grad_norm": 0.17953762412071228,
      "learning_rate": 0.00015817611239396772,
      "loss": 1.0307,
      "step": 2031
    },
    {
      "epoch": 0.3035554227666567,
      "grad_norm": 0.1381669044494629,
      "learning_rate": 0.00015813787669280855,
      "loss": 0.6869,
      "step": 2032
    },
    {
      "epoch": 0.3037048102778608,
      "grad_norm": 0.1326492428779602,
      "learning_rate": 0.0001580996281480715,
      "loss": 0.7534,
      "step": 2033
    },
    {
      "epoch": 0.30385419778906486,
      "grad_norm": 0.13373532891273499,
      "learning_rate": 0.00015806136676820639,
      "loss": 0.987,
      "step": 2034
    },
    {
      "epoch": 0.3040035853002689,
      "grad_norm": 0.13426072895526886,
      "learning_rate": 0.0001580230925616657,
      "loss": 0.6961,
      "step": 2035
    },
    {
      "epoch": 0.30415297281147297,
      "grad_norm": 0.15611180663108826,
      "learning_rate": 0.00015798480553690482,
      "loss": 0.7639,
      "step": 2036
    },
    {
      "epoch": 0.30430236032267705,
      "grad_norm": 0.13876785337924957,
      "learning_rate": 0.000157946505702382,
      "loss": 0.7782,
      "step": 2037
    },
    {
      "epoch": 0.30445174783388107,
      "grad_norm": 0.15186084806919098,
      "learning_rate": 0.0001579081930665582,
      "loss": 0.8154,
      "step": 2038
    },
    {
      "epoch": 0.30460113534508515,
      "grad_norm": 0.1986602544784546,
      "learning_rate": 0.00015786986763789736,
      "loss": 0.9731,
      "step": 2039
    },
    {
      "epoch": 0.30475052285628923,
      "grad_norm": 0.17375710606575012,
      "learning_rate": 0.00015783152942486613,
      "loss": 0.7783,
      "step": 2040
    },
    {
      "epoch": 0.30489991036749325,
      "grad_norm": 0.13821902871131897,
      "learning_rate": 0.00015779317843593406,
      "loss": 0.8962,
      "step": 2041
    },
    {
      "epoch": 0.30504929787869733,
      "grad_norm": 0.19690564274787903,
      "learning_rate": 0.0001577548146795735,
      "loss": 0.6165,
      "step": 2042
    },
    {
      "epoch": 0.3051986853899014,
      "grad_norm": 0.34292152523994446,
      "learning_rate": 0.00015771643816425953,
      "loss": 0.9558,
      "step": 2043
    },
    {
      "epoch": 0.3053480729011055,
      "grad_norm": 0.15843629837036133,
      "learning_rate": 0.00015767804889847025,
      "loss": 0.7504,
      "step": 2044
    },
    {
      "epoch": 0.3054974604123095,
      "grad_norm": 0.12701314687728882,
      "learning_rate": 0.00015763964689068636,
      "loss": 0.5962,
      "step": 2045
    },
    {
      "epoch": 0.3056468479235136,
      "grad_norm": 0.2860795259475708,
      "learning_rate": 0.00015760123214939148,
      "loss": 0.9803,
      "step": 2046
    },
    {
      "epoch": 0.3057962354347177,
      "grad_norm": 0.16724273562431335,
      "learning_rate": 0.0001575628046830721,
      "loss": 0.6602,
      "step": 2047
    },
    {
      "epoch": 0.3059456229459217,
      "grad_norm": 0.15602046251296997,
      "learning_rate": 0.00015752436450021742,
      "loss": 0.8576,
      "step": 2048
    },
    {
      "epoch": 0.3060950104571258,
      "grad_norm": 0.18656249344348907,
      "learning_rate": 0.00015748591160931946,
      "loss": 0.6884,
      "step": 2049
    },
    {
      "epoch": 0.30624439796832986,
      "grad_norm": 0.10958966612815857,
      "learning_rate": 0.0001574474460188731,
      "loss": 0.6705,
      "step": 2050
    },
    {
      "epoch": 0.3063937854795339,
      "grad_norm": 0.31523603200912476,
      "learning_rate": 0.00015740896773737601,
      "loss": 1.096,
      "step": 2051
    },
    {
      "epoch": 0.30654317299073797,
      "grad_norm": 0.20349806547164917,
      "learning_rate": 0.00015737047677332863,
      "loss": 0.9721,
      "step": 2052
    },
    {
      "epoch": 0.30669256050194205,
      "grad_norm": 0.18805855512619019,
      "learning_rate": 0.00015733197313523423,
      "loss": 0.5629,
      "step": 2053
    },
    {
      "epoch": 0.3068419480131461,
      "grad_norm": 0.19701740145683289,
      "learning_rate": 0.0001572934568315989,
      "loss": 0.526,
      "step": 2054
    },
    {
      "epoch": 0.30699133552435015,
      "grad_norm": 0.10095413029193878,
      "learning_rate": 0.00015725492787093144,
      "loss": 0.4077,
      "step": 2055
    },
    {
      "epoch": 0.30714072303555423,
      "grad_norm": 0.11596342921257019,
      "learning_rate": 0.00015721638626174354,
      "loss": 0.7053,
      "step": 2056
    },
    {
      "epoch": 0.3072901105467583,
      "grad_norm": 0.13187706470489502,
      "learning_rate": 0.00015717783201254967,
      "loss": 0.6944,
      "step": 2057
    },
    {
      "epoch": 0.30743949805796233,
      "grad_norm": 0.1936214119195938,
      "learning_rate": 0.00015713926513186702,
      "loss": 0.9514,
      "step": 2058
    },
    {
      "epoch": 0.3075888855691664,
      "grad_norm": 0.12927375733852386,
      "learning_rate": 0.00015710068562821568,
      "loss": 0.8447,
      "step": 2059
    },
    {
      "epoch": 0.3077382730803705,
      "grad_norm": 0.2957967221736908,
      "learning_rate": 0.00015706209351011848,
      "loss": 0.7323,
      "step": 2060
    },
    {
      "epoch": 0.3078876605915745,
      "grad_norm": 0.29521241784095764,
      "learning_rate": 0.00015702348878610103,
      "loss": 0.9674,
      "step": 2061
    },
    {
      "epoch": 0.3080370481027786,
      "grad_norm": 0.10689933598041534,
      "learning_rate": 0.00015698487146469163,
      "loss": 0.591,
      "step": 2062
    },
    {
      "epoch": 0.3081864356139827,
      "grad_norm": 0.14676429331302643,
      "learning_rate": 0.00015694624155442158,
      "loss": 0.5536,
      "step": 2063
    },
    {
      "epoch": 0.30833582312518676,
      "grad_norm": 0.2026292085647583,
      "learning_rate": 0.0001569075990638248,
      "loss": 0.9233,
      "step": 2064
    },
    {
      "epoch": 0.3084852106363908,
      "grad_norm": 0.14372330904006958,
      "learning_rate": 0.00015686894400143798,
      "loss": 0.8334,
      "step": 2065
    },
    {
      "epoch": 0.30863459814759486,
      "grad_norm": 0.20169852674007416,
      "learning_rate": 0.00015683027637580066,
      "loss": 0.717,
      "step": 2066
    },
    {
      "epoch": 0.30878398565879894,
      "grad_norm": 0.1340600699186325,
      "learning_rate": 0.00015679159619545518,
      "loss": 0.6994,
      "step": 2067
    },
    {
      "epoch": 0.30893337317000297,
      "grad_norm": 0.6489109396934509,
      "learning_rate": 0.00015675290346894657,
      "loss": 1.4394,
      "step": 2068
    },
    {
      "epoch": 0.30908276068120705,
      "grad_norm": 0.18207131326198578,
      "learning_rate": 0.00015671419820482264,
      "loss": 0.8153,
      "step": 2069
    },
    {
      "epoch": 0.3092321481924111,
      "grad_norm": 0.12122584134340286,
      "learning_rate": 0.00015667548041163406,
      "loss": 0.8143,
      "step": 2070
    },
    {
      "epoch": 0.30938153570361515,
      "grad_norm": 0.1290636658668518,
      "learning_rate": 0.00015663675009793413,
      "loss": 0.7167,
      "step": 2071
    },
    {
      "epoch": 0.30953092321481923,
      "grad_norm": 0.13312415778636932,
      "learning_rate": 0.00015659800727227903,
      "loss": 0.917,
      "step": 2072
    },
    {
      "epoch": 0.3096803107260233,
      "grad_norm": 0.15888121724128723,
      "learning_rate": 0.00015655925194322764,
      "loss": 0.639,
      "step": 2073
    },
    {
      "epoch": 0.3098296982372274,
      "grad_norm": 0.1925979107618332,
      "learning_rate": 0.00015652048411934167,
      "loss": 0.5766,
      "step": 2074
    },
    {
      "epoch": 0.3099790857484314,
      "grad_norm": 0.14733384549617767,
      "learning_rate": 0.0001564817038091855,
      "loss": 0.655,
      "step": 2075
    },
    {
      "epoch": 0.3101284732596355,
      "grad_norm": 0.1112808883190155,
      "learning_rate": 0.00015644291102132635,
      "loss": 0.6861,
      "step": 2076
    },
    {
      "epoch": 0.3102778607708396,
      "grad_norm": 0.19479699432849884,
      "learning_rate": 0.00015640410576433416,
      "loss": 0.9946,
      "step": 2077
    },
    {
      "epoch": 0.3104272482820436,
      "grad_norm": 0.1497500091791153,
      "learning_rate": 0.0001563652880467816,
      "loss": 0.8795,
      "step": 2078
    },
    {
      "epoch": 0.3105766357932477,
      "grad_norm": 0.22533303499221802,
      "learning_rate": 0.0001563264578772441,
      "loss": 0.773,
      "step": 2079
    },
    {
      "epoch": 0.31072602330445176,
      "grad_norm": 0.1410921961069107,
      "learning_rate": 0.00015628761526429992,
      "loss": 0.7418,
      "step": 2080
    },
    {
      "epoch": 0.3108754108156558,
      "grad_norm": 0.16990308463573456,
      "learning_rate": 0.00015624876021652997,
      "loss": 0.744,
      "step": 2081
    },
    {
      "epoch": 0.31102479832685986,
      "grad_norm": 0.3026455044746399,
      "learning_rate": 0.00015620989274251797,
      "loss": 1.2825,
      "step": 2082
    },
    {
      "epoch": 0.31117418583806394,
      "grad_norm": 0.16028766334056854,
      "learning_rate": 0.00015617101285085033,
      "loss": 0.6529,
      "step": 2083
    },
    {
      "epoch": 0.311323573349268,
      "grad_norm": 0.16228477656841278,
      "learning_rate": 0.00015613212055011624,
      "loss": 0.7174,
      "step": 2084
    },
    {
      "epoch": 0.31147296086047205,
      "grad_norm": 0.23121260106563568,
      "learning_rate": 0.00015609321584890765,
      "loss": 1.2887,
      "step": 2085
    },
    {
      "epoch": 0.31162234837167613,
      "grad_norm": 0.22243498265743256,
      "learning_rate": 0.0001560542987558192,
      "loss": 0.7668,
      "step": 2086
    },
    {
      "epoch": 0.3117717358828802,
      "grad_norm": 0.13863208889961243,
      "learning_rate": 0.0001560153692794483,
      "loss": 0.8454,
      "step": 2087
    },
    {
      "epoch": 0.31192112339408423,
      "grad_norm": 0.1197424978017807,
      "learning_rate": 0.00015597642742839506,
      "loss": 0.6192,
      "step": 2088
    },
    {
      "epoch": 0.3120705109052883,
      "grad_norm": 0.11627914756536484,
      "learning_rate": 0.0001559374732112624,
      "loss": 0.7492,
      "step": 2089
    },
    {
      "epoch": 0.3122198984164924,
      "grad_norm": 0.1267491579055786,
      "learning_rate": 0.00015589850663665593,
      "loss": 0.4508,
      "step": 2090
    },
    {
      "epoch": 0.31236928592769647,
      "grad_norm": 0.296199768781662,
      "learning_rate": 0.00015585952771318394,
      "loss": 0.7639,
      "step": 2091
    },
    {
      "epoch": 0.3125186734389005,
      "grad_norm": 0.14849886298179626,
      "learning_rate": 0.0001558205364494575,
      "loss": 0.6327,
      "step": 2092
    },
    {
      "epoch": 0.3126680609501046,
      "grad_norm": 0.3462176024913788,
      "learning_rate": 0.00015578153285409043,
      "loss": 0.9845,
      "step": 2093
    },
    {
      "epoch": 0.31281744846130866,
      "grad_norm": 0.18521694839000702,
      "learning_rate": 0.0001557425169356992,
      "loss": 0.7221,
      "step": 2094
    },
    {
      "epoch": 0.3129668359725127,
      "grad_norm": 0.30432364344596863,
      "learning_rate": 0.00015570348870290305,
      "loss": 1.1774,
      "step": 2095
    },
    {
      "epoch": 0.31311622348371676,
      "grad_norm": 0.18962708115577698,
      "learning_rate": 0.000155664448164324,
      "loss": 0.9006,
      "step": 2096
    },
    {
      "epoch": 0.31326561099492084,
      "grad_norm": 0.16737207770347595,
      "learning_rate": 0.00015562539532858663,
      "loss": 0.6755,
      "step": 2097
    },
    {
      "epoch": 0.31341499850612486,
      "grad_norm": 0.2279043197631836,
      "learning_rate": 0.00015558633020431835,
      "loss": 0.5415,
      "step": 2098
    },
    {
      "epoch": 0.31356438601732894,
      "grad_norm": 0.16743901371955872,
      "learning_rate": 0.00015554725280014935,
      "loss": 0.9926,
      "step": 2099
    },
    {
      "epoch": 0.313713773528533,
      "grad_norm": 0.19655898213386536,
      "learning_rate": 0.00015550816312471234,
      "loss": 0.9702,
      "step": 2100
    },
    {
      "epoch": 0.3138631610397371,
      "grad_norm": 0.12318786978721619,
      "learning_rate": 0.0001554690611866429,
      "loss": 0.826,
      "step": 2101
    },
    {
      "epoch": 0.31401254855094113,
      "grad_norm": 0.1411537528038025,
      "learning_rate": 0.00015542994699457925,
      "loss": 0.5781,
      "step": 2102
    },
    {
      "epoch": 0.3141619360621452,
      "grad_norm": 0.21302370727062225,
      "learning_rate": 0.00015539082055716235,
      "loss": 0.8996,
      "step": 2103
    },
    {
      "epoch": 0.3143113235733493,
      "grad_norm": 0.12233003228902817,
      "learning_rate": 0.00015535168188303585,
      "loss": 0.8194,
      "step": 2104
    },
    {
      "epoch": 0.3144607110845533,
      "grad_norm": 0.14661891758441925,
      "learning_rate": 0.00015531253098084608,
      "loss": 0.8645,
      "step": 2105
    },
    {
      "epoch": 0.3146100985957574,
      "grad_norm": 0.20048436522483826,
      "learning_rate": 0.00015527336785924213,
      "loss": 0.7498,
      "step": 2106
    },
    {
      "epoch": 0.3147594861069615,
      "grad_norm": 0.11773763597011566,
      "learning_rate": 0.00015523419252687573,
      "loss": 0.5234,
      "step": 2107
    },
    {
      "epoch": 0.3149088736181655,
      "grad_norm": 0.1294790655374527,
      "learning_rate": 0.00015519500499240133,
      "loss": 0.6089,
      "step": 2108
    },
    {
      "epoch": 0.3150582611293696,
      "grad_norm": 0.17981868982315063,
      "learning_rate": 0.00015515580526447611,
      "loss": 0.7482,
      "step": 2109
    },
    {
      "epoch": 0.31520764864057366,
      "grad_norm": 0.13476106524467468,
      "learning_rate": 0.00015511659335175985,
      "loss": 0.9713,
      "step": 2110
    },
    {
      "epoch": 0.31535703615177774,
      "grad_norm": 0.19272902607917786,
      "learning_rate": 0.00015507736926291513,
      "loss": 0.5962,
      "step": 2111
    },
    {
      "epoch": 0.31550642366298176,
      "grad_norm": 0.13049067556858063,
      "learning_rate": 0.00015503813300660717,
      "loss": 0.5775,
      "step": 2112
    },
    {
      "epoch": 0.31565581117418584,
      "grad_norm": 0.33260369300842285,
      "learning_rate": 0.00015499888459150386,
      "loss": 0.9088,
      "step": 2113
    },
    {
      "epoch": 0.3158051986853899,
      "grad_norm": 0.12220057100057602,
      "learning_rate": 0.0001549596240262758,
      "loss": 0.6327,
      "step": 2114
    },
    {
      "epoch": 0.31595458619659395,
      "grad_norm": 1.2551153898239136,
      "learning_rate": 0.0001549203513195963,
      "loss": 1.5567,
      "step": 2115
    },
    {
      "epoch": 0.316103973707798,
      "grad_norm": 0.17383141815662384,
      "learning_rate": 0.00015488106648014127,
      "loss": 0.5836,
      "step": 2116
    },
    {
      "epoch": 0.3162533612190021,
      "grad_norm": 0.11264350265264511,
      "learning_rate": 0.0001548417695165894,
      "loss": 0.6948,
      "step": 2117
    },
    {
      "epoch": 0.31640274873020613,
      "grad_norm": 0.14316920936107635,
      "learning_rate": 0.00015480246043762198,
      "loss": 0.7636,
      "step": 2118
    },
    {
      "epoch": 0.3165521362414102,
      "grad_norm": 0.1528894603252411,
      "learning_rate": 0.00015476313925192304,
      "loss": 0.7999,
      "step": 2119
    },
    {
      "epoch": 0.3167015237526143,
      "grad_norm": 0.15061505138874054,
      "learning_rate": 0.00015472380596817922,
      "loss": 0.6164,
      "step": 2120
    },
    {
      "epoch": 0.31685091126381837,
      "grad_norm": 0.16563770174980164,
      "learning_rate": 0.0001546844605950799,
      "loss": 0.8071,
      "step": 2121
    },
    {
      "epoch": 0.3170002987750224,
      "grad_norm": 0.1965642124414444,
      "learning_rate": 0.0001546451031413171,
      "loss": 1.0775,
      "step": 2122
    },
    {
      "epoch": 0.3171496862862265,
      "grad_norm": 0.15104445815086365,
      "learning_rate": 0.00015460573361558542,
      "loss": 0.4892,
      "step": 2123
    },
    {
      "epoch": 0.31729907379743055,
      "grad_norm": 0.17511765658855438,
      "learning_rate": 0.0001545663520265823,
      "loss": 0.9075,
      "step": 2124
    },
    {
      "epoch": 0.3174484613086346,
      "grad_norm": 0.18730461597442627,
      "learning_rate": 0.00015452695838300774,
      "loss": 0.7765,
      "step": 2125
    },
    {
      "epoch": 0.31759784881983866,
      "grad_norm": 0.12714232504367828,
      "learning_rate": 0.00015448755269356442,
      "loss": 0.7867,
      "step": 2126
    },
    {
      "epoch": 0.31774723633104274,
      "grad_norm": 0.16589601337909698,
      "learning_rate": 0.00015444813496695768,
      "loss": 0.9272,
      "step": 2127
    },
    {
      "epoch": 0.31789662384224676,
      "grad_norm": 0.116688072681427,
      "learning_rate": 0.00015440870521189547,
      "loss": 0.7728,
      "step": 2128
    },
    {
      "epoch": 0.31804601135345084,
      "grad_norm": 0.1571618765592575,
      "learning_rate": 0.0001543692634370885,
      "loss": 0.5823,
      "step": 2129
    },
    {
      "epoch": 0.3181953988646549,
      "grad_norm": 0.14393973350524902,
      "learning_rate": 0.00015432980965125008,
      "loss": 0.725,
      "step": 2130
    },
    {
      "epoch": 0.318344786375859,
      "grad_norm": 0.1295754611492157,
      "learning_rate": 0.0001542903438630962,
      "loss": 0.8236,
      "step": 2131
    },
    {
      "epoch": 0.318494173887063,
      "grad_norm": 0.15953676402568817,
      "learning_rate": 0.0001542508660813454,
      "loss": 0.7303,
      "step": 2132
    },
    {
      "epoch": 0.3186435613982671,
      "grad_norm": 0.11740988492965698,
      "learning_rate": 0.00015421137631471901,
      "loss": 0.6903,
      "step": 2133
    },
    {
      "epoch": 0.3187929489094712,
      "grad_norm": 0.16701951622962952,
      "learning_rate": 0.00015417187457194092,
      "loss": 0.4059,
      "step": 2134
    },
    {
      "epoch": 0.3189423364206752,
      "grad_norm": 0.12519478797912598,
      "learning_rate": 0.00015413236086173773,
      "loss": 0.7466,
      "step": 2135
    },
    {
      "epoch": 0.3190917239318793,
      "grad_norm": 0.39568090438842773,
      "learning_rate": 0.00015409283519283857,
      "loss": 1.1699,
      "step": 2136
    },
    {
      "epoch": 0.31924111144308337,
      "grad_norm": 0.13197210431098938,
      "learning_rate": 0.00015405329757397536,
      "loss": 0.7426,
      "step": 2137
    },
    {
      "epoch": 0.3193904989542874,
      "grad_norm": 0.26793086528778076,
      "learning_rate": 0.00015401374801388254,
      "loss": 0.8037,
      "step": 2138
    },
    {
      "epoch": 0.3195398864654915,
      "grad_norm": 0.1415427178144455,
      "learning_rate": 0.00015397418652129727,
      "loss": 0.7311,
      "step": 2139
    },
    {
      "epoch": 0.31968927397669555,
      "grad_norm": 0.29071441292762756,
      "learning_rate": 0.00015393461310495926,
      "loss": 0.7324,
      "step": 2140
    },
    {
      "epoch": 0.31983866148789963,
      "grad_norm": 0.13702036440372467,
      "learning_rate": 0.00015389502777361097,
      "loss": 0.6265,
      "step": 2141
    },
    {
      "epoch": 0.31998804899910366,
      "grad_norm": 0.20132002234458923,
      "learning_rate": 0.0001538554305359974,
      "loss": 0.6028,
      "step": 2142
    },
    {
      "epoch": 0.32013743651030774,
      "grad_norm": 0.18386778235435486,
      "learning_rate": 0.00015381582140086615,
      "loss": 0.4095,
      "step": 2143
    },
    {
      "epoch": 0.3202868240215118,
      "grad_norm": 0.12306158244609833,
      "learning_rate": 0.00015377620037696757,
      "loss": 0.6053,
      "step": 2144
    },
    {
      "epoch": 0.32043621153271584,
      "grad_norm": 0.1582852452993393,
      "learning_rate": 0.00015373656747305457,
      "loss": 0.6819,
      "step": 2145
    },
    {
      "epoch": 0.3205855990439199,
      "grad_norm": 0.18054120242595673,
      "learning_rate": 0.00015369692269788266,
      "loss": 0.6782,
      "step": 2146
    },
    {
      "epoch": 0.320734986555124,
      "grad_norm": 1.1181217432022095,
      "learning_rate": 0.00015365726606021,
      "loss": 2.4661,
      "step": 2147
    },
    {
      "epoch": 0.3208843740663281,
      "grad_norm": 0.24032537639141083,
      "learning_rate": 0.0001536175975687974,
      "loss": 1.0103,
      "step": 2148
    },
    {
      "epoch": 0.3210337615775321,
      "grad_norm": 0.19975623488426208,
      "learning_rate": 0.00015357791723240825,
      "loss": 0.9478,
      "step": 2149
    },
    {
      "epoch": 0.3211831490887362,
      "grad_norm": 0.14880773425102234,
      "learning_rate": 0.00015353822505980854,
      "loss": 0.6756,
      "step": 2150
    },
    {
      "epoch": 0.32133253659994027,
      "grad_norm": 0.14271578192710876,
      "learning_rate": 0.0001534985210597669,
      "loss": 0.4425,
      "step": 2151
    },
    {
      "epoch": 0.3214819241111443,
      "grad_norm": 0.11834029108285904,
      "learning_rate": 0.00015345880524105462,
      "loss": 0.7266,
      "step": 2152
    },
    {
      "epoch": 0.32163131162234837,
      "grad_norm": 0.20353318750858307,
      "learning_rate": 0.00015341907761244549,
      "loss": 0.7732,
      "step": 2153
    },
    {
      "epoch": 0.32178069913355245,
      "grad_norm": 0.15648643672466278,
      "learning_rate": 0.00015337933818271597,
      "loss": 0.9631,
      "step": 2154
    },
    {
      "epoch": 0.3219300866447565,
      "grad_norm": 0.12701861560344696,
      "learning_rate": 0.00015333958696064518,
      "loss": 0.7171,
      "step": 2155
    },
    {
      "epoch": 0.32207947415596055,
      "grad_norm": 0.14326916635036469,
      "learning_rate": 0.00015329982395501478,
      "loss": 0.5361,
      "step": 2156
    },
    {
      "epoch": 0.32222886166716463,
      "grad_norm": 0.14299695193767548,
      "learning_rate": 0.0001532600491746091,
      "loss": 0.9336,
      "step": 2157
    },
    {
      "epoch": 0.3223782491783687,
      "grad_norm": 0.16783787310123444,
      "learning_rate": 0.00015322026262821488,
      "loss": 0.7901,
      "step": 2158
    },
    {
      "epoch": 0.32252763668957274,
      "grad_norm": 0.28383246064186096,
      "learning_rate": 0.00015318046432462171,
      "loss": 1.201,
      "step": 2159
    },
    {
      "epoch": 0.3226770242007768,
      "grad_norm": 0.15813380479812622,
      "learning_rate": 0.00015314065427262166,
      "loss": 0.6809,
      "step": 2160
    },
    {
      "epoch": 0.3228264117119809,
      "grad_norm": 0.1553335040807724,
      "learning_rate": 0.00015310083248100935,
      "loss": 0.9311,
      "step": 2161
    },
    {
      "epoch": 0.3229757992231849,
      "grad_norm": 0.15257221460342407,
      "learning_rate": 0.00015306099895858206,
      "loss": 0.9086,
      "step": 2162
    },
    {
      "epoch": 0.323125186734389,
      "grad_norm": 0.10464900732040405,
      "learning_rate": 0.00015302115371413968,
      "loss": 0.5379,
      "step": 2163
    },
    {
      "epoch": 0.3232745742455931,
      "grad_norm": 0.2939471900463104,
      "learning_rate": 0.00015298129675648462,
      "loss": 0.9939,
      "step": 2164
    },
    {
      "epoch": 0.3234239617567971,
      "grad_norm": 0.1119488924741745,
      "learning_rate": 0.00015294142809442197,
      "loss": 0.6316,
      "step": 2165
    },
    {
      "epoch": 0.3235733492680012,
      "grad_norm": 0.13687579333782196,
      "learning_rate": 0.00015290154773675923,
      "loss": 0.7862,
      "step": 2166
    },
    {
      "epoch": 0.32372273677920527,
      "grad_norm": 0.3407014012336731,
      "learning_rate": 0.00015286165569230675,
      "loss": 0.8528,
      "step": 2167
    },
    {
      "epoch": 0.32387212429040935,
      "grad_norm": 0.16920167207717896,
      "learning_rate": 0.00015282175196987721,
      "loss": 0.8838,
      "step": 2168
    },
    {
      "epoch": 0.32402151180161337,
      "grad_norm": 0.1541024148464203,
      "learning_rate": 0.000152781836578286,
      "loss": 0.8387,
      "step": 2169
    },
    {
      "epoch": 0.32417089931281745,
      "grad_norm": 0.15079593658447266,
      "learning_rate": 0.00015274190952635106,
      "loss": 0.7809,
      "step": 2170
    },
    {
      "epoch": 0.32432028682402153,
      "grad_norm": 0.15514825284481049,
      "learning_rate": 0.00015270197082289293,
      "loss": 0.6239,
      "step": 2171
    },
    {
      "epoch": 0.32446967433522556,
      "grad_norm": 0.13850729167461395,
      "learning_rate": 0.00015266202047673467,
      "loss": 0.7308,
      "step": 2172
    },
    {
      "epoch": 0.32461906184642964,
      "grad_norm": 0.11541914194822311,
      "learning_rate": 0.00015262205849670202,
      "loss": 0.7403,
      "step": 2173
    },
    {
      "epoch": 0.3247684493576337,
      "grad_norm": 0.1851481944322586,
      "learning_rate": 0.00015258208489162312,
      "loss": 0.7634,
      "step": 2174
    },
    {
      "epoch": 0.32491783686883774,
      "grad_norm": 0.16813145577907562,
      "learning_rate": 0.0001525420996703288,
      "loss": 1.1267,
      "step": 2175
    },
    {
      "epoch": 0.3250672243800418,
      "grad_norm": 0.10599573701620102,
      "learning_rate": 0.00015250210284165246,
      "loss": 0.6155,
      "step": 2176
    },
    {
      "epoch": 0.3252166118912459,
      "grad_norm": 0.12516306340694427,
      "learning_rate": 0.00015246209441443001,
      "loss": 0.8394,
      "step": 2177
    },
    {
      "epoch": 0.32536599940245,
      "grad_norm": 0.15646129846572876,
      "learning_rate": 0.00015242207439749992,
      "loss": 0.8634,
      "step": 2178
    },
    {
      "epoch": 0.325515386913654,
      "grad_norm": 0.20016029477119446,
      "learning_rate": 0.0001523820427997033,
      "loss": 0.7401,
      "step": 2179
    },
    {
      "epoch": 0.3256647744248581,
      "grad_norm": 0.14763489365577698,
      "learning_rate": 0.0001523419996298837,
      "loss": 0.656,
      "step": 2180
    },
    {
      "epoch": 0.32581416193606216,
      "grad_norm": 0.086511991918087,
      "learning_rate": 0.00015230194489688738,
      "loss": 0.3255,
      "step": 2181
    },
    {
      "epoch": 0.3259635494472662,
      "grad_norm": 0.13473878800868988,
      "learning_rate": 0.00015226187860956295,
      "loss": 0.8509,
      "step": 2182
    },
    {
      "epoch": 0.32611293695847027,
      "grad_norm": 0.13440148532390594,
      "learning_rate": 0.0001522218007767618,
      "loss": 0.4161,
      "step": 2183
    },
    {
      "epoch": 0.32626232446967435,
      "grad_norm": 0.17197415232658386,
      "learning_rate": 0.00015218171140733773,
      "loss": 1.0334,
      "step": 2184
    },
    {
      "epoch": 0.32641171198087837,
      "grad_norm": 0.13701216876506805,
      "learning_rate": 0.00015214161051014705,
      "loss": 0.6186,
      "step": 2185
    },
    {
      "epoch": 0.32656109949208245,
      "grad_norm": 0.11823068559169769,
      "learning_rate": 0.00015210149809404875,
      "loss": 0.7333,
      "step": 2186
    },
    {
      "epoch": 0.32671048700328653,
      "grad_norm": 0.16329899430274963,
      "learning_rate": 0.00015206137416790428,
      "loss": 0.6203,
      "step": 2187
    },
    {
      "epoch": 0.3268598745144906,
      "grad_norm": 0.2963871955871582,
      "learning_rate": 0.00015202123874057761,
      "loss": 1.2211,
      "step": 2188
    },
    {
      "epoch": 0.32700926202569464,
      "grad_norm": 0.152197927236557,
      "learning_rate": 0.00015198109182093533,
      "loss": 0.5791,
      "step": 2189
    },
    {
      "epoch": 0.3271586495368987,
      "grad_norm": 0.27296942472457886,
      "learning_rate": 0.00015194093341784655,
      "loss": 0.9555,
      "step": 2190
    },
    {
      "epoch": 0.3273080370481028,
      "grad_norm": 0.13215193152427673,
      "learning_rate": 0.00015190076354018288,
      "loss": 0.9015,
      "step": 2191
    },
    {
      "epoch": 0.3274574245593068,
      "grad_norm": 0.14458350837230682,
      "learning_rate": 0.00015186058219681848,
      "loss": 0.4232,
      "step": 2192
    },
    {
      "epoch": 0.3276068120705109,
      "grad_norm": 0.15759699046611786,
      "learning_rate": 0.00015182038939663003,
      "loss": 0.6486,
      "step": 2193
    },
    {
      "epoch": 0.327756199581715,
      "grad_norm": 0.179714635014534,
      "learning_rate": 0.00015178018514849678,
      "loss": 0.8581,
      "step": 2194
    },
    {
      "epoch": 0.32790558709291906,
      "grad_norm": 0.16566520929336548,
      "learning_rate": 0.00015173996946130048,
      "loss": 0.8443,
      "step": 2195
    },
    {
      "epoch": 0.3280549746041231,
      "grad_norm": 0.11166373640298843,
      "learning_rate": 0.00015169974234392538,
      "loss": 0.4754,
      "step": 2196
    },
    {
      "epoch": 0.32820436211532716,
      "grad_norm": 0.1186353787779808,
      "learning_rate": 0.00015165950380525834,
      "loss": 0.791,
      "step": 2197
    },
    {
      "epoch": 0.32835374962653124,
      "grad_norm": 0.14318035542964935,
      "learning_rate": 0.00015161925385418867,
      "loss": 0.6756,
      "step": 2198
    },
    {
      "epoch": 0.32850313713773527,
      "grad_norm": 0.1903889775276184,
      "learning_rate": 0.0001515789924996082,
      "loss": 1.1436,
      "step": 2199
    },
    {
      "epoch": 0.32865252464893935,
      "grad_norm": 0.5121449828147888,
      "learning_rate": 0.00015153871975041131,
      "loss": 0.8837,
      "step": 2200
    },
    {
      "epoch": 0.32880191216014343,
      "grad_norm": 0.2169773280620575,
      "learning_rate": 0.00015149843561549492,
      "loss": 1.2616,
      "step": 2201
    },
    {
      "epoch": 0.32895129967134745,
      "grad_norm": 0.15341217815876007,
      "learning_rate": 0.00015145814010375841,
      "loss": 0.4713,
      "step": 2202
    },
    {
      "epoch": 0.32910068718255153,
      "grad_norm": 0.12478365749120712,
      "learning_rate": 0.0001514178332241037,
      "loss": 0.7957,
      "step": 2203
    },
    {
      "epoch": 0.3292500746937556,
      "grad_norm": 0.1309196501970291,
      "learning_rate": 0.00015137751498543517,
      "loss": 0.369,
      "step": 2204
    },
    {
      "epoch": 0.3293994622049597,
      "grad_norm": 0.1589868813753128,
      "learning_rate": 0.0001513371853966598,
      "loss": 0.8516,
      "step": 2205
    },
    {
      "epoch": 0.3295488497161637,
      "grad_norm": 0.19178910553455353,
      "learning_rate": 0.00015129684446668713,
      "loss": 0.7398,
      "step": 2206
    },
    {
      "epoch": 0.3296982372273678,
      "grad_norm": 0.18907012045383453,
      "learning_rate": 0.0001512564922044289,
      "loss": 0.6329,
      "step": 2207
    },
    {
      "epoch": 0.3298476247385719,
      "grad_norm": 0.27931201457977295,
      "learning_rate": 0.00015121612861879974,
      "loss": 1.4891,
      "step": 2208
    },
    {
      "epoch": 0.3299970122497759,
      "grad_norm": 0.13590273261070251,
      "learning_rate": 0.00015117575371871656,
      "loss": 0.6954,
      "step": 2209
    },
    {
      "epoch": 0.33014639976098,
      "grad_norm": 0.15948185324668884,
      "learning_rate": 0.00015113536751309878,
      "loss": 0.5663,
      "step": 2210
    },
    {
      "epoch": 0.33029578727218406,
      "grad_norm": 0.18005739152431488,
      "learning_rate": 0.00015109497001086837,
      "loss": 0.6757,
      "step": 2211
    },
    {
      "epoch": 0.3304451747833881,
      "grad_norm": 0.13694941997528076,
      "learning_rate": 0.00015105456122094983,
      "loss": 0.8489,
      "step": 2212
    },
    {
      "epoch": 0.33059456229459216,
      "grad_norm": 0.14397189021110535,
      "learning_rate": 0.00015101414115226998,
      "loss": 0.5567,
      "step": 2213
    },
    {
      "epoch": 0.33074394980579624,
      "grad_norm": 0.1509937047958374,
      "learning_rate": 0.00015097370981375838,
      "loss": 0.8535,
      "step": 2214
    },
    {
      "epoch": 0.3308933373170003,
      "grad_norm": 0.49491196870803833,
      "learning_rate": 0.00015093326721434685,
      "loss": 0.8976,
      "step": 2215
    },
    {
      "epoch": 0.33104272482820435,
      "grad_norm": 0.15288975834846497,
      "learning_rate": 0.0001508928133629699,
      "loss": 0.5157,
      "step": 2216
    },
    {
      "epoch": 0.33119211233940843,
      "grad_norm": 0.1827441155910492,
      "learning_rate": 0.00015085234826856436,
      "loss": 0.4898,
      "step": 2217
    },
    {
      "epoch": 0.3313414998506125,
      "grad_norm": 0.14467212557792664,
      "learning_rate": 0.00015081187194006962,
      "loss": 0.6615,
      "step": 2218
    },
    {
      "epoch": 0.33149088736181653,
      "grad_norm": 0.15302497148513794,
      "learning_rate": 0.00015077138438642756,
      "loss": 0.6549,
      "step": 2219
    },
    {
      "epoch": 0.3316402748730206,
      "grad_norm": 0.22060945630073547,
      "learning_rate": 0.0001507308856165825,
      "loss": 0.8503,
      "step": 2220
    },
    {
      "epoch": 0.3317896623842247,
      "grad_norm": 0.1676500141620636,
      "learning_rate": 0.0001506903756394813,
      "loss": 0.8627,
      "step": 2221
    },
    {
      "epoch": 0.3319390498954287,
      "grad_norm": 0.15056689083576202,
      "learning_rate": 0.00015064985446407321,
      "loss": 0.7991,
      "step": 2222
    },
    {
      "epoch": 0.3320884374066328,
      "grad_norm": 0.1680218130350113,
      "learning_rate": 0.00015060932209931001,
      "loss": 0.7028,
      "step": 2223
    },
    {
      "epoch": 0.3322378249178369,
      "grad_norm": 0.17818184196949005,
      "learning_rate": 0.00015056877855414594,
      "loss": 0.6659,
      "step": 2224
    },
    {
      "epoch": 0.33238721242904096,
      "grad_norm": 0.1500559151172638,
      "learning_rate": 0.00015052822383753776,
      "loss": 0.8635,
      "step": 2225
    },
    {
      "epoch": 0.332536599940245,
      "grad_norm": 0.12771417200565338,
      "learning_rate": 0.00015048765795844457,
      "loss": 0.5247,
      "step": 2226
    },
    {
      "epoch": 0.33268598745144906,
      "grad_norm": 0.12329693138599396,
      "learning_rate": 0.00015044708092582806,
      "loss": 0.7296,
      "step": 2227
    },
    {
      "epoch": 0.33283537496265314,
      "grad_norm": 0.18867084383964539,
      "learning_rate": 0.00015040649274865238,
      "loss": 0.829,
      "step": 2228
    },
    {
      "epoch": 0.33298476247385717,
      "grad_norm": 0.13382841646671295,
      "learning_rate": 0.000150365893435884,
      "loss": 0.9773,
      "step": 2229
    },
    {
      "epoch": 0.33313414998506125,
      "grad_norm": 0.13404740393161774,
      "learning_rate": 0.000150325282996492,
      "loss": 1.0215,
      "step": 2230
    },
    {
      "epoch": 0.3332835374962653,
      "grad_norm": 0.11712568253278732,
      "learning_rate": 0.00015028466143944788,
      "loss": 0.6633,
      "step": 2231
    },
    {
      "epoch": 0.33343292500746935,
      "grad_norm": 0.22773051261901855,
      "learning_rate": 0.00015024402877372562,
      "loss": 0.9991,
      "step": 2232
    },
    {
      "epoch": 0.33358231251867343,
      "grad_norm": 0.1770501434803009,
      "learning_rate": 0.00015020338500830156,
      "loss": 0.7405,
      "step": 2233
    },
    {
      "epoch": 0.3337317000298775,
      "grad_norm": 0.1672230064868927,
      "learning_rate": 0.00015016273015215455,
      "loss": 0.5908,
      "step": 2234
    },
    {
      "epoch": 0.3338810875410816,
      "grad_norm": 0.29762253165245056,
      "learning_rate": 0.00015012206421426591,
      "loss": 1.3059,
      "step": 2235
    },
    {
      "epoch": 0.3340304750522856,
      "grad_norm": 0.12029761075973511,
      "learning_rate": 0.00015008138720361942,
      "loss": 0.5016,
      "step": 2236
    },
    {
      "epoch": 0.3341798625634897,
      "grad_norm": 0.1249702200293541,
      "learning_rate": 0.00015004069912920122,
      "loss": 0.6987,
      "step": 2237
    },
    {
      "epoch": 0.3343292500746938,
      "grad_norm": 0.12574952840805054,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.6811,
      "step": 2238
    },
    {
      "epoch": 0.3344786375858978,
      "grad_norm": 0.13587528467178345,
      "learning_rate": 0.00014995928982500684,
      "loss": 0.798,
      "step": 2239
    },
    {
      "epoch": 0.3346280250971019,
      "grad_norm": 0.1179809719324112,
      "learning_rate": 0.0001499185686132152,
      "loss": 0.6066,
      "step": 2240
    },
    {
      "epoch": 0.33477741260830596,
      "grad_norm": 0.24678409099578857,
      "learning_rate": 0.0001498778363736211,
      "loss": 1.0321,
      "step": 2241
    },
    {
      "epoch": 0.33492680011951,
      "grad_norm": 0.13504454493522644,
      "learning_rate": 0.00014983709311522297,
      "loss": 0.4708,
      "step": 2242
    },
    {
      "epoch": 0.33507618763071406,
      "grad_norm": 0.1156691312789917,
      "learning_rate": 0.00014979633884702151,
      "loss": 0.6969,
      "step": 2243
    },
    {
      "epoch": 0.33522557514191814,
      "grad_norm": 0.13647282123565674,
      "learning_rate": 0.0001497555735780201,
      "loss": 0.5103,
      "step": 2244
    },
    {
      "epoch": 0.3353749626531222,
      "grad_norm": 0.7194836735725403,
      "learning_rate": 0.0001497147973172244,
      "loss": 1.1432,
      "step": 2245
    },
    {
      "epoch": 0.33552435016432625,
      "grad_norm": 0.11568637937307358,
      "learning_rate": 0.00014967401007364255,
      "loss": 0.6271,
      "step": 2246
    },
    {
      "epoch": 0.3356737376755303,
      "grad_norm": 0.13965779542922974,
      "learning_rate": 0.00014963321185628505,
      "loss": 0.7152,
      "step": 2247
    },
    {
      "epoch": 0.3358231251867344,
      "grad_norm": 0.13680095970630646,
      "learning_rate": 0.0001495924026741649,
      "loss": 0.9004,
      "step": 2248
    },
    {
      "epoch": 0.33597251269793843,
      "grad_norm": 0.13491761684417725,
      "learning_rate": 0.00014955158253629754,
      "loss": 0.9181,
      "step": 2249
    },
    {
      "epoch": 0.3361219002091425,
      "grad_norm": 0.17654938995838165,
      "learning_rate": 0.0001495107514517007,
      "loss": 0.6854,
      "step": 2250
    },
    {
      "epoch": 0.3362712877203466,
      "grad_norm": 0.1615394502878189,
      "learning_rate": 0.00014946990942939467,
      "loss": 0.9516,
      "step": 2251
    },
    {
      "epoch": 0.33642067523155067,
      "grad_norm": 0.19500978291034698,
      "learning_rate": 0.00014942905647840206,
      "loss": 0.7857,
      "step": 2252
    },
    {
      "epoch": 0.3365700627427547,
      "grad_norm": 0.15302661061286926,
      "learning_rate": 0.00014938819260774797,
      "loss": 0.6748,
      "step": 2253
    },
    {
      "epoch": 0.3367194502539588,
      "grad_norm": 0.11134330928325653,
      "learning_rate": 0.0001493473178264599,
      "loss": 0.516,
      "step": 2254
    },
    {
      "epoch": 0.33686883776516285,
      "grad_norm": 0.786919891834259,
      "learning_rate": 0.00014930643214356769,
      "loss": 1.6704,
      "step": 2255
    },
    {
      "epoch": 0.3370182252763669,
      "grad_norm": 0.12148578464984894,
      "learning_rate": 0.0001492655355681036,
      "loss": 0.6768,
      "step": 2256
    },
    {
      "epoch": 0.33716761278757096,
      "grad_norm": 0.20351570844650269,
      "learning_rate": 0.00014922462810910243,
      "loss": 0.497,
      "step": 2257
    },
    {
      "epoch": 0.33731700029877504,
      "grad_norm": 0.13617151975631714,
      "learning_rate": 0.00014918370977560122,
      "loss": 0.6556,
      "step": 2258
    },
    {
      "epoch": 0.33746638780997906,
      "grad_norm": 0.17014221847057343,
      "learning_rate": 0.00014914278057663946,
      "loss": 0.7847,
      "step": 2259
    },
    {
      "epoch": 0.33761577532118314,
      "grad_norm": 0.14155036211013794,
      "learning_rate": 0.0001491018405212591,
      "loss": 0.7608,
      "step": 2260
    },
    {
      "epoch": 0.3377651628323872,
      "grad_norm": 0.17408017814159393,
      "learning_rate": 0.00014906088961850445,
      "loss": 0.972,
      "step": 2261
    },
    {
      "epoch": 0.3379145503435913,
      "grad_norm": 0.11812028288841248,
      "learning_rate": 0.00014901992787742219,
      "loss": 0.7282,
      "step": 2262
    },
    {
      "epoch": 0.3380639378547953,
      "grad_norm": 0.1497081220149994,
      "learning_rate": 0.00014897895530706143,
      "loss": 0.7121,
      "step": 2263
    },
    {
      "epoch": 0.3382133253659994,
      "grad_norm": 0.16701044142246246,
      "learning_rate": 0.00014893797191647368,
      "loss": 0.7234,
      "step": 2264
    },
    {
      "epoch": 0.3383627128772035,
      "grad_norm": 0.1282917708158493,
      "learning_rate": 0.00014889697771471277,
      "loss": 0.8029,
      "step": 2265
    },
    {
      "epoch": 0.3385121003884075,
      "grad_norm": 0.14407439529895782,
      "learning_rate": 0.00014885597271083499,
      "loss": 0.6208,
      "step": 2266
    },
    {
      "epoch": 0.3386614878996116,
      "grad_norm": 0.1551835834980011,
      "learning_rate": 0.00014881495691389902,
      "loss": 0.5759,
      "step": 2267
    },
    {
      "epoch": 0.33881087541081567,
      "grad_norm": 0.19830705225467682,
      "learning_rate": 0.00014877393033296585,
      "loss": 0.779,
      "step": 2268
    },
    {
      "epoch": 0.3389602629220197,
      "grad_norm": 0.10387979447841644,
      "learning_rate": 0.00014873289297709895,
      "loss": 0.4921,
      "step": 2269
    },
    {
      "epoch": 0.3391096504332238,
      "grad_norm": 0.22923505306243896,
      "learning_rate": 0.00014869184485536408,
      "loss": 0.5294,
      "step": 2270
    },
    {
      "epoch": 0.33925903794442785,
      "grad_norm": 0.10682381689548492,
      "learning_rate": 0.00014865078597682947,
      "loss": 0.6181,
      "step": 2271
    },
    {
      "epoch": 0.33940842545563193,
      "grad_norm": 0.13146300613880157,
      "learning_rate": 0.00014860971635056563,
      "loss": 0.762,
      "step": 2272
    },
    {
      "epoch": 0.33955781296683596,
      "grad_norm": 0.829555094242096,
      "learning_rate": 0.00014856863598564554,
      "loss": 1.313,
      "step": 2273
    },
    {
      "epoch": 0.33970720047804004,
      "grad_norm": 0.14699624478816986,
      "learning_rate": 0.00014852754489114444,
      "loss": 0.7296,
      "step": 2274
    },
    {
      "epoch": 0.3398565879892441,
      "grad_norm": 0.12859925627708435,
      "learning_rate": 0.00014848644307614007,
      "loss": 0.6295,
      "step": 2275
    },
    {
      "epoch": 0.34000597550044814,
      "grad_norm": 0.16007597744464874,
      "learning_rate": 0.0001484453305497124,
      "loss": 0.9533,
      "step": 2276
    },
    {
      "epoch": 0.3401553630116522,
      "grad_norm": 0.13484562933444977,
      "learning_rate": 0.0001484042073209439,
      "loss": 0.7907,
      "step": 2277
    },
    {
      "epoch": 0.3403047505228563,
      "grad_norm": 0.42613568902015686,
      "learning_rate": 0.00014836307339891934,
      "loss": 0.8884,
      "step": 2278
    },
    {
      "epoch": 0.3404541380340603,
      "grad_norm": 0.12938427925109863,
      "learning_rate": 0.00014832192879272579,
      "loss": 0.7177,
      "step": 2279
    },
    {
      "epoch": 0.3406035255452644,
      "grad_norm": 0.21188515424728394,
      "learning_rate": 0.00014828077351145282,
      "loss": 0.8826,
      "step": 2280
    },
    {
      "epoch": 0.3407529130564685,
      "grad_norm": 0.15950484573841095,
      "learning_rate": 0.00014823960756419222,
      "loss": 0.4308,
      "step": 2281
    },
    {
      "epoch": 0.34090230056767257,
      "grad_norm": 0.7921125292778015,
      "learning_rate": 0.00014819843096003824,
      "loss": 1.8442,
      "step": 2282
    },
    {
      "epoch": 0.3410516880788766,
      "grad_norm": 0.19723160564899445,
      "learning_rate": 0.0001481572437080875,
      "loss": 1.0448,
      "step": 2283
    },
    {
      "epoch": 0.34120107559008067,
      "grad_norm": 0.15879294276237488,
      "learning_rate": 0.0001481160458174388,
      "loss": 0.8834,
      "step": 2284
    },
    {
      "epoch": 0.34135046310128475,
      "grad_norm": 0.18377555906772614,
      "learning_rate": 0.00014807483729719346,
      "loss": 0.8285,
      "step": 2285
    },
    {
      "epoch": 0.3414998506124888,
      "grad_norm": 0.21528570353984833,
      "learning_rate": 0.0001480336181564551,
      "loss": 0.4335,
      "step": 2286
    },
    {
      "epoch": 0.34164923812369286,
      "grad_norm": 1.1315922737121582,
      "learning_rate": 0.0001479923884043297,
      "loss": 1.6964,
      "step": 2287
    },
    {
      "epoch": 0.34179862563489694,
      "grad_norm": 0.15542832016944885,
      "learning_rate": 0.0001479511480499255,
      "loss": 0.709,
      "step": 2288
    },
    {
      "epoch": 0.34194801314610096,
      "grad_norm": 0.247155100107193,
      "learning_rate": 0.0001479098971023532,
      "loss": 0.5991,
      "step": 2289
    },
    {
      "epoch": 0.34209740065730504,
      "grad_norm": 0.16007624566555023,
      "learning_rate": 0.00014786863557072582,
      "loss": 0.6625,
      "step": 2290
    },
    {
      "epoch": 0.3422467881685091,
      "grad_norm": 0.14167092740535736,
      "learning_rate": 0.00014782736346415857,
      "loss": 0.6226,
      "step": 2291
    },
    {
      "epoch": 0.3423961756797132,
      "grad_norm": 0.1698661744594574,
      "learning_rate": 0.00014778608079176923,
      "loss": 0.6817,
      "step": 2292
    },
    {
      "epoch": 0.3425455631909172,
      "grad_norm": 0.20441316068172455,
      "learning_rate": 0.00014774478756267775,
      "loss": 0.7386,
      "step": 2293
    },
    {
      "epoch": 0.3426949507021213,
      "grad_norm": 0.17780330777168274,
      "learning_rate": 0.00014770348378600646,
      "loss": 0.6273,
      "step": 2294
    },
    {
      "epoch": 0.3428443382133254,
      "grad_norm": 0.14582781493663788,
      "learning_rate": 0.00014766216947088002,
      "loss": 0.599,
      "step": 2295
    },
    {
      "epoch": 0.3429937257245294,
      "grad_norm": 0.2473122626543045,
      "learning_rate": 0.00014762084462642539,
      "loss": 0.998,
      "step": 2296
    },
    {
      "epoch": 0.3431431132357335,
      "grad_norm": 0.14406158030033112,
      "learning_rate": 0.00014757950926177192,
      "loss": 0.9101,
      "step": 2297
    },
    {
      "epoch": 0.34329250074693757,
      "grad_norm": 0.18513797223567963,
      "learning_rate": 0.00014753816338605123,
      "loss": 0.5664,
      "step": 2298
    },
    {
      "epoch": 0.3434418882581416,
      "grad_norm": 0.13709723949432373,
      "learning_rate": 0.0001474968070083973,
      "loss": 0.5203,
      "step": 2299
    },
    {
      "epoch": 0.34359127576934567,
      "grad_norm": 0.1416139453649521,
      "learning_rate": 0.00014745544013794636,
      "loss": 0.8693,
      "step": 2300
    },
    {
      "epoch": 0.34374066328054975,
      "grad_norm": 0.20001940429210663,
      "learning_rate": 0.00014741406278383706,
      "loss": 0.7592,
      "step": 2301
    },
    {
      "epoch": 0.34389005079175383,
      "grad_norm": 0.20120392739772797,
      "learning_rate": 0.0001473726749552103,
      "loss": 0.9879,
      "step": 2302
    },
    {
      "epoch": 0.34403943830295786,
      "grad_norm": 0.22447481751441956,
      "learning_rate": 0.0001473312766612093,
      "loss": 0.7953,
      "step": 2303
    },
    {
      "epoch": 0.34418882581416194,
      "grad_norm": 0.13864095509052277,
      "learning_rate": 0.00014728986791097957,
      "loss": 0.6685,
      "step": 2304
    },
    {
      "epoch": 0.344338213325366,
      "grad_norm": 0.18510349094867706,
      "learning_rate": 0.00014724844871366897,
      "loss": 0.5626,
      "step": 2305
    },
    {
      "epoch": 0.34448760083657004,
      "grad_norm": 0.1396680623292923,
      "learning_rate": 0.00014720701907842772,
      "loss": 0.9187,
      "step": 2306
    },
    {
      "epoch": 0.3446369883477741,
      "grad_norm": 0.22750426828861237,
      "learning_rate": 0.00014716557901440822,
      "loss": 0.7216,
      "step": 2307
    },
    {
      "epoch": 0.3447863758589782,
      "grad_norm": 0.14426091313362122,
      "learning_rate": 0.00014712412853076524,
      "loss": 0.5973,
      "step": 2308
    },
    {
      "epoch": 0.3449357633701823,
      "grad_norm": 0.10019828379154205,
      "learning_rate": 0.00014708266763665588,
      "loss": 0.4711,
      "step": 2309
    },
    {
      "epoch": 0.3450851508813863,
      "grad_norm": 0.4579393267631531,
      "learning_rate": 0.00014704119634123948,
      "loss": 1.2181,
      "step": 2310
    },
    {
      "epoch": 0.3452345383925904,
      "grad_norm": 0.1417282521724701,
      "learning_rate": 0.00014699971465367771,
      "loss": 0.905,
      "step": 2311
    },
    {
      "epoch": 0.34538392590379446,
      "grad_norm": 0.19049352407455444,
      "learning_rate": 0.00014695822258313455,
      "loss": 0.467,
      "step": 2312
    },
    {
      "epoch": 0.3455333134149985,
      "grad_norm": 0.18309010565280914,
      "learning_rate": 0.00014691672013877622,
      "loss": 0.8782,
      "step": 2313
    },
    {
      "epoch": 0.34568270092620257,
      "grad_norm": 0.14359761774539948,
      "learning_rate": 0.00014687520732977128,
      "loss": 0.3978,
      "step": 2314
    },
    {
      "epoch": 0.34583208843740665,
      "grad_norm": 0.1837110072374344,
      "learning_rate": 0.0001468336841652906,
      "loss": 0.8672,
      "step": 2315
    },
    {
      "epoch": 0.3459814759486107,
      "grad_norm": 0.14767298102378845,
      "learning_rate": 0.00014679215065450726,
      "loss": 0.85,
      "step": 2316
    },
    {
      "epoch": 0.34613086345981475,
      "grad_norm": 0.2434052675962448,
      "learning_rate": 0.00014675060680659669,
      "loss": 1.0772,
      "step": 2317
    },
    {
      "epoch": 0.34628025097101883,
      "grad_norm": 0.230267733335495,
      "learning_rate": 0.0001467090526307366,
      "loss": 0.449,
      "step": 2318
    },
    {
      "epoch": 0.3464296384822229,
      "grad_norm": 0.259000301361084,
      "learning_rate": 0.00014666748813610699,
      "loss": 1.2701,
      "step": 2319
    },
    {
      "epoch": 0.34657902599342694,
      "grad_norm": 0.12033212929964066,
      "learning_rate": 0.00014662591333189,
      "loss": 0.7961,
      "step": 2320
    },
    {
      "epoch": 0.346728413504631,
      "grad_norm": 0.3120269775390625,
      "learning_rate": 0.0001465843282272703,
      "loss": 1.068,
      "step": 2321
    },
    {
      "epoch": 0.3468778010158351,
      "grad_norm": 0.4133605360984802,
      "learning_rate": 0.0001465427328314346,
      "loss": 1.2725,
      "step": 2322
    },
    {
      "epoch": 0.3470271885270391,
      "grad_norm": 0.1393682211637497,
      "learning_rate": 0.00014650112715357204,
      "loss": 0.7734,
      "step": 2323
    },
    {
      "epoch": 0.3471765760382432,
      "grad_norm": 0.16707074642181396,
      "learning_rate": 0.0001464595112028739,
      "loss": 0.8834,
      "step": 2324
    },
    {
      "epoch": 0.3473259635494473,
      "grad_norm": 0.1558665782213211,
      "learning_rate": 0.00014641788498853393,
      "loss": 0.8371,
      "step": 2325
    },
    {
      "epoch": 0.3474753510606513,
      "grad_norm": 0.2127390205860138,
      "learning_rate": 0.0001463762485197479,
      "loss": 1.0614,
      "step": 2326
    },
    {
      "epoch": 0.3476247385718554,
      "grad_norm": 0.15837760269641876,
      "learning_rate": 0.000146334601805714,
      "loss": 0.788,
      "step": 2327
    },
    {
      "epoch": 0.34777412608305946,
      "grad_norm": 0.16346672177314758,
      "learning_rate": 0.00014629294485563271,
      "loss": 0.7688,
      "step": 2328
    },
    {
      "epoch": 0.34792351359426354,
      "grad_norm": 0.25042709708213806,
      "learning_rate": 0.0001462512776787066,
      "loss": 0.8124,
      "step": 2329
    },
    {
      "epoch": 0.34807290110546757,
      "grad_norm": 0.2147558331489563,
      "learning_rate": 0.00014620960028414074,
      "loss": 0.6709,
      "step": 2330
    },
    {
      "epoch": 0.34822228861667165,
      "grad_norm": 0.12575598061084747,
      "learning_rate": 0.0001461679126811422,
      "loss": 0.6898,
      "step": 2331
    },
    {
      "epoch": 0.34837167612787573,
      "grad_norm": 0.24064159393310547,
      "learning_rate": 0.0001461262148789205,
      "loss": 0.9161,
      "step": 2332
    },
    {
      "epoch": 0.34852106363907975,
      "grad_norm": 0.30165067315101624,
      "learning_rate": 0.00014608450688668732,
      "loss": 0.9826,
      "step": 2333
    },
    {
      "epoch": 0.34867045115028383,
      "grad_norm": 0.2656305730342865,
      "learning_rate": 0.00014604278871365662,
      "loss": 0.5895,
      "step": 2334
    },
    {
      "epoch": 0.3488198386614879,
      "grad_norm": 0.19641396403312683,
      "learning_rate": 0.00014600106036904466,
      "loss": 0.7341,
      "step": 2335
    },
    {
      "epoch": 0.34896922617269194,
      "grad_norm": 0.13855178654193878,
      "learning_rate": 0.0001459593218620698,
      "loss": 0.6434,
      "step": 2336
    },
    {
      "epoch": 0.349118613683896,
      "grad_norm": 0.28205788135528564,
      "learning_rate": 0.0001459175732019528,
      "loss": 0.9504,
      "step": 2337
    },
    {
      "epoch": 0.3492680011951001,
      "grad_norm": 0.22744609415531158,
      "learning_rate": 0.0001458758143979166,
      "loss": 0.7483,
      "step": 2338
    },
    {
      "epoch": 0.3494173887063042,
      "grad_norm": 0.12420664727687836,
      "learning_rate": 0.00014583404545918634,
      "loss": 0.7162,
      "step": 2339
    },
    {
      "epoch": 0.3495667762175082,
      "grad_norm": 0.11175963282585144,
      "learning_rate": 0.00014579226639498946,
      "loss": 0.6747,
      "step": 2340
    },
    {
      "epoch": 0.3497161637287123,
      "grad_norm": 0.3252045810222626,
      "learning_rate": 0.00014575047721455562,
      "loss": 0.8335,
      "step": 2341
    },
    {
      "epoch": 0.34986555123991636,
      "grad_norm": 0.15948821604251862,
      "learning_rate": 0.00014570867792711674,
      "loss": 0.8628,
      "step": 2342
    },
    {
      "epoch": 0.3500149387511204,
      "grad_norm": 0.2466660439968109,
      "learning_rate": 0.0001456668685419069,
      "loss": 1.0494,
      "step": 2343
    },
    {
      "epoch": 0.35016432626232447,
      "grad_norm": 0.17932328581809998,
      "learning_rate": 0.0001456250490681625,
      "loss": 0.8786,
      "step": 2344
    },
    {
      "epoch": 0.35031371377352855,
      "grad_norm": 0.19382989406585693,
      "learning_rate": 0.00014558321951512205,
      "loss": 1.0716,
      "step": 2345
    },
    {
      "epoch": 0.35046310128473257,
      "grad_norm": 0.12419094145298004,
      "learning_rate": 0.00014554137989202643,
      "loss": 0.8633,
      "step": 2346
    },
    {
      "epoch": 0.35061248879593665,
      "grad_norm": 0.15344765782356262,
      "learning_rate": 0.00014549953020811867,
      "loss": 0.8646,
      "step": 2347
    },
    {
      "epoch": 0.35076187630714073,
      "grad_norm": 0.19439557194709778,
      "learning_rate": 0.000145457670472644,
      "loss": 0.5951,
      "step": 2348
    },
    {
      "epoch": 0.3509112638183448,
      "grad_norm": 0.13660065829753876,
      "learning_rate": 0.00014541580069484992,
      "loss": 0.8178,
      "step": 2349
    },
    {
      "epoch": 0.35106065132954883,
      "grad_norm": 0.15537413954734802,
      "learning_rate": 0.00014537392088398608,
      "loss": 0.7985,
      "step": 2350
    },
    {
      "epoch": 0.3512100388407529,
      "grad_norm": 0.22345857322216034,
      "learning_rate": 0.00014533203104930445,
      "loss": 0.7629,
      "step": 2351
    },
    {
      "epoch": 0.351359426351957,
      "grad_norm": 0.2205721139907837,
      "learning_rate": 0.00014529013120005916,
      "loss": 0.99,
      "step": 2352
    },
    {
      "epoch": 0.351508813863161,
      "grad_norm": 0.9867051243782043,
      "learning_rate": 0.00014524822134550651,
      "loss": 1.6639,
      "step": 2353
    },
    {
      "epoch": 0.3516582013743651,
      "grad_norm": 0.1274566650390625,
      "learning_rate": 0.0001452063014949051,
      "loss": 0.8408,
      "step": 2354
    },
    {
      "epoch": 0.3518075888855692,
      "grad_norm": 0.12840072810649872,
      "learning_rate": 0.00014516437165751562,
      "loss": 0.7849,
      "step": 2355
    },
    {
      "epoch": 0.3519569763967732,
      "grad_norm": 0.16408976912498474,
      "learning_rate": 0.0001451224318426011,
      "loss": 0.7865,
      "step": 2356
    },
    {
      "epoch": 0.3521063639079773,
      "grad_norm": 0.24575677514076233,
      "learning_rate": 0.00014508048205942665,
      "loss": 0.6155,
      "step": 2357
    },
    {
      "epoch": 0.35225575141918136,
      "grad_norm": 0.13022038340568542,
      "learning_rate": 0.0001450385223172597,
      "loss": 0.4543,
      "step": 2358
    },
    {
      "epoch": 0.35240513893038544,
      "grad_norm": 0.2615824043750763,
      "learning_rate": 0.0001449965526253698,
      "loss": 0.728,
      "step": 2359
    },
    {
      "epoch": 0.35255452644158947,
      "grad_norm": 0.16605320572853088,
      "learning_rate": 0.0001449545729930287,
      "loss": 0.7351,
      "step": 2360
    },
    {
      "epoch": 0.35270391395279355,
      "grad_norm": 0.2628827691078186,
      "learning_rate": 0.0001449125834295104,
      "loss": 0.8374,
      "step": 2361
    },
    {
      "epoch": 0.3528533014639976,
      "grad_norm": 0.2556038200855255,
      "learning_rate": 0.00014487058394409104,
      "loss": 1.0356,
      "step": 2362
    },
    {
      "epoch": 0.35300268897520165,
      "grad_norm": 0.16808576881885529,
      "learning_rate": 0.00014482857454604895,
      "loss": 0.7134,
      "step": 2363
    },
    {
      "epoch": 0.35315207648640573,
      "grad_norm": 0.13573585450649261,
      "learning_rate": 0.00014478655524466475,
      "loss": 0.5456,
      "step": 2364
    },
    {
      "epoch": 0.3533014639976098,
      "grad_norm": 0.11733470112085342,
      "learning_rate": 0.00014474452604922107,
      "loss": 0.7421,
      "step": 2365
    },
    {
      "epoch": 0.3534508515088139,
      "grad_norm": 0.29591408371925354,
      "learning_rate": 0.00014470248696900285,
      "loss": 1.3908,
      "step": 2366
    },
    {
      "epoch": 0.3536002390200179,
      "grad_norm": 0.30100902915000916,
      "learning_rate": 0.00014466043801329725,
      "loss": 1.0222,
      "step": 2367
    },
    {
      "epoch": 0.353749626531222,
      "grad_norm": 0.1572057455778122,
      "learning_rate": 0.00014461837919139348,
      "loss": 0.6108,
      "step": 2368
    },
    {
      "epoch": 0.3538990140424261,
      "grad_norm": 0.24329088628292084,
      "learning_rate": 0.00014457631051258303,
      "loss": 0.7188,
      "step": 2369
    },
    {
      "epoch": 0.3540484015536301,
      "grad_norm": 0.291996031999588,
      "learning_rate": 0.00014453423198615957,
      "loss": 0.9339,
      "step": 2370
    },
    {
      "epoch": 0.3541977890648342,
      "grad_norm": 0.12791137397289276,
      "learning_rate": 0.00014449214362141882,
      "loss": 0.8645,
      "step": 2371
    },
    {
      "epoch": 0.35434717657603826,
      "grad_norm": 0.3090847134590149,
      "learning_rate": 0.00014445004542765888,
      "loss": 0.7204,
      "step": 2372
    },
    {
      "epoch": 0.3544965640872423,
      "grad_norm": 0.3392612338066101,
      "learning_rate": 0.00014440793741417985,
      "loss": 1.0048,
      "step": 2373
    },
    {
      "epoch": 0.35464595159844636,
      "grad_norm": 0.41094449162483215,
      "learning_rate": 0.00014436581959028405,
      "loss": 0.7095,
      "step": 2374
    },
    {
      "epoch": 0.35479533910965044,
      "grad_norm": 0.12895432114601135,
      "learning_rate": 0.00014432369196527595,
      "loss": 0.8101,
      "step": 2375
    },
    {
      "epoch": 0.3549447266208545,
      "grad_norm": 0.16293783485889435,
      "learning_rate": 0.00014428155454846225,
      "loss": 0.958,
      "step": 2376
    },
    {
      "epoch": 0.35509411413205855,
      "grad_norm": 0.1935635358095169,
      "learning_rate": 0.0001442394073491518,
      "loss": 0.7621,
      "step": 2377
    },
    {
      "epoch": 0.3552435016432626,
      "grad_norm": 0.23022136092185974,
      "learning_rate": 0.0001441972503766555,
      "loss": 0.5687,
      "step": 2378
    },
    {
      "epoch": 0.3553928891544667,
      "grad_norm": 0.12238634377717972,
      "learning_rate": 0.0001441550836402866,
      "loss": 0.8099,
      "step": 2379
    },
    {
      "epoch": 0.35554227666567073,
      "grad_norm": 0.1600809097290039,
      "learning_rate": 0.00014411290714936033,
      "loss": 0.8935,
      "step": 2380
    },
    {
      "epoch": 0.3556916641768748,
      "grad_norm": 0.13895153999328613,
      "learning_rate": 0.00014407072091319415,
      "loss": 0.7407,
      "step": 2381
    },
    {
      "epoch": 0.3558410516880789,
      "grad_norm": 0.23314234614372253,
      "learning_rate": 0.00014402852494110768,
      "loss": 1.2477,
      "step": 2382
    },
    {
      "epoch": 0.3559904391992829,
      "grad_norm": 0.1355046033859253,
      "learning_rate": 0.00014398631924242267,
      "loss": 0.7014,
      "step": 2383
    },
    {
      "epoch": 0.356139826710487,
      "grad_norm": 0.25120407342910767,
      "learning_rate": 0.00014394410382646304,
      "loss": 1.0142,
      "step": 2384
    },
    {
      "epoch": 0.3562892142216911,
      "grad_norm": 0.14794789254665375,
      "learning_rate": 0.00014390187870255483,
      "loss": 0.6787,
      "step": 2385
    },
    {
      "epoch": 0.35643860173289516,
      "grad_norm": 0.1715037226676941,
      "learning_rate": 0.00014385964388002623,
      "loss": 0.6621,
      "step": 2386
    },
    {
      "epoch": 0.3565879892440992,
      "grad_norm": 0.182341530919075,
      "learning_rate": 0.0001438173993682076,
      "loss": 0.8658,
      "step": 2387
    },
    {
      "epoch": 0.35673737675530326,
      "grad_norm": 0.13495273888111115,
      "learning_rate": 0.00014377514517643144,
      "loss": 0.7215,
      "step": 2388
    },
    {
      "epoch": 0.35688676426650734,
      "grad_norm": 0.1517488956451416,
      "learning_rate": 0.00014373288131403232,
      "loss": 0.5074,
      "step": 2389
    },
    {
      "epoch": 0.35703615177771136,
      "grad_norm": 0.23197542130947113,
      "learning_rate": 0.00014369060779034708,
      "loss": 0.7701,
      "step": 2390
    },
    {
      "epoch": 0.35718553928891544,
      "grad_norm": 0.18405930697917938,
      "learning_rate": 0.00014364832461471452,
      "loss": 0.8195,
      "step": 2391
    },
    {
      "epoch": 0.3573349268001195,
      "grad_norm": 0.15993326902389526,
      "learning_rate": 0.00014360603179647567,
      "loss": 0.8224,
      "step": 2392
    },
    {
      "epoch": 0.35748431431132355,
      "grad_norm": 0.2076801210641861,
      "learning_rate": 0.00014356372934497377,
      "loss": 0.7723,
      "step": 2393
    },
    {
      "epoch": 0.3576337018225276,
      "grad_norm": 0.144704669713974,
      "learning_rate": 0.000143521417269554,
      "loss": 0.7462,
      "step": 2394
    },
    {
      "epoch": 0.3577830893337317,
      "grad_norm": 0.1373116672039032,
      "learning_rate": 0.00014347909557956384,
      "loss": 0.6932,
      "step": 2395
    },
    {
      "epoch": 0.3579324768449358,
      "grad_norm": 0.2745577096939087,
      "learning_rate": 0.00014343676428435275,
      "loss": 0.9025,
      "step": 2396
    },
    {
      "epoch": 0.3580818643561398,
      "grad_norm": 0.13737031817436218,
      "learning_rate": 0.00014339442339327246,
      "loss": 0.7861,
      "step": 2397
    },
    {
      "epoch": 0.3582312518673439,
      "grad_norm": 0.1721804141998291,
      "learning_rate": 0.0001433520729156767,
      "loss": 0.5584,
      "step": 2398
    },
    {
      "epoch": 0.35838063937854797,
      "grad_norm": 0.13475766777992249,
      "learning_rate": 0.00014330971286092138,
      "loss": 0.6657,
      "step": 2399
    },
    {
      "epoch": 0.358530026889752,
      "grad_norm": 0.2044338881969452,
      "learning_rate": 0.0001432673432383645,
      "loss": 1.0157,
      "step": 2400
    },
    {
      "epoch": 0.3586794144009561,
      "grad_norm": 0.27857422828674316,
      "learning_rate": 0.00014322496405736617,
      "loss": 0.7371,
      "step": 2401
    },
    {
      "epoch": 0.35882880191216016,
      "grad_norm": 0.13335542380809784,
      "learning_rate": 0.00014318257532728866,
      "loss": 0.5442,
      "step": 2402
    },
    {
      "epoch": 0.3589781894233642,
      "grad_norm": 0.15120452642440796,
      "learning_rate": 0.00014314017705749626,
      "loss": 0.6071,
      "step": 2403
    },
    {
      "epoch": 0.35912757693456826,
      "grad_norm": 0.233122780919075,
      "learning_rate": 0.0001430977692573554,
      "loss": 0.7562,
      "step": 2404
    },
    {
      "epoch": 0.35927696444577234,
      "grad_norm": 0.23194383084774017,
      "learning_rate": 0.00014305535193623474,
      "loss": 0.5493,
      "step": 2405
    },
    {
      "epoch": 0.3594263519569764,
      "grad_norm": 0.18080557882785797,
      "learning_rate": 0.00014301292510350485,
      "loss": 0.9932,
      "step": 2406
    },
    {
      "epoch": 0.35957573946818044,
      "grad_norm": 0.1705053746700287,
      "learning_rate": 0.00014297048876853852,
      "loss": 0.7797,
      "step": 2407
    },
    {
      "epoch": 0.3597251269793845,
      "grad_norm": 0.1859169900417328,
      "learning_rate": 0.0001429280429407106,
      "loss": 0.4589,
      "step": 2408
    },
    {
      "epoch": 0.3598745144905886,
      "grad_norm": 0.13879810273647308,
      "learning_rate": 0.00014288558762939806,
      "loss": 0.758,
      "step": 2409
    },
    {
      "epoch": 0.36002390200179263,
      "grad_norm": 0.14433880150318146,
      "learning_rate": 0.00014284312284397994,
      "loss": 0.9777,
      "step": 2410
    },
    {
      "epoch": 0.3601732895129967,
      "grad_norm": 0.134858176112175,
      "learning_rate": 0.00014280064859383739,
      "loss": 0.5217,
      "step": 2411
    },
    {
      "epoch": 0.3603226770242008,
      "grad_norm": 0.14556215703487396,
      "learning_rate": 0.00014275816488835364,
      "loss": 0.9242,
      "step": 2412
    },
    {
      "epoch": 0.3604720645354048,
      "grad_norm": 0.1300947368144989,
      "learning_rate": 0.00014271567173691397,
      "loss": 0.5911,
      "step": 2413
    },
    {
      "epoch": 0.3606214520466089,
      "grad_norm": 0.20407763123512268,
      "learning_rate": 0.00014267316914890583,
      "loss": 0.7447,
      "step": 2414
    },
    {
      "epoch": 0.360770839557813,
      "grad_norm": 0.11976472288370132,
      "learning_rate": 0.00014263065713371876,
      "loss": 0.7955,
      "step": 2415
    },
    {
      "epoch": 0.36092022706901705,
      "grad_norm": 0.17073266208171844,
      "learning_rate": 0.00014258813570074429,
      "loss": 0.7444,
      "step": 2416
    },
    {
      "epoch": 0.3610696145802211,
      "grad_norm": 0.1640082150697708,
      "learning_rate": 0.00014254560485937607,
      "loss": 0.5093,
      "step": 2417
    },
    {
      "epoch": 0.36121900209142516,
      "grad_norm": 0.1276959925889969,
      "learning_rate": 0.00014250306461900984,
      "loss": 0.67,
      "step": 2418
    },
    {
      "epoch": 0.36136838960262924,
      "grad_norm": 0.1521582454442978,
      "learning_rate": 0.00014246051498904345,
      "loss": 0.6601,
      "step": 2419
    },
    {
      "epoch": 0.36151777711383326,
      "grad_norm": 0.17073354125022888,
      "learning_rate": 0.00014241795597887675,
      "loss": 0.9145,
      "step": 2420
    },
    {
      "epoch": 0.36166716462503734,
      "grad_norm": 0.22814391553401947,
      "learning_rate": 0.0001423753875979117,
      "loss": 0.9405,
      "step": 2421
    },
    {
      "epoch": 0.3618165521362414,
      "grad_norm": 0.1149008497595787,
      "learning_rate": 0.00014233280985555234,
      "loss": 0.5919,
      "step": 2422
    },
    {
      "epoch": 0.3619659396474455,
      "grad_norm": 0.12494846433401108,
      "learning_rate": 0.00014229022276120477,
      "loss": 0.733,
      "step": 2423
    },
    {
      "epoch": 0.3621153271586495,
      "grad_norm": 0.14213088154792786,
      "learning_rate": 0.00014224762632427713,
      "loss": 0.7663,
      "step": 2424
    },
    {
      "epoch": 0.3622647146698536,
      "grad_norm": 0.1726178079843521,
      "learning_rate": 0.00014220502055417968,
      "loss": 0.5154,
      "step": 2425
    },
    {
      "epoch": 0.3624141021810577,
      "grad_norm": 0.20680487155914307,
      "learning_rate": 0.0001421624054603247,
      "loss": 0.9292,
      "step": 2426
    },
    {
      "epoch": 0.3625634896922617,
      "grad_norm": 0.1911637783050537,
      "learning_rate": 0.00014211978105212653,
      "loss": 0.6538,
      "step": 2427
    },
    {
      "epoch": 0.3627128772034658,
      "grad_norm": 0.16632306575775146,
      "learning_rate": 0.00014207714733900162,
      "loss": 0.4626,
      "step": 2428
    },
    {
      "epoch": 0.36286226471466987,
      "grad_norm": 0.15098689496517181,
      "learning_rate": 0.00014203450433036835,
      "loss": 0.7309,
      "step": 2429
    },
    {
      "epoch": 0.3630116522258739,
      "grad_norm": 0.14323461055755615,
      "learning_rate": 0.00014199185203564728,
      "loss": 0.5956,
      "step": 2430
    },
    {
      "epoch": 0.363161039737078,
      "grad_norm": 0.47918006777763367,
      "learning_rate": 0.000141949190464261,
      "loss": 1.2064,
      "step": 2431
    },
    {
      "epoch": 0.36331042724828205,
      "grad_norm": 0.29491880536079407,
      "learning_rate": 0.00014190651962563407,
      "loss": 1.329,
      "step": 2432
    },
    {
      "epoch": 0.36345981475948613,
      "grad_norm": 0.1892276108264923,
      "learning_rate": 0.00014186383952919323,
      "loss": 0.605,
      "step": 2433
    },
    {
      "epoch": 0.36360920227069016,
      "grad_norm": 0.15556937456130981,
      "learning_rate": 0.00014182115018436715,
      "loss": 0.5756,
      "step": 2434
    },
    {
      "epoch": 0.36375858978189424,
      "grad_norm": 0.22487881779670715,
      "learning_rate": 0.00014177845160058658,
      "loss": 0.9732,
      "step": 2435
    },
    {
      "epoch": 0.3639079772930983,
      "grad_norm": 0.13250693678855896,
      "learning_rate": 0.0001417357437872843,
      "loss": 0.8,
      "step": 2436
    },
    {
      "epoch": 0.36405736480430234,
      "grad_norm": 0.18051151931285858,
      "learning_rate": 0.00014169302675389516,
      "loss": 1.0482,
      "step": 2437
    },
    {
      "epoch": 0.3642067523155064,
      "grad_norm": 0.27097687125205994,
      "learning_rate": 0.00014165030050985604,
      "loss": 1.1912,
      "step": 2438
    },
    {
      "epoch": 0.3643561398267105,
      "grad_norm": 0.13915418088436127,
      "learning_rate": 0.0001416075650646058,
      "loss": 0.5344,
      "step": 2439
    },
    {
      "epoch": 0.3645055273379145,
      "grad_norm": 0.3761531412601471,
      "learning_rate": 0.00014156482042758544,
      "loss": 0.8674,
      "step": 2440
    },
    {
      "epoch": 0.3646549148491186,
      "grad_norm": 0.1343490481376648,
      "learning_rate": 0.00014152206660823788,
      "loss": 0.734,
      "step": 2441
    },
    {
      "epoch": 0.3648043023603227,
      "grad_norm": 1.089711308479309,
      "learning_rate": 0.0001414793036160081,
      "loss": 1.7908,
      "step": 2442
    },
    {
      "epoch": 0.36495368987152677,
      "grad_norm": 0.32068824768066406,
      "learning_rate": 0.0001414365314603432,
      "loss": 1.0298,
      "step": 2443
    },
    {
      "epoch": 0.3651030773827308,
      "grad_norm": 0.15771561861038208,
      "learning_rate": 0.00014139375015069215,
      "loss": 0.5826,
      "step": 2444
    },
    {
      "epoch": 0.36525246489393487,
      "grad_norm": 0.22867271304130554,
      "learning_rate": 0.0001413509596965061,
      "loss": 0.5699,
      "step": 2445
    },
    {
      "epoch": 0.36540185240513895,
      "grad_norm": 0.19706255197525024,
      "learning_rate": 0.00014130816010723805,
      "loss": 0.5931,
      "step": 2446
    },
    {
      "epoch": 0.365551239916343,
      "grad_norm": 0.14952997863292694,
      "learning_rate": 0.0001412653513923431,
      "loss": 0.7402,
      "step": 2447
    },
    {
      "epoch": 0.36570062742754705,
      "grad_norm": 0.19710053503513336,
      "learning_rate": 0.0001412225335612785,
      "loss": 0.3707,
      "step": 2448
    },
    {
      "epoch": 0.36585001493875113,
      "grad_norm": 0.11646145582199097,
      "learning_rate": 0.00014117970662350327,
      "loss": 0.712,
      "step": 2449
    },
    {
      "epoch": 0.36599940244995516,
      "grad_norm": 0.2085895538330078,
      "learning_rate": 0.00014113687058847857,
      "loss": 0.9136,
      "step": 2450
    },
    {
      "epoch": 0.36614878996115924,
      "grad_norm": 0.1410319209098816,
      "learning_rate": 0.0001410940254656676,
      "loss": 0.9974,
      "step": 2451
    },
    {
      "epoch": 0.3662981774723633,
      "grad_norm": 0.13308507204055786,
      "learning_rate": 0.00014105117126453554,
      "loss": 0.7333,
      "step": 2452
    },
    {
      "epoch": 0.3664475649835674,
      "grad_norm": 0.139163076877594,
      "learning_rate": 0.0001410083079945495,
      "loss": 0.6924,
      "step": 2453
    },
    {
      "epoch": 0.3665969524947714,
      "grad_norm": 0.17223769426345825,
      "learning_rate": 0.00014096543566517871,
      "loss": 0.7073,
      "step": 2454
    },
    {
      "epoch": 0.3667463400059755,
      "grad_norm": 0.11783270537853241,
      "learning_rate": 0.00014092255428589429,
      "loss": 0.8244,
      "step": 2455
    },
    {
      "epoch": 0.3668957275171796,
      "grad_norm": 0.13273076713085175,
      "learning_rate": 0.00014087966386616945,
      "loss": 0.8539,
      "step": 2456
    },
    {
      "epoch": 0.3670451150283836,
      "grad_norm": 0.12106109410524368,
      "learning_rate": 0.0001408367644154794,
      "loss": 0.509,
      "step": 2457
    },
    {
      "epoch": 0.3671945025395877,
      "grad_norm": 0.173287034034729,
      "learning_rate": 0.00014079385594330121,
      "loss": 0.6208,
      "step": 2458
    },
    {
      "epoch": 0.36734389005079177,
      "grad_norm": 0.15107597410678864,
      "learning_rate": 0.00014075093845911414,
      "loss": 0.9295,
      "step": 2459
    },
    {
      "epoch": 0.3674932775619958,
      "grad_norm": 0.21176166832447052,
      "learning_rate": 0.00014070801197239928,
      "loss": 0.9364,
      "step": 2460
    },
    {
      "epoch": 0.36764266507319987,
      "grad_norm": 0.23147724568843842,
      "learning_rate": 0.00014066507649263983,
      "loss": 0.5677,
      "step": 2461
    },
    {
      "epoch": 0.36779205258440395,
      "grad_norm": 0.13638894259929657,
      "learning_rate": 0.00014062213202932085,
      "loss": 0.5624,
      "step": 2462
    },
    {
      "epoch": 0.36794144009560803,
      "grad_norm": 0.17373988032341003,
      "learning_rate": 0.0001405791785919295,
      "loss": 0.7032,
      "step": 2463
    },
    {
      "epoch": 0.36809082760681205,
      "grad_norm": 0.13953036069869995,
      "learning_rate": 0.00014053621618995488,
      "loss": 0.6328,
      "step": 2464
    },
    {
      "epoch": 0.36824021511801613,
      "grad_norm": 0.12342777848243713,
      "learning_rate": 0.00014049324483288798,
      "loss": 0.7745,
      "step": 2465
    },
    {
      "epoch": 0.3683896026292202,
      "grad_norm": 0.23031273484230042,
      "learning_rate": 0.00014045026453022197,
      "loss": 0.5934,
      "step": 2466
    },
    {
      "epoch": 0.36853899014042424,
      "grad_norm": 0.26746872067451477,
      "learning_rate": 0.00014040727529145182,
      "loss": 1.1887,
      "step": 2467
    },
    {
      "epoch": 0.3686883776516283,
      "grad_norm": 11.205217361450195,
      "learning_rate": 0.00014036427712607453,
      "loss": 2.7668,
      "step": 2468
    },
    {
      "epoch": 0.3688377651628324,
      "grad_norm": 0.26219356060028076,
      "learning_rate": 0.00014032127004358905,
      "loss": 0.8974,
      "step": 2469
    },
    {
      "epoch": 0.3689871526740365,
      "grad_norm": 0.1428171992301941,
      "learning_rate": 0.00014027825405349642,
      "loss": 0.7776,
      "step": 2470
    },
    {
      "epoch": 0.3691365401852405,
      "grad_norm": 0.1508057862520218,
      "learning_rate": 0.00014023522916529948,
      "loss": 1.0241,
      "step": 2471
    },
    {
      "epoch": 0.3692859276964446,
      "grad_norm": 0.17132999002933502,
      "learning_rate": 0.0001401921953885031,
      "loss": 0.6854,
      "step": 2472
    },
    {
      "epoch": 0.36943531520764866,
      "grad_norm": 0.13361242413520813,
      "learning_rate": 0.00014014915273261418,
      "loss": 0.8027,
      "step": 2473
    },
    {
      "epoch": 0.3695847027188527,
      "grad_norm": 0.11395853012800217,
      "learning_rate": 0.00014010610120714147,
      "loss": 0.7048,
      "step": 2474
    },
    {
      "epoch": 0.36973409023005677,
      "grad_norm": 0.20443783700466156,
      "learning_rate": 0.00014006304082159575,
      "loss": 1.0703,
      "step": 2475
    },
    {
      "epoch": 0.36988347774126085,
      "grad_norm": 0.1404961496591568,
      "learning_rate": 0.00014001997158548973,
      "loss": 0.8376,
      "step": 2476
    },
    {
      "epoch": 0.37003286525246487,
      "grad_norm": 0.148174449801445,
      "learning_rate": 0.00013997689350833814,
      "loss": 0.6878,
      "step": 2477
    },
    {
      "epoch": 0.37018225276366895,
      "grad_norm": 0.12239596247673035,
      "learning_rate": 0.00013993380659965755,
      "loss": 0.4875,
      "step": 2478
    },
    {
      "epoch": 0.37033164027487303,
      "grad_norm": 0.13538095355033875,
      "learning_rate": 0.0001398907108689666,
      "loss": 0.6911,
      "step": 2479
    },
    {
      "epoch": 0.3704810277860771,
      "grad_norm": 0.17555901408195496,
      "learning_rate": 0.00013984760632578577,
      "loss": 1.0559,
      "step": 2480
    },
    {
      "epoch": 0.37063041529728113,
      "grad_norm": 0.2748256325721741,
      "learning_rate": 0.00013980449297963752,
      "loss": 0.7446,
      "step": 2481
    },
    {
      "epoch": 0.3707798028084852,
      "grad_norm": 0.16466206312179565,
      "learning_rate": 0.00013976137084004633,
      "loss": 0.7396,
      "step": 2482
    },
    {
      "epoch": 0.3709291903196893,
      "grad_norm": 0.28830182552337646,
      "learning_rate": 0.0001397182399165385,
      "loss": 0.6485,
      "step": 2483
    },
    {
      "epoch": 0.3710785778308933,
      "grad_norm": 0.33990252017974854,
      "learning_rate": 0.0001396751002186424,
      "loss": 1.1408,
      "step": 2484
    },
    {
      "epoch": 0.3712279653420974,
      "grad_norm": 0.1105538159608841,
      "learning_rate": 0.00013963195175588824,
      "loss": 0.557,
      "step": 2485
    },
    {
      "epoch": 0.3713773528533015,
      "grad_norm": 0.22302602231502533,
      "learning_rate": 0.00013958879453780817,
      "loss": 0.6195,
      "step": 2486
    },
    {
      "epoch": 0.3715267403645055,
      "grad_norm": 0.13026951253414154,
      "learning_rate": 0.00013954562857393637,
      "loss": 0.85,
      "step": 2487
    },
    {
      "epoch": 0.3716761278757096,
      "grad_norm": 0.14023616909980774,
      "learning_rate": 0.00013950245387380882,
      "loss": 0.4141,
      "step": 2488
    },
    {
      "epoch": 0.37182551538691366,
      "grad_norm": 0.14884553849697113,
      "learning_rate": 0.00013945927044696355,
      "loss": 0.869,
      "step": 2489
    },
    {
      "epoch": 0.37197490289811774,
      "grad_norm": 0.12515412271022797,
      "learning_rate": 0.00013941607830294042,
      "loss": 0.6266,
      "step": 2490
    },
    {
      "epoch": 0.37212429040932177,
      "grad_norm": 0.16218025982379913,
      "learning_rate": 0.00013937287745128128,
      "loss": 0.9316,
      "step": 2491
    },
    {
      "epoch": 0.37227367792052585,
      "grad_norm": 0.35289645195007324,
      "learning_rate": 0.00013932966790152987,
      "loss": 1.0348,
      "step": 2492
    },
    {
      "epoch": 0.3724230654317299,
      "grad_norm": 0.21793174743652344,
      "learning_rate": 0.00013928644966323188,
      "loss": 0.4258,
      "step": 2493
    },
    {
      "epoch": 0.37257245294293395,
      "grad_norm": 0.18603777885437012,
      "learning_rate": 0.00013924322274593486,
      "loss": 0.8372,
      "step": 2494
    },
    {
      "epoch": 0.37272184045413803,
      "grad_norm": 0.14018608629703522,
      "learning_rate": 0.00013919998715918836,
      "loss": 0.9426,
      "step": 2495
    },
    {
      "epoch": 0.3728712279653421,
      "grad_norm": 0.11894482374191284,
      "learning_rate": 0.00013915674291254383,
      "loss": 0.6959,
      "step": 2496
    },
    {
      "epoch": 0.37302061547654614,
      "grad_norm": 0.15760189294815063,
      "learning_rate": 0.0001391134900155546,
      "loss": 0.706,
      "step": 2497
    },
    {
      "epoch": 0.3731700029877502,
      "grad_norm": 0.12037986516952515,
      "learning_rate": 0.00013907022847777585,
      "loss": 0.7656,
      "step": 2498
    },
    {
      "epoch": 0.3733193904989543,
      "grad_norm": 0.17428357899188995,
      "learning_rate": 0.00013902695830876483,
      "loss": 0.6979,
      "step": 2499
    },
    {
      "epoch": 0.3734687780101584,
      "grad_norm": 0.13507042825222015,
      "learning_rate": 0.00013898367951808052,
      "loss": 0.5623,
      "step": 2500
    },
    {
      "epoch": 0.3736181655213624,
      "grad_norm": 0.21382325887680054,
      "learning_rate": 0.00013894039211528395,
      "loss": 0.7443,
      "step": 2501
    },
    {
      "epoch": 0.3737675530325665,
      "grad_norm": 0.21533359587192535,
      "learning_rate": 0.000138897096109938,
      "loss": 1.0596,
      "step": 2502
    },
    {
      "epoch": 0.37391694054377056,
      "grad_norm": 0.19932664930820465,
      "learning_rate": 0.0001388537915116074,
      "loss": 0.6498,
      "step": 2503
    },
    {
      "epoch": 0.3740663280549746,
      "grad_norm": 0.30534040927886963,
      "learning_rate": 0.00013881047832985886,
      "loss": 0.9741,
      "step": 2504
    },
    {
      "epoch": 0.37421571556617866,
      "grad_norm": 0.14218056201934814,
      "learning_rate": 0.0001387671565742609,
      "loss": 0.8208,
      "step": 2505
    },
    {
      "epoch": 0.37436510307738274,
      "grad_norm": 0.2969071567058563,
      "learning_rate": 0.00013872382625438405,
      "loss": 1.2549,
      "step": 2506
    },
    {
      "epoch": 0.37451449058858677,
      "grad_norm": 4.911860942840576,
      "learning_rate": 0.00013868048737980063,
      "loss": 2.5317,
      "step": 2507
    },
    {
      "epoch": 0.37466387809979085,
      "grad_norm": 0.29426854848861694,
      "learning_rate": 0.00013863713996008483,
      "loss": 0.9459,
      "step": 2508
    },
    {
      "epoch": 0.37481326561099493,
      "grad_norm": 0.20989391207695007,
      "learning_rate": 0.0001385937840048129,
      "loss": 0.7703,
      "step": 2509
    },
    {
      "epoch": 0.374962653122199,
      "grad_norm": 0.15865615010261536,
      "learning_rate": 0.00013855041952356273,
      "loss": 0.7891,
      "step": 2510
    },
    {
      "epoch": 0.37511204063340303,
      "grad_norm": 0.09545331448316574,
      "learning_rate": 0.0001385070465259143,
      "loss": 0.426,
      "step": 2511
    },
    {
      "epoch": 0.3752614281446071,
      "grad_norm": 0.1343315839767456,
      "learning_rate": 0.00013846366502144936,
      "loss": 0.6051,
      "step": 2512
    },
    {
      "epoch": 0.3754108156558112,
      "grad_norm": 0.2765781581401825,
      "learning_rate": 0.00013842027501975162,
      "loss": 1.1162,
      "step": 2513
    },
    {
      "epoch": 0.3755602031670152,
      "grad_norm": 0.3213646709918976,
      "learning_rate": 0.00013837687653040653,
      "loss": 1.1929,
      "step": 2514
    },
    {
      "epoch": 0.3757095906782193,
      "grad_norm": 0.2544189989566803,
      "learning_rate": 0.0001383334695630016,
      "loss": 0.9352,
      "step": 2515
    },
    {
      "epoch": 0.3758589781894234,
      "grad_norm": 0.12397495657205582,
      "learning_rate": 0.00013829005412712607,
      "loss": 0.7535,
      "step": 2516
    },
    {
      "epoch": 0.3760083657006274,
      "grad_norm": 0.2225325107574463,
      "learning_rate": 0.0001382466302323711,
      "loss": 0.8956,
      "step": 2517
    },
    {
      "epoch": 0.3761577532118315,
      "grad_norm": 0.43180760741233826,
      "learning_rate": 0.00013820319788832968,
      "loss": 0.9872,
      "step": 2518
    },
    {
      "epoch": 0.37630714072303556,
      "grad_norm": 0.20235896110534668,
      "learning_rate": 0.0001381597571045968,
      "loss": 0.8725,
      "step": 2519
    },
    {
      "epoch": 0.37645652823423964,
      "grad_norm": 0.12181615084409714,
      "learning_rate": 0.0001381163078907691,
      "loss": 0.6685,
      "step": 2520
    },
    {
      "epoch": 0.37660591574544366,
      "grad_norm": 0.20705847442150116,
      "learning_rate": 0.00013807285025644526,
      "loss": 0.9449,
      "step": 2521
    },
    {
      "epoch": 0.37675530325664774,
      "grad_norm": 0.16851113736629486,
      "learning_rate": 0.0001380293842112258,
      "loss": 0.7579,
      "step": 2522
    },
    {
      "epoch": 0.3769046907678518,
      "grad_norm": 0.12044961750507355,
      "learning_rate": 0.00013798590976471297,
      "loss": 0.6829,
      "step": 2523
    },
    {
      "epoch": 0.37705407827905585,
      "grad_norm": 0.1249905601143837,
      "learning_rate": 0.00013794242692651102,
      "loss": 0.5983,
      "step": 2524
    },
    {
      "epoch": 0.37720346579025993,
      "grad_norm": 0.1865132749080658,
      "learning_rate": 0.000137898935706226,
      "loss": 0.6271,
      "step": 2525
    },
    {
      "epoch": 0.377352853301464,
      "grad_norm": 0.17507433891296387,
      "learning_rate": 0.00013785543611346578,
      "loss": 0.5406,
      "step": 2526
    },
    {
      "epoch": 0.3775022408126681,
      "grad_norm": 0.1108279675245285,
      "learning_rate": 0.00013781192815784014,
      "loss": 0.751,
      "step": 2527
    },
    {
      "epoch": 0.3776516283238721,
      "grad_norm": 0.1221894696354866,
      "learning_rate": 0.00013776841184896064,
      "loss": 0.6087,
      "step": 2528
    },
    {
      "epoch": 0.3778010158350762,
      "grad_norm": 0.15383478999137878,
      "learning_rate": 0.00013772488719644075,
      "loss": 0.7539,
      "step": 2529
    },
    {
      "epoch": 0.3779504033462803,
      "grad_norm": 0.1354319304227829,
      "learning_rate": 0.00013768135420989577,
      "loss": 0.6343,
      "step": 2530
    },
    {
      "epoch": 0.3780997908574843,
      "grad_norm": 0.29546427726745605,
      "learning_rate": 0.00013763781289894274,
      "loss": 0.9065,
      "step": 2531
    },
    {
      "epoch": 0.3782491783686884,
      "grad_norm": 0.12710508704185486,
      "learning_rate": 0.00013759426327320074,
      "loss": 1.0793,
      "step": 2532
    },
    {
      "epoch": 0.37839856587989246,
      "grad_norm": 0.1346661001443863,
      "learning_rate": 0.0001375507053422905,
      "loss": 0.6063,
      "step": 2533
    },
    {
      "epoch": 0.3785479533910965,
      "grad_norm": 0.27018964290618896,
      "learning_rate": 0.0001375071391158347,
      "loss": 1.1654,
      "step": 2534
    },
    {
      "epoch": 0.37869734090230056,
      "grad_norm": 0.11668441444635391,
      "learning_rate": 0.00013746356460345779,
      "loss": 0.7728,
      "step": 2535
    },
    {
      "epoch": 0.37884672841350464,
      "grad_norm": 0.2194286435842514,
      "learning_rate": 0.00013741998181478603,
      "loss": 0.826,
      "step": 2536
    },
    {
      "epoch": 0.3789961159247087,
      "grad_norm": 0.14430202543735504,
      "learning_rate": 0.0001373763907594476,
      "loss": 0.7729,
      "step": 2537
    },
    {
      "epoch": 0.37914550343591275,
      "grad_norm": 0.13049617409706116,
      "learning_rate": 0.00013733279144707245,
      "loss": 0.7842,
      "step": 2538
    },
    {
      "epoch": 0.3792948909471168,
      "grad_norm": 0.18665525317192078,
      "learning_rate": 0.00013728918388729232,
      "loss": 0.6885,
      "step": 2539
    },
    {
      "epoch": 0.3794442784583209,
      "grad_norm": 0.1557392179965973,
      "learning_rate": 0.00013724556808974086,
      "loss": 0.9572,
      "step": 2540
    },
    {
      "epoch": 0.37959366596952493,
      "grad_norm": 0.15846869349479675,
      "learning_rate": 0.00013720194406405348,
      "loss": 0.8906,
      "step": 2541
    },
    {
      "epoch": 0.379743053480729,
      "grad_norm": 0.13207389414310455,
      "learning_rate": 0.0001371583118198674,
      "loss": 0.9396,
      "step": 2542
    },
    {
      "epoch": 0.3798924409919331,
      "grad_norm": 0.21497125923633575,
      "learning_rate": 0.00013711467136682168,
      "loss": 0.9738,
      "step": 2543
    },
    {
      "epoch": 0.3800418285031371,
      "grad_norm": 0.1559433937072754,
      "learning_rate": 0.0001370710227145572,
      "loss": 0.482,
      "step": 2544
    },
    {
      "epoch": 0.3801912160143412,
      "grad_norm": 0.24656884372234344,
      "learning_rate": 0.00013702736587271663,
      "loss": 1.1823,
      "step": 2545
    },
    {
      "epoch": 0.3803406035255453,
      "grad_norm": 0.12045598775148392,
      "learning_rate": 0.00013698370085094442,
      "loss": 0.8941,
      "step": 2546
    },
    {
      "epoch": 0.38048999103674935,
      "grad_norm": 0.17790566384792328,
      "learning_rate": 0.00013694002765888693,
      "loss": 0.5948,
      "step": 2547
    },
    {
      "epoch": 0.3806393785479534,
      "grad_norm": 0.1559494584798813,
      "learning_rate": 0.0001368963463061922,
      "loss": 0.7468,
      "step": 2548
    },
    {
      "epoch": 0.38078876605915746,
      "grad_norm": 0.13779549300670624,
      "learning_rate": 0.00013685265680251022,
      "loss": 0.8985,
      "step": 2549
    },
    {
      "epoch": 0.38093815357036154,
      "grad_norm": 0.17108717560768127,
      "learning_rate": 0.0001368089591574926,
      "loss": 0.6023,
      "step": 2550
    },
    {
      "epoch": 0.38108754108156556,
      "grad_norm": 0.23677372932434082,
      "learning_rate": 0.00013676525338079286,
      "loss": 1.1638,
      "step": 2551
    },
    {
      "epoch": 0.38123692859276964,
      "grad_norm": 0.108987957239151,
      "learning_rate": 0.00013672153948206635,
      "loss": 0.4463,
      "step": 2552
    },
    {
      "epoch": 0.3813863161039737,
      "grad_norm": 0.23067811131477356,
      "learning_rate": 0.0001366778174709701,
      "loss": 1.0109,
      "step": 2553
    },
    {
      "epoch": 0.38153570361517775,
      "grad_norm": 0.1385107785463333,
      "learning_rate": 0.00013663408735716307,
      "loss": 0.772,
      "step": 2554
    },
    {
      "epoch": 0.3816850911263818,
      "grad_norm": 0.35186049342155457,
      "learning_rate": 0.00013659034915030585,
      "loss": 1.0516,
      "step": 2555
    },
    {
      "epoch": 0.3818344786375859,
      "grad_norm": 0.13650234043598175,
      "learning_rate": 0.00013654660286006095,
      "loss": 0.7608,
      "step": 2556
    },
    {
      "epoch": 0.38198386614879,
      "grad_norm": 0.21628083288669586,
      "learning_rate": 0.00013650284849609263,
      "loss": 0.9622,
      "step": 2557
    },
    {
      "epoch": 0.382133253659994,
      "grad_norm": 0.1910853236913681,
      "learning_rate": 0.0001364590860680669,
      "loss": 0.5907,
      "step": 2558
    },
    {
      "epoch": 0.3822826411711981,
      "grad_norm": 0.14919957518577576,
      "learning_rate": 0.00013641531558565152,
      "loss": 0.5221,
      "step": 2559
    },
    {
      "epoch": 0.38243202868240217,
      "grad_norm": 0.12867121398448944,
      "learning_rate": 0.00013637153705851616,
      "loss": 0.8938,
      "step": 2560
    },
    {
      "epoch": 0.3825814161936062,
      "grad_norm": 0.183865487575531,
      "learning_rate": 0.0001363277504963322,
      "loss": 0.4792,
      "step": 2561
    },
    {
      "epoch": 0.3827308037048103,
      "grad_norm": 0.23723919689655304,
      "learning_rate": 0.00013628395590877277,
      "loss": 0.6296,
      "step": 2562
    },
    {
      "epoch": 0.38288019121601435,
      "grad_norm": 0.46233540773391724,
      "learning_rate": 0.00013624015330551272,
      "loss": 0.8328,
      "step": 2563
    },
    {
      "epoch": 0.3830295787272184,
      "grad_norm": 0.13171587884426117,
      "learning_rate": 0.00013619634269622884,
      "loss": 0.6896,
      "step": 2564
    },
    {
      "epoch": 0.38317896623842246,
      "grad_norm": 0.12042468041181564,
      "learning_rate": 0.0001361525240905995,
      "loss": 0.5986,
      "step": 2565
    },
    {
      "epoch": 0.38332835374962654,
      "grad_norm": 0.14616362750530243,
      "learning_rate": 0.00013610869749830498,
      "loss": 0.8106,
      "step": 2566
    },
    {
      "epoch": 0.3834777412608306,
      "grad_norm": 0.20069532096385956,
      "learning_rate": 0.00013606486292902723,
      "loss": 0.8731,
      "step": 2567
    },
    {
      "epoch": 0.38362712877203464,
      "grad_norm": 0.144715815782547,
      "learning_rate": 0.00013602102039245002,
      "loss": 0.6771,
      "step": 2568
    },
    {
      "epoch": 0.3837765162832387,
      "grad_norm": 5.180905342102051,
      "learning_rate": 0.00013597716989825886,
      "loss": 1.6743,
      "step": 2569
    },
    {
      "epoch": 0.3839259037944428,
      "grad_norm": 0.165913388133049,
      "learning_rate": 0.00013593331145614104,
      "loss": 0.6524,
      "step": 2570
    },
    {
      "epoch": 0.3840752913056468,
      "grad_norm": 0.25694411993026733,
      "learning_rate": 0.0001358894450757856,
      "loss": 0.738,
      "step": 2571
    },
    {
      "epoch": 0.3842246788168509,
      "grad_norm": 0.19058038294315338,
      "learning_rate": 0.00013584557076688322,
      "loss": 0.8229,
      "step": 2572
    },
    {
      "epoch": 0.384374066328055,
      "grad_norm": 0.16156141459941864,
      "learning_rate": 0.00013580168853912655,
      "loss": 0.4887,
      "step": 2573
    },
    {
      "epoch": 0.384523453839259,
      "grad_norm": 0.14655321836471558,
      "learning_rate": 0.00013575779840220976,
      "loss": 0.6537,
      "step": 2574
    },
    {
      "epoch": 0.3846728413504631,
      "grad_norm": 0.8720521330833435,
      "learning_rate": 0.00013571390036582894,
      "loss": 1.9229,
      "step": 2575
    },
    {
      "epoch": 0.38482222886166717,
      "grad_norm": 0.21166038513183594,
      "learning_rate": 0.00013566999443968185,
      "loss": 1.005,
      "step": 2576
    },
    {
      "epoch": 0.38497161637287125,
      "grad_norm": 0.20068562030792236,
      "learning_rate": 0.00013562608063346802,
      "loss": 0.7407,
      "step": 2577
    },
    {
      "epoch": 0.3851210038840753,
      "grad_norm": 0.16301977634429932,
      "learning_rate": 0.00013558215895688867,
      "loss": 0.7166,
      "step": 2578
    },
    {
      "epoch": 0.38527039139527935,
      "grad_norm": 0.24074499309062958,
      "learning_rate": 0.00013553822941964682,
      "loss": 0.6777,
      "step": 2579
    },
    {
      "epoch": 0.38541977890648343,
      "grad_norm": 0.18944929540157318,
      "learning_rate": 0.00013549429203144723,
      "loss": 0.6196,
      "step": 2580
    },
    {
      "epoch": 0.38556916641768746,
      "grad_norm": 0.24977436661720276,
      "learning_rate": 0.0001354503468019963,
      "loss": 0.7757,
      "step": 2581
    },
    {
      "epoch": 0.38571855392889154,
      "grad_norm": 0.1581128090620041,
      "learning_rate": 0.00013540639374100226,
      "loss": 0.4881,
      "step": 2582
    },
    {
      "epoch": 0.3858679414400956,
      "grad_norm": 0.14354291558265686,
      "learning_rate": 0.00013536243285817504,
      "loss": 0.9198,
      "step": 2583
    },
    {
      "epoch": 0.3860173289512997,
      "grad_norm": 0.41500425338745117,
      "learning_rate": 0.00013531846416322627,
      "loss": 1.3847,
      "step": 2584
    },
    {
      "epoch": 0.3861667164625037,
      "grad_norm": 0.12938615679740906,
      "learning_rate": 0.00013527448766586934,
      "loss": 0.7046,
      "step": 2585
    },
    {
      "epoch": 0.3863161039737078,
      "grad_norm": 0.13857179880142212,
      "learning_rate": 0.00013523050337581943,
      "loss": 0.6458,
      "step": 2586
    },
    {
      "epoch": 0.3864654914849119,
      "grad_norm": 0.18126429617404938,
      "learning_rate": 0.00013518651130279324,
      "loss": 1.0078,
      "step": 2587
    },
    {
      "epoch": 0.3866148789961159,
      "grad_norm": 0.5621423125267029,
      "learning_rate": 0.0001351425114565094,
      "loss": 1.1803,
      "step": 2588
    },
    {
      "epoch": 0.38676426650732,
      "grad_norm": 0.28276127576828003,
      "learning_rate": 0.00013509850384668814,
      "loss": 0.6218,
      "step": 2589
    },
    {
      "epoch": 0.38691365401852407,
      "grad_norm": 0.18387152254581451,
      "learning_rate": 0.0001350544884830515,
      "loss": 0.6942,
      "step": 2590
    },
    {
      "epoch": 0.3870630415297281,
      "grad_norm": 0.25450313091278076,
      "learning_rate": 0.00013501046537532305,
      "loss": 1.0552,
      "step": 2591
    },
    {
      "epoch": 0.38721242904093217,
      "grad_norm": 0.14817726612091064,
      "learning_rate": 0.00013496643453322828,
      "loss": 0.8778,
      "step": 2592
    },
    {
      "epoch": 0.38736181655213625,
      "grad_norm": 0.43990394473075867,
      "learning_rate": 0.0001349223959664943,
      "loss": 1.12,
      "step": 2593
    },
    {
      "epoch": 0.38751120406334033,
      "grad_norm": 0.16388940811157227,
      "learning_rate": 0.0001348783496848499,
      "loss": 0.7339,
      "step": 2594
    },
    {
      "epoch": 0.38766059157454436,
      "grad_norm": 0.17454546689987183,
      "learning_rate": 0.0001348342956980256,
      "loss": 0.7154,
      "step": 2595
    },
    {
      "epoch": 0.38780997908574844,
      "grad_norm": 0.22581076622009277,
      "learning_rate": 0.00013479023401575366,
      "loss": 0.9446,
      "step": 2596
    },
    {
      "epoch": 0.3879593665969525,
      "grad_norm": 0.19662749767303467,
      "learning_rate": 0.00013474616464776795,
      "loss": 0.9129,
      "step": 2597
    },
    {
      "epoch": 0.38810875410815654,
      "grad_norm": 0.14187973737716675,
      "learning_rate": 0.00013470208760380412,
      "loss": 0.7935,
      "step": 2598
    },
    {
      "epoch": 0.3882581416193606,
      "grad_norm": 0.15428054332733154,
      "learning_rate": 0.00013465800289359951,
      "loss": 0.7923,
      "step": 2599
    },
    {
      "epoch": 0.3884075291305647,
      "grad_norm": 0.19818037748336792,
      "learning_rate": 0.0001346139105268931,
      "loss": 0.756,
      "step": 2600
    },
    {
      "epoch": 0.3885569166417687,
      "grad_norm": 0.19121751189231873,
      "learning_rate": 0.0001345698105134256,
      "loss": 0.6549,
      "step": 2601
    },
    {
      "epoch": 0.3887063041529728,
      "grad_norm": 0.12198566645383835,
      "learning_rate": 0.00013452570286293938,
      "loss": 0.5057,
      "step": 2602
    },
    {
      "epoch": 0.3888556916641769,
      "grad_norm": 0.4115545451641083,
      "learning_rate": 0.00013448158758517853,
      "loss": 0.7019,
      "step": 2603
    },
    {
      "epoch": 0.38900507917538096,
      "grad_norm": 0.14466340839862823,
      "learning_rate": 0.00013443746468988884,
      "loss": 0.8026,
      "step": 2604
    },
    {
      "epoch": 0.389154466686585,
      "grad_norm": 0.121718630194664,
      "learning_rate": 0.00013439333418681772,
      "loss": 0.8005,
      "step": 2605
    },
    {
      "epoch": 0.38930385419778907,
      "grad_norm": 0.12980741262435913,
      "learning_rate": 0.00013434919608571437,
      "loss": 0.6259,
      "step": 2606
    },
    {
      "epoch": 0.38945324170899315,
      "grad_norm": 0.19823727011680603,
      "learning_rate": 0.0001343050503963295,
      "loss": 0.9948,
      "step": 2607
    },
    {
      "epoch": 0.38960262922019717,
      "grad_norm": 0.14001765847206116,
      "learning_rate": 0.00013426089712841564,
      "loss": 0.8101,
      "step": 2608
    },
    {
      "epoch": 0.38975201673140125,
      "grad_norm": 0.17087489366531372,
      "learning_rate": 0.00013421673629172696,
      "loss": 0.697,
      "step": 2609
    },
    {
      "epoch": 0.38990140424260533,
      "grad_norm": 0.139466792345047,
      "learning_rate": 0.00013417256789601925,
      "loss": 0.6222,
      "step": 2610
    },
    {
      "epoch": 0.39005079175380936,
      "grad_norm": 0.18401959538459778,
      "learning_rate": 0.00013412839195105,
      "loss": 0.5941,
      "step": 2611
    },
    {
      "epoch": 0.39020017926501344,
      "grad_norm": 0.16685070097446442,
      "learning_rate": 0.00013408420846657844,
      "loss": 0.8743,
      "step": 2612
    },
    {
      "epoch": 0.3903495667762175,
      "grad_norm": 0.3349725306034088,
      "learning_rate": 0.00013404001745236533,
      "loss": 1.3105,
      "step": 2613
    },
    {
      "epoch": 0.3904989542874216,
      "grad_norm": 0.11015414446592331,
      "learning_rate": 0.00013399581891817324,
      "loss": 0.5633,
      "step": 2614
    },
    {
      "epoch": 0.3906483417986256,
      "grad_norm": 0.19119904935359955,
      "learning_rate": 0.00013395161287376624,
      "loss": 0.7372,
      "step": 2615
    },
    {
      "epoch": 0.3907977293098297,
      "grad_norm": 0.1373324692249298,
      "learning_rate": 0.00013390739932891022,
      "loss": 0.7534,
      "step": 2616
    },
    {
      "epoch": 0.3909471168210338,
      "grad_norm": 0.2964041531085968,
      "learning_rate": 0.00013386317829337262,
      "loss": 1.0623,
      "step": 2617
    },
    {
      "epoch": 0.3910965043322378,
      "grad_norm": 0.18005995452404022,
      "learning_rate": 0.00013381894977692257,
      "loss": 0.8697,
      "step": 2618
    },
    {
      "epoch": 0.3912458918434419,
      "grad_norm": 0.13531728088855743,
      "learning_rate": 0.00013377471378933086,
      "loss": 0.713,
      "step": 2619
    },
    {
      "epoch": 0.39139527935464596,
      "grad_norm": 0.15257832407951355,
      "learning_rate": 0.00013373047034036988,
      "loss": 0.6063,
      "step": 2620
    },
    {
      "epoch": 0.39154466686585,
      "grad_norm": 0.1412448287010193,
      "learning_rate": 0.00013368621943981375,
      "loss": 0.6743,
      "step": 2621
    },
    {
      "epoch": 0.39169405437705407,
      "grad_norm": 0.14522215723991394,
      "learning_rate": 0.0001336419610974382,
      "loss": 0.5655,
      "step": 2622
    },
    {
      "epoch": 0.39184344188825815,
      "grad_norm": 0.17962020635604858,
      "learning_rate": 0.00013359769532302057,
      "loss": 0.6511,
      "step": 2623
    },
    {
      "epoch": 0.39199282939946223,
      "grad_norm": 0.1276795119047165,
      "learning_rate": 0.00013355342212633986,
      "loss": 0.6817,
      "step": 2624
    },
    {
      "epoch": 0.39214221691066625,
      "grad_norm": 0.1390298455953598,
      "learning_rate": 0.00013350914151717683,
      "loss": 0.6588,
      "step": 2625
    },
    {
      "epoch": 0.39229160442187033,
      "grad_norm": 0.16389770805835724,
      "learning_rate": 0.0001334648535053136,
      "loss": 0.7673,
      "step": 2626
    },
    {
      "epoch": 0.3924409919330744,
      "grad_norm": 0.23969946801662445,
      "learning_rate": 0.0001334205581005342,
      "loss": 0.9776,
      "step": 2627
    },
    {
      "epoch": 0.39259037944427844,
      "grad_norm": 0.16722801327705383,
      "learning_rate": 0.00013337625531262414,
      "loss": 0.4678,
      "step": 2628
    },
    {
      "epoch": 0.3927397669554825,
      "grad_norm": 0.13670620322227478,
      "learning_rate": 0.00013333194515137065,
      "loss": 0.8396,
      "step": 2629
    },
    {
      "epoch": 0.3928891544666866,
      "grad_norm": 0.10439969599246979,
      "learning_rate": 0.0001332876276265625,
      "loss": 0.6072,
      "step": 2630
    },
    {
      "epoch": 0.3930385419778906,
      "grad_norm": 0.12676899135112762,
      "learning_rate": 0.00013324330274799018,
      "loss": 0.7611,
      "step": 2631
    },
    {
      "epoch": 0.3931879294890947,
      "grad_norm": 0.25304096937179565,
      "learning_rate": 0.00013319897052544577,
      "loss": 0.782,
      "step": 2632
    },
    {
      "epoch": 0.3933373170002988,
      "grad_norm": 0.8042634725570679,
      "learning_rate": 0.00013315463096872287,
      "loss": 0.9017,
      "step": 2633
    },
    {
      "epoch": 0.39348670451150286,
      "grad_norm": 0.14679522812366486,
      "learning_rate": 0.00013311028408761688,
      "loss": 0.6606,
      "step": 2634
    },
    {
      "epoch": 0.3936360920227069,
      "grad_norm": 0.16115805506706238,
      "learning_rate": 0.0001330659298919247,
      "loss": 0.576,
      "step": 2635
    },
    {
      "epoch": 0.39378547953391096,
      "grad_norm": 0.1483192890882492,
      "learning_rate": 0.00013302156839144484,
      "loss": 0.7323,
      "step": 2636
    },
    {
      "epoch": 0.39393486704511504,
      "grad_norm": 0.16038136184215546,
      "learning_rate": 0.0001329771995959775,
      "loss": 0.7193,
      "step": 2637
    },
    {
      "epoch": 0.39408425455631907,
      "grad_norm": 0.14783136546611786,
      "learning_rate": 0.00013293282351532442,
      "loss": 0.762,
      "step": 2638
    },
    {
      "epoch": 0.39423364206752315,
      "grad_norm": 0.35292062163352966,
      "learning_rate": 0.00013288844015928901,
      "loss": 0.9694,
      "step": 2639
    },
    {
      "epoch": 0.39438302957872723,
      "grad_norm": 0.1505533903837204,
      "learning_rate": 0.00013284404953767625,
      "loss": 0.7306,
      "step": 2640
    },
    {
      "epoch": 0.3945324170899313,
      "grad_norm": 0.11992014199495316,
      "learning_rate": 0.00013279965166029274,
      "loss": 0.6739,
      "step": 2641
    },
    {
      "epoch": 0.39468180460113533,
      "grad_norm": 0.161918506026268,
      "learning_rate": 0.00013275524653694665,
      "loss": 0.6941,
      "step": 2642
    },
    {
      "epoch": 0.3948311921123394,
      "grad_norm": 0.16933605074882507,
      "learning_rate": 0.00013271083417744776,
      "loss": 0.52,
      "step": 2643
    },
    {
      "epoch": 0.3949805796235435,
      "grad_norm": 0.14251480996608734,
      "learning_rate": 0.00013266641459160753,
      "loss": 0.5986,
      "step": 2644
    },
    {
      "epoch": 0.3951299671347475,
      "grad_norm": 0.19696931540966034,
      "learning_rate": 0.00013262198778923886,
      "loss": 1.09,
      "step": 2645
    },
    {
      "epoch": 0.3952793546459516,
      "grad_norm": 0.2510409951210022,
      "learning_rate": 0.0001325775537801564,
      "loss": 0.6033,
      "step": 2646
    },
    {
      "epoch": 0.3954287421571557,
      "grad_norm": 0.2563779354095459,
      "learning_rate": 0.0001325331125741763,
      "loss": 1.0942,
      "step": 2647
    },
    {
      "epoch": 0.3955781296683597,
      "grad_norm": 0.26792702078819275,
      "learning_rate": 0.00013248866418111635,
      "loss": 0.7593,
      "step": 2648
    },
    {
      "epoch": 0.3957275171795638,
      "grad_norm": 0.17927950620651245,
      "learning_rate": 0.00013244420861079588,
      "loss": 1.0743,
      "step": 2649
    },
    {
      "epoch": 0.39587690469076786,
      "grad_norm": 0.1726602166891098,
      "learning_rate": 0.00013239974587303584,
      "loss": 0.6306,
      "step": 2650
    },
    {
      "epoch": 0.39602629220197194,
      "grad_norm": 0.21016177535057068,
      "learning_rate": 0.00013235527597765876,
      "loss": 0.833,
      "step": 2651
    },
    {
      "epoch": 0.39617567971317597,
      "grad_norm": 0.1278001368045807,
      "learning_rate": 0.00013231079893448873,
      "loss": 0.7121,
      "step": 2652
    },
    {
      "epoch": 0.39632506722438005,
      "grad_norm": 0.1910894513130188,
      "learning_rate": 0.00013226631475335142,
      "loss": 0.719,
      "step": 2653
    },
    {
      "epoch": 0.3964744547355841,
      "grad_norm": 0.125779390335083,
      "learning_rate": 0.00013222182344407415,
      "loss": 0.7768,
      "step": 2654
    },
    {
      "epoch": 0.39662384224678815,
      "grad_norm": 0.1998329758644104,
      "learning_rate": 0.00013217732501648568,
      "loss": 0.8989,
      "step": 2655
    },
    {
      "epoch": 0.39677322975799223,
      "grad_norm": 0.13692845404148102,
      "learning_rate": 0.00013213281948041647,
      "loss": 0.8524,
      "step": 2656
    },
    {
      "epoch": 0.3969226172691963,
      "grad_norm": 0.12631644308567047,
      "learning_rate": 0.00013208830684569847,
      "loss": 0.5603,
      "step": 2657
    },
    {
      "epoch": 0.39707200478040033,
      "grad_norm": 0.11097691208124161,
      "learning_rate": 0.0001320437871221652,
      "loss": 0.6293,
      "step": 2658
    },
    {
      "epoch": 0.3972213922916044,
      "grad_norm": 0.17241108417510986,
      "learning_rate": 0.00013199926031965187,
      "loss": 0.7668,
      "step": 2659
    },
    {
      "epoch": 0.3973707798028085,
      "grad_norm": 0.12255235761404037,
      "learning_rate": 0.00013195472644799504,
      "loss": 0.7087,
      "step": 2660
    },
    {
      "epoch": 0.3975201673140126,
      "grad_norm": 0.11954877525568008,
      "learning_rate": 0.00013191018551703305,
      "loss": 0.6361,
      "step": 2661
    },
    {
      "epoch": 0.3976695548252166,
      "grad_norm": 0.12530867755413055,
      "learning_rate": 0.00013186563753660562,
      "loss": 0.7217,
      "step": 2662
    },
    {
      "epoch": 0.3978189423364207,
      "grad_norm": 0.12268290668725967,
      "learning_rate": 0.00013182108251655415,
      "loss": 0.475,
      "step": 2663
    },
    {
      "epoch": 0.39796832984762476,
      "grad_norm": 0.1455891728401184,
      "learning_rate": 0.0001317765204667215,
      "loss": 0.7436,
      "step": 2664
    },
    {
      "epoch": 0.3981177173588288,
      "grad_norm": 0.1256171613931656,
      "learning_rate": 0.0001317319513969522,
      "loss": 0.544,
      "step": 2665
    },
    {
      "epoch": 0.39826710487003286,
      "grad_norm": 0.20818553864955902,
      "learning_rate": 0.0001316873753170922,
      "loss": 1.0755,
      "step": 2666
    },
    {
      "epoch": 0.39841649238123694,
      "grad_norm": 0.2578659653663635,
      "learning_rate": 0.0001316427922369891,
      "loss": 0.9188,
      "step": 2667
    },
    {
      "epoch": 0.39856587989244097,
      "grad_norm": 0.1123015433549881,
      "learning_rate": 0.00013159820216649198,
      "loss": 0.6371,
      "step": 2668
    },
    {
      "epoch": 0.39871526740364505,
      "grad_norm": 0.14669334888458252,
      "learning_rate": 0.00013155360511545155,
      "loss": 0.6954,
      "step": 2669
    },
    {
      "epoch": 0.3988646549148491,
      "grad_norm": 0.14442336559295654,
      "learning_rate": 0.00013150900109371998,
      "loss": 0.409,
      "step": 2670
    },
    {
      "epoch": 0.3990140424260532,
      "grad_norm": 0.18954886496067047,
      "learning_rate": 0.00013146439011115096,
      "loss": 0.8776,
      "step": 2671
    },
    {
      "epoch": 0.39916342993725723,
      "grad_norm": 0.15388745069503784,
      "learning_rate": 0.00013141977217759977,
      "loss": 0.8512,
      "step": 2672
    },
    {
      "epoch": 0.3993128174484613,
      "grad_norm": 0.1552528738975525,
      "learning_rate": 0.00013137514730292326,
      "loss": 0.7248,
      "step": 2673
    },
    {
      "epoch": 0.3994622049596654,
      "grad_norm": 0.17782266438007355,
      "learning_rate": 0.00013133051549697977,
      "loss": 0.8985,
      "step": 2674
    },
    {
      "epoch": 0.3996115924708694,
      "grad_norm": 0.13761469721794128,
      "learning_rate": 0.00013128587676962912,
      "loss": 0.8545,
      "step": 2675
    },
    {
      "epoch": 0.3997609799820735,
      "grad_norm": 0.1979491114616394,
      "learning_rate": 0.00013124123113073278,
      "loss": 0.9282,
      "step": 2676
    },
    {
      "epoch": 0.3999103674932776,
      "grad_norm": 0.1521802395582199,
      "learning_rate": 0.0001311965785901537,
      "loss": 0.7071,
      "step": 2677
    },
    {
      "epoch": 0.4000597550044816,
      "grad_norm": 0.21194405853748322,
      "learning_rate": 0.0001311519191577562,
      "loss": 0.9731,
      "step": 2678
    },
    {
      "epoch": 0.4002091425156857,
      "grad_norm": 0.16967430710792542,
      "learning_rate": 0.00013110725284340636,
      "loss": 0.7712,
      "step": 2679
    },
    {
      "epoch": 0.40035853002688976,
      "grad_norm": 0.21047207713127136,
      "learning_rate": 0.0001310625796569717,
      "loss": 0.9286,
      "step": 2680
    },
    {
      "epoch": 0.40050791753809384,
      "grad_norm": 0.14066767692565918,
      "learning_rate": 0.00013101789960832111,
      "loss": 0.5469,
      "step": 2681
    },
    {
      "epoch": 0.40065730504929786,
      "grad_norm": 0.15885429084300995,
      "learning_rate": 0.00013097321270732524,
      "loss": 0.8327,
      "step": 2682
    },
    {
      "epoch": 0.40080669256050194,
      "grad_norm": 0.18348488211631775,
      "learning_rate": 0.00013092851896385606,
      "loss": 0.6746,
      "step": 2683
    },
    {
      "epoch": 0.400956080071706,
      "grad_norm": 0.182549849152565,
      "learning_rate": 0.0001308838183877872,
      "loss": 0.9097,
      "step": 2684
    },
    {
      "epoch": 0.40110546758291005,
      "grad_norm": 0.1817612498998642,
      "learning_rate": 0.00013083911098899366,
      "loss": 0.8671,
      "step": 2685
    },
    {
      "epoch": 0.4012548550941141,
      "grad_norm": 0.5262943506240845,
      "learning_rate": 0.00013079439677735207,
      "loss": 1.075,
      "step": 2686
    },
    {
      "epoch": 0.4014042426053182,
      "grad_norm": 0.13640695810317993,
      "learning_rate": 0.00013074967576274043,
      "loss": 0.7433,
      "step": 2687
    },
    {
      "epoch": 0.40155363011652223,
      "grad_norm": 0.22469180822372437,
      "learning_rate": 0.0001307049479550384,
      "loss": 0.8477,
      "step": 2688
    },
    {
      "epoch": 0.4017030176277263,
      "grad_norm": 0.24286799132823944,
      "learning_rate": 0.00013066021336412698,
      "loss": 0.911,
      "step": 2689
    },
    {
      "epoch": 0.4018524051389304,
      "grad_norm": 0.16932006180286407,
      "learning_rate": 0.00013061547199988885,
      "loss": 0.8501,
      "step": 2690
    },
    {
      "epoch": 0.40200179265013447,
      "grad_norm": 0.1616639643907547,
      "learning_rate": 0.00013057072387220802,
      "loss": 0.7756,
      "step": 2691
    },
    {
      "epoch": 0.4021511801613385,
      "grad_norm": 0.6991116404533386,
      "learning_rate": 0.00013052596899097005,
      "loss": 1.6588,
      "step": 2692
    },
    {
      "epoch": 0.4023005676725426,
      "grad_norm": 0.1160588413476944,
      "learning_rate": 0.000130481207366062,
      "loss": 0.6841,
      "step": 2693
    },
    {
      "epoch": 0.40244995518374665,
      "grad_norm": 0.13019360601902008,
      "learning_rate": 0.0001304364390073725,
      "loss": 0.6629,
      "step": 2694
    },
    {
      "epoch": 0.4025993426949507,
      "grad_norm": 0.12842808663845062,
      "learning_rate": 0.0001303916639247915,
      "loss": 0.5842,
      "step": 2695
    },
    {
      "epoch": 0.40274873020615476,
      "grad_norm": 0.16104257106781006,
      "learning_rate": 0.00013034688212821058,
      "loss": 0.7371,
      "step": 2696
    },
    {
      "epoch": 0.40289811771735884,
      "grad_norm": 0.13033579289913177,
      "learning_rate": 0.00013030209362752273,
      "loss": 0.7908,
      "step": 2697
    },
    {
      "epoch": 0.4030475052285629,
      "grad_norm": 0.16546832025051117,
      "learning_rate": 0.00013025729843262241,
      "loss": 0.8029,
      "step": 2698
    },
    {
      "epoch": 0.40319689273976694,
      "grad_norm": 0.13665848970413208,
      "learning_rate": 0.00013021249655340565,
      "loss": 0.5085,
      "step": 2699
    },
    {
      "epoch": 0.403346280250971,
      "grad_norm": 0.15850643813610077,
      "learning_rate": 0.00013016768799976983,
      "loss": 0.9416,
      "step": 2700
    },
    {
      "epoch": 0.4034956677621751,
      "grad_norm": 0.13148969411849976,
      "learning_rate": 0.0001301228727816139,
      "loss": 0.7612,
      "step": 2701
    },
    {
      "epoch": 0.4036450552733791,
      "grad_norm": 0.35034894943237305,
      "learning_rate": 0.00013007805090883826,
      "loss": 0.8406,
      "step": 2702
    },
    {
      "epoch": 0.4037944427845832,
      "grad_norm": 0.15132221579551697,
      "learning_rate": 0.00013003322239134475,
      "loss": 0.763,
      "step": 2703
    },
    {
      "epoch": 0.4039438302957873,
      "grad_norm": 0.10346682369709015,
      "learning_rate": 0.00012998838723903675,
      "loss": 0.579,
      "step": 2704
    },
    {
      "epoch": 0.4040932178069913,
      "grad_norm": 0.17298509180545807,
      "learning_rate": 0.00012994354546181898,
      "loss": 0.675,
      "step": 2705
    },
    {
      "epoch": 0.4042426053181954,
      "grad_norm": 0.20535051822662354,
      "learning_rate": 0.00012989869706959777,
      "loss": 0.5868,
      "step": 2706
    },
    {
      "epoch": 0.40439199282939947,
      "grad_norm": 0.20123709738254547,
      "learning_rate": 0.00012985384207228078,
      "loss": 0.5872,
      "step": 2707
    },
    {
      "epoch": 0.40454138034060355,
      "grad_norm": 0.18401536345481873,
      "learning_rate": 0.0001298089804797772,
      "loss": 0.7643,
      "step": 2708
    },
    {
      "epoch": 0.4046907678518076,
      "grad_norm": 0.16344019770622253,
      "learning_rate": 0.00012976411230199774,
      "loss": 0.7011,
      "step": 2709
    },
    {
      "epoch": 0.40484015536301166,
      "grad_norm": 0.12780562043190002,
      "learning_rate": 0.00012971923754885438,
      "loss": 0.5972,
      "step": 2710
    },
    {
      "epoch": 0.40498954287421574,
      "grad_norm": 0.14979572594165802,
      "learning_rate": 0.0001296743562302607,
      "loss": 0.584,
      "step": 2711
    },
    {
      "epoch": 0.40513893038541976,
      "grad_norm": 0.13349317014217377,
      "learning_rate": 0.0001296294683561318,
      "loss": 0.8066,
      "step": 2712
    },
    {
      "epoch": 0.40528831789662384,
      "grad_norm": 0.2305103987455368,
      "learning_rate": 0.00012958457393638396,
      "loss": 0.7833,
      "step": 2713
    },
    {
      "epoch": 0.4054377054078279,
      "grad_norm": 0.13477012515068054,
      "learning_rate": 0.00012953967298093513,
      "loss": 0.8807,
      "step": 2714
    },
    {
      "epoch": 0.40558709291903194,
      "grad_norm": 0.2081054449081421,
      "learning_rate": 0.00012949476549970474,
      "loss": 1.1975,
      "step": 2715
    },
    {
      "epoch": 0.405736480430236,
      "grad_norm": 0.1232752725481987,
      "learning_rate": 0.00012944985150261341,
      "loss": 0.6946,
      "step": 2716
    },
    {
      "epoch": 0.4058858679414401,
      "grad_norm": 0.16085590422153473,
      "learning_rate": 0.00012940493099958343,
      "loss": 1.2668,
      "step": 2717
    },
    {
      "epoch": 0.4060352554526442,
      "grad_norm": 0.11771124601364136,
      "learning_rate": 0.00012936000400053845,
      "loss": 0.6682,
      "step": 2718
    },
    {
      "epoch": 0.4061846429638482,
      "grad_norm": 0.1437877118587494,
      "learning_rate": 0.00012931507051540354,
      "loss": 0.5094,
      "step": 2719
    },
    {
      "epoch": 0.4063340304750523,
      "grad_norm": 0.16264525055885315,
      "learning_rate": 0.00012927013055410522,
      "loss": 0.8985,
      "step": 2720
    },
    {
      "epoch": 0.40648341798625637,
      "grad_norm": 0.16355526447296143,
      "learning_rate": 0.00012922518412657145,
      "loss": 0.5834,
      "step": 2721
    },
    {
      "epoch": 0.4066328054974604,
      "grad_norm": 0.7323223352432251,
      "learning_rate": 0.00012918023124273165,
      "loss": 1.4513,
      "step": 2722
    },
    {
      "epoch": 0.40678219300866447,
      "grad_norm": 0.31279024481773376,
      "learning_rate": 0.0001291352719125165,
      "loss": 1.275,
      "step": 2723
    },
    {
      "epoch": 0.40693158051986855,
      "grad_norm": 0.16710931062698364,
      "learning_rate": 0.00012909030614585836,
      "loss": 0.6563,
      "step": 2724
    },
    {
      "epoch": 0.4070809680310726,
      "grad_norm": 0.4445287883281708,
      "learning_rate": 0.00012904533395269083,
      "loss": 1.105,
      "step": 2725
    },
    {
      "epoch": 0.40723035554227666,
      "grad_norm": 0.2966986298561096,
      "learning_rate": 0.00012900035534294893,
      "loss": 1.1945,
      "step": 2726
    },
    {
      "epoch": 0.40737974305348074,
      "grad_norm": 0.12481849640607834,
      "learning_rate": 0.00012895537032656923,
      "loss": 0.6538,
      "step": 2727
    },
    {
      "epoch": 0.4075291305646848,
      "grad_norm": 0.15355361998081207,
      "learning_rate": 0.00012891037891348957,
      "loss": 0.8507,
      "step": 2728
    },
    {
      "epoch": 0.40767851807588884,
      "grad_norm": 0.2790848910808563,
      "learning_rate": 0.00012886538111364933,
      "loss": 1.004,
      "step": 2729
    },
    {
      "epoch": 0.4078279055870929,
      "grad_norm": 0.36557501554489136,
      "learning_rate": 0.00012882037693698917,
      "loss": 1.0527,
      "step": 2730
    },
    {
      "epoch": 0.407977293098297,
      "grad_norm": 0.1678098440170288,
      "learning_rate": 0.00012877536639345122,
      "loss": 0.5943,
      "step": 2731
    },
    {
      "epoch": 0.408126680609501,
      "grad_norm": 0.16713346540927887,
      "learning_rate": 0.00012873034949297912,
      "loss": 0.5807,
      "step": 2732
    },
    {
      "epoch": 0.4082760681207051,
      "grad_norm": 0.11321716755628586,
      "learning_rate": 0.00012868532624551773,
      "loss": 0.656,
      "step": 2733
    },
    {
      "epoch": 0.4084254556319092,
      "grad_norm": 0.12688088417053223,
      "learning_rate": 0.0001286402966610134,
      "loss": 0.7693,
      "step": 2734
    },
    {
      "epoch": 0.4085748431431132,
      "grad_norm": 0.12967096269130707,
      "learning_rate": 0.00012859526074941393,
      "loss": 0.7154,
      "step": 2735
    },
    {
      "epoch": 0.4087242306543173,
      "grad_norm": 0.14440783858299255,
      "learning_rate": 0.00012855021852066842,
      "loss": 0.7999,
      "step": 2736
    },
    {
      "epoch": 0.40887361816552137,
      "grad_norm": 0.08039014786481857,
      "learning_rate": 0.00012850516998472742,
      "loss": 0.3766,
      "step": 2737
    },
    {
      "epoch": 0.40902300567672545,
      "grad_norm": 0.1711304783821106,
      "learning_rate": 0.00012846011515154287,
      "loss": 0.894,
      "step": 2738
    },
    {
      "epoch": 0.4091723931879295,
      "grad_norm": 0.1515461653470993,
      "learning_rate": 0.0001284150540310681,
      "loss": 0.8942,
      "step": 2739
    },
    {
      "epoch": 0.40932178069913355,
      "grad_norm": 0.2400161474943161,
      "learning_rate": 0.00012836998663325782,
      "loss": 0.6129,
      "step": 2740
    },
    {
      "epoch": 0.40947116821033763,
      "grad_norm": 0.15345792472362518,
      "learning_rate": 0.00012832491296806817,
      "loss": 0.7611,
      "step": 2741
    },
    {
      "epoch": 0.40962055572154166,
      "grad_norm": 0.1785755753517151,
      "learning_rate": 0.00012827983304545656,
      "loss": 0.6098,
      "step": 2742
    },
    {
      "epoch": 0.40976994323274574,
      "grad_norm": 0.17453451454639435,
      "learning_rate": 0.0001282347468753819,
      "loss": 0.6678,
      "step": 2743
    },
    {
      "epoch": 0.4099193307439498,
      "grad_norm": 0.19147516787052155,
      "learning_rate": 0.00012818965446780448,
      "loss": 0.5183,
      "step": 2744
    },
    {
      "epoch": 0.4100687182551539,
      "grad_norm": 0.21111948788166046,
      "learning_rate": 0.00012814455583268588,
      "loss": 0.7062,
      "step": 2745
    },
    {
      "epoch": 0.4102181057663579,
      "grad_norm": 0.17327934503555298,
      "learning_rate": 0.00012809945097998907,
      "loss": 0.7819,
      "step": 2746
    },
    {
      "epoch": 0.410367493277562,
      "grad_norm": 0.12339986115694046,
      "learning_rate": 0.00012805433991967848,
      "loss": 0.7196,
      "step": 2747
    },
    {
      "epoch": 0.4105168807887661,
      "grad_norm": 0.18432828783988953,
      "learning_rate": 0.00012800922266171987,
      "loss": 0.7388,
      "step": 2748
    },
    {
      "epoch": 0.4106662682999701,
      "grad_norm": 0.34459012746810913,
      "learning_rate": 0.0001279640992160803,
      "loss": 0.8311,
      "step": 2749
    },
    {
      "epoch": 0.4108156558111742,
      "grad_norm": 0.2298446148633957,
      "learning_rate": 0.0001279189695927283,
      "loss": 0.8606,
      "step": 2750
    },
    {
      "epoch": 0.41096504332237827,
      "grad_norm": 0.3446527123451233,
      "learning_rate": 0.00012787383380163372,
      "loss": 0.9635,
      "step": 2751
    },
    {
      "epoch": 0.4111144308335823,
      "grad_norm": 0.20561830699443817,
      "learning_rate": 0.0001278286918527677,
      "loss": 0.4813,
      "step": 2752
    },
    {
      "epoch": 0.41126381834478637,
      "grad_norm": 0.16291874647140503,
      "learning_rate": 0.00012778354375610294,
      "loss": 0.5007,
      "step": 2753
    },
    {
      "epoch": 0.41141320585599045,
      "grad_norm": 0.14061903953552246,
      "learning_rate": 0.00012773838952161322,
      "loss": 1.0204,
      "step": 2754
    },
    {
      "epoch": 0.41156259336719453,
      "grad_norm": 0.14922526478767395,
      "learning_rate": 0.00012769322915927395,
      "loss": 0.5922,
      "step": 2755
    },
    {
      "epoch": 0.41171198087839855,
      "grad_norm": 0.13793185353279114,
      "learning_rate": 0.0001276480626790617,
      "loss": 0.9527,
      "step": 2756
    },
    {
      "epoch": 0.41186136838960263,
      "grad_norm": 0.23414811491966248,
      "learning_rate": 0.0001276028900909545,
      "loss": 0.8249,
      "step": 2757
    },
    {
      "epoch": 0.4120107559008067,
      "grad_norm": 0.20121406018733978,
      "learning_rate": 0.00012755771140493167,
      "loss": 0.9573,
      "step": 2758
    },
    {
      "epoch": 0.41216014341201074,
      "grad_norm": 0.21266387403011322,
      "learning_rate": 0.0001275125266309739,
      "loss": 0.6615,
      "step": 2759
    },
    {
      "epoch": 0.4123095309232148,
      "grad_norm": 0.12896525859832764,
      "learning_rate": 0.0001274673357790632,
      "loss": 0.3989,
      "step": 2760
    },
    {
      "epoch": 0.4124589184344189,
      "grad_norm": 5.503906726837158,
      "learning_rate": 0.000127422138859183,
      "loss": 3.1524,
      "step": 2761
    },
    {
      "epoch": 0.4126083059456229,
      "grad_norm": 0.1865500658750534,
      "learning_rate": 0.00012737693588131793,
      "loss": 0.6656,
      "step": 2762
    },
    {
      "epoch": 0.412757693456827,
      "grad_norm": 0.18052983283996582,
      "learning_rate": 0.00012733172685545414,
      "loss": 0.5917,
      "step": 2763
    },
    {
      "epoch": 0.4129070809680311,
      "grad_norm": 0.17281179130077362,
      "learning_rate": 0.00012728651179157895,
      "loss": 0.7881,
      "step": 2764
    },
    {
      "epoch": 0.41305646847923516,
      "grad_norm": 0.1971537321805954,
      "learning_rate": 0.00012724129069968112,
      "loss": 0.7932,
      "step": 2765
    },
    {
      "epoch": 0.4132058559904392,
      "grad_norm": 0.20541293919086456,
      "learning_rate": 0.00012719606358975073,
      "loss": 0.7341,
      "step": 2766
    },
    {
      "epoch": 0.41335524350164327,
      "grad_norm": 0.2898564040660858,
      "learning_rate": 0.0001271508304717791,
      "loss": 0.8166,
      "step": 2767
    },
    {
      "epoch": 0.41350463101284735,
      "grad_norm": 0.28142088651657104,
      "learning_rate": 0.00012710559135575895,
      "loss": 0.8473,
      "step": 2768
    },
    {
      "epoch": 0.41365401852405137,
      "grad_norm": 0.14628003537654877,
      "learning_rate": 0.00012706034625168435,
      "loss": 0.6946,
      "step": 2769
    },
    {
      "epoch": 0.41380340603525545,
      "grad_norm": 0.14080846309661865,
      "learning_rate": 0.00012701509516955067,
      "loss": 0.7228,
      "step": 2770
    },
    {
      "epoch": 0.41395279354645953,
      "grad_norm": 0.3143552243709564,
      "learning_rate": 0.00012696983811935457,
      "loss": 0.6538,
      "step": 2771
    },
    {
      "epoch": 0.41410218105766355,
      "grad_norm": 0.15786129236221313,
      "learning_rate": 0.00012692457511109402,
      "loss": 0.7463,
      "step": 2772
    },
    {
      "epoch": 0.41425156856886763,
      "grad_norm": 0.1806403249502182,
      "learning_rate": 0.00012687930615476838,
      "loss": 0.9242,
      "step": 2773
    },
    {
      "epoch": 0.4144009560800717,
      "grad_norm": 0.188722163438797,
      "learning_rate": 0.00012683403126037825,
      "loss": 0.913,
      "step": 2774
    },
    {
      "epoch": 0.4145503435912758,
      "grad_norm": 0.1534443199634552,
      "learning_rate": 0.0001267887504379256,
      "loss": 0.7678,
      "step": 2775
    },
    {
      "epoch": 0.4146997311024798,
      "grad_norm": 0.17588689923286438,
      "learning_rate": 0.00012674346369741365,
      "loss": 0.3719,
      "step": 2776
    },
    {
      "epoch": 0.4148491186136839,
      "grad_norm": 0.22392036020755768,
      "learning_rate": 0.00012669817104884703,
      "loss": 0.8156,
      "step": 2777
    },
    {
      "epoch": 0.414998506124888,
      "grad_norm": 0.1883402168750763,
      "learning_rate": 0.0001266528725022315,
      "loss": 0.7338,
      "step": 2778
    },
    {
      "epoch": 0.415147893636092,
      "grad_norm": 0.13259749114513397,
      "learning_rate": 0.00012660756806757428,
      "loss": 0.7613,
      "step": 2779
    },
    {
      "epoch": 0.4152972811472961,
      "grad_norm": 0.13270725309848785,
      "learning_rate": 0.00012656225775488383,
      "loss": 0.8243,
      "step": 2780
    },
    {
      "epoch": 0.41544666865850016,
      "grad_norm": 0.22006285190582275,
      "learning_rate": 0.00012651694157416994,
      "loss": 0.7873,
      "step": 2781
    },
    {
      "epoch": 0.4155960561697042,
      "grad_norm": 0.19291041791439056,
      "learning_rate": 0.0001264716195354436,
      "loss": 0.8715,
      "step": 2782
    },
    {
      "epoch": 0.41574544368090827,
      "grad_norm": 0.12593582272529602,
      "learning_rate": 0.00012642629164871726,
      "loss": 0.6518,
      "step": 2783
    },
    {
      "epoch": 0.41589483119211235,
      "grad_norm": 0.6027681231498718,
      "learning_rate": 0.00012638095792400452,
      "loss": 1.314,
      "step": 2784
    },
    {
      "epoch": 0.4160442187033164,
      "grad_norm": 0.11635372042655945,
      "learning_rate": 0.0001263356183713203,
      "loss": 0.7563,
      "step": 2785
    },
    {
      "epoch": 0.41619360621452045,
      "grad_norm": 0.1896970570087433,
      "learning_rate": 0.00012629027300068088,
      "loss": 0.9477,
      "step": 2786
    },
    {
      "epoch": 0.41634299372572453,
      "grad_norm": 0.4362841248512268,
      "learning_rate": 0.00012624492182210372,
      "loss": 0.9925,
      "step": 2787
    },
    {
      "epoch": 0.4164923812369286,
      "grad_norm": 0.18889276683330536,
      "learning_rate": 0.0001261995648456076,
      "loss": 0.7465,
      "step": 2788
    },
    {
      "epoch": 0.41664176874813263,
      "grad_norm": 0.3795793652534485,
      "learning_rate": 0.00012615420208121267,
      "loss": 0.9248,
      "step": 2789
    },
    {
      "epoch": 0.4167911562593367,
      "grad_norm": 0.11698241531848907,
      "learning_rate": 0.00012610883353894026,
      "loss": 0.6894,
      "step": 2790
    },
    {
      "epoch": 0.4169405437705408,
      "grad_norm": 0.15589016675949097,
      "learning_rate": 0.0001260634592288129,
      "loss": 0.4277,
      "step": 2791
    },
    {
      "epoch": 0.4170899312817448,
      "grad_norm": 0.14596420526504517,
      "learning_rate": 0.00012601807916085461,
      "loss": 0.8346,
      "step": 2792
    },
    {
      "epoch": 0.4172393187929489,
      "grad_norm": 0.18475967645645142,
      "learning_rate": 0.00012597269334509056,
      "loss": 0.7691,
      "step": 2793
    },
    {
      "epoch": 0.417388706304153,
      "grad_norm": 0.11927938461303711,
      "learning_rate": 0.00012592730179154712,
      "loss": 0.4278,
      "step": 2794
    },
    {
      "epoch": 0.41753809381535706,
      "grad_norm": 0.13390043377876282,
      "learning_rate": 0.00012588190451025207,
      "loss": 0.7773,
      "step": 2795
    },
    {
      "epoch": 0.4176874813265611,
      "grad_norm": 0.1760786473751068,
      "learning_rate": 0.0001258365015112344,
      "loss": 0.8258,
      "step": 2796
    },
    {
      "epoch": 0.41783686883776516,
      "grad_norm": 0.1638936996459961,
      "learning_rate": 0.00012579109280452425,
      "loss": 1.0548,
      "step": 2797
    },
    {
      "epoch": 0.41798625634896924,
      "grad_norm": 0.1905653178691864,
      "learning_rate": 0.00012574567840015324,
      "loss": 0.7376,
      "step": 2798
    },
    {
      "epoch": 0.41813564386017327,
      "grad_norm": 0.2321435511112213,
      "learning_rate": 0.00012570025830815405,
      "loss": 1.1223,
      "step": 2799
    },
    {
      "epoch": 0.41828503137137735,
      "grad_norm": 0.186323344707489,
      "learning_rate": 0.00012565483253856071,
      "loss": 0.9795,
      "step": 2800
    },
    {
      "epoch": 0.4184344188825814,
      "grad_norm": 0.15830577909946442,
      "learning_rate": 0.00012560940110140854,
      "loss": 0.5669,
      "step": 2801
    },
    {
      "epoch": 0.4185838063937855,
      "grad_norm": 0.1189170703291893,
      "learning_rate": 0.00012556396400673403,
      "loss": 0.7681,
      "step": 2802
    },
    {
      "epoch": 0.41873319390498953,
      "grad_norm": 0.2227000743150711,
      "learning_rate": 0.000125518521264575,
      "loss": 1.0181,
      "step": 2803
    },
    {
      "epoch": 0.4188825814161936,
      "grad_norm": 0.17998264729976654,
      "learning_rate": 0.00012547307288497035,
      "loss": 0.5145,
      "step": 2804
    },
    {
      "epoch": 0.4190319689273977,
      "grad_norm": 0.2130528837442398,
      "learning_rate": 0.0001254276188779605,
      "loss": 0.9355,
      "step": 2805
    },
    {
      "epoch": 0.4191813564386017,
      "grad_norm": 0.1300712525844574,
      "learning_rate": 0.00012538215925358688,
      "loss": 0.8792,
      "step": 2806
    },
    {
      "epoch": 0.4193307439498058,
      "grad_norm": 0.18171724677085876,
      "learning_rate": 0.0001253366940218922,
      "loss": 0.8837,
      "step": 2807
    },
    {
      "epoch": 0.4194801314610099,
      "grad_norm": 0.21381869912147522,
      "learning_rate": 0.00012529122319292053,
      "loss": 0.7601,
      "step": 2808
    },
    {
      "epoch": 0.4196295189722139,
      "grad_norm": 0.1427227407693863,
      "learning_rate": 0.00012524574677671705,
      "loss": 0.6711,
      "step": 2809
    },
    {
      "epoch": 0.419778906483418,
      "grad_norm": 0.1249597892165184,
      "learning_rate": 0.00012520026478332822,
      "loss": 0.7086,
      "step": 2810
    },
    {
      "epoch": 0.41992829399462206,
      "grad_norm": 0.11634568125009537,
      "learning_rate": 0.00012515477722280178,
      "loss": 0.5877,
      "step": 2811
    },
    {
      "epoch": 0.42007768150582614,
      "grad_norm": 0.20771601796150208,
      "learning_rate": 0.00012510928410518663,
      "loss": 0.9871,
      "step": 2812
    },
    {
      "epoch": 0.42022706901703016,
      "grad_norm": 0.20207735896110535,
      "learning_rate": 0.0001250637854405329,
      "loss": 0.7579,
      "step": 2813
    },
    {
      "epoch": 0.42037645652823424,
      "grad_norm": 0.15419501066207886,
      "learning_rate": 0.00012501828123889194,
      "loss": 0.6867,
      "step": 2814
    },
    {
      "epoch": 0.4205258440394383,
      "grad_norm": 0.364545613527298,
      "learning_rate": 0.00012497277151031643,
      "loss": 0.8273,
      "step": 2815
    },
    {
      "epoch": 0.42067523155064235,
      "grad_norm": 0.21446232497692108,
      "learning_rate": 0.00012492725626486013,
      "loss": 1.0899,
      "step": 2816
    },
    {
      "epoch": 0.4208246190618464,
      "grad_norm": 0.13538731634616852,
      "learning_rate": 0.0001248817355125781,
      "loss": 0.6771,
      "step": 2817
    },
    {
      "epoch": 0.4209740065730505,
      "grad_norm": 0.3066132366657257,
      "learning_rate": 0.00012483620926352656,
      "loss": 0.7166,
      "step": 2818
    },
    {
      "epoch": 0.42112339408425453,
      "grad_norm": 0.2135513424873352,
      "learning_rate": 0.00012479067752776304,
      "loss": 0.9082,
      "step": 2819
    },
    {
      "epoch": 0.4212727815954586,
      "grad_norm": 0.22986231744289398,
      "learning_rate": 0.00012474514031534617,
      "loss": 1.0247,
      "step": 2820
    },
    {
      "epoch": 0.4214221691066627,
      "grad_norm": 0.12677530944347382,
      "learning_rate": 0.00012469959763633585,
      "loss": 0.6002,
      "step": 2821
    },
    {
      "epoch": 0.42157155661786677,
      "grad_norm": 0.11655603349208832,
      "learning_rate": 0.00012465404950079325,
      "loss": 0.5927,
      "step": 2822
    },
    {
      "epoch": 0.4217209441290708,
      "grad_norm": 0.2599896490573883,
      "learning_rate": 0.00012460849591878055,
      "loss": 0.7353,
      "step": 2823
    },
    {
      "epoch": 0.4218703316402749,
      "grad_norm": 0.13026916980743408,
      "learning_rate": 0.00012456293690036135,
      "loss": 0.7728,
      "step": 2824
    },
    {
      "epoch": 0.42201971915147896,
      "grad_norm": 0.15472781658172607,
      "learning_rate": 0.00012451737245560037,
      "loss": 0.7458,
      "step": 2825
    },
    {
      "epoch": 0.422169106662683,
      "grad_norm": 0.13531015813350677,
      "learning_rate": 0.00012447180259456342,
      "loss": 0.9021,
      "step": 2826
    },
    {
      "epoch": 0.42231849417388706,
      "grad_norm": 0.15706327557563782,
      "learning_rate": 0.00012442622732731766,
      "loss": 0.8921,
      "step": 2827
    },
    {
      "epoch": 0.42246788168509114,
      "grad_norm": 0.2359904795885086,
      "learning_rate": 0.00012438064666393144,
      "loss": 1.0115,
      "step": 2828
    },
    {
      "epoch": 0.42261726919629516,
      "grad_norm": 0.31738442182540894,
      "learning_rate": 0.00012433506061447418,
      "loss": 0.898,
      "step": 2829
    },
    {
      "epoch": 0.42276665670749924,
      "grad_norm": 0.13776662945747375,
      "learning_rate": 0.00012428946918901655,
      "loss": 0.601,
      "step": 2830
    },
    {
      "epoch": 0.4229160442187033,
      "grad_norm": 0.15527896583080292,
      "learning_rate": 0.0001242438723976305,
      "loss": 0.9006,
      "step": 2831
    },
    {
      "epoch": 0.4230654317299074,
      "grad_norm": 0.1306951344013214,
      "learning_rate": 0.00012419827025038905,
      "loss": 0.8107,
      "step": 2832
    },
    {
      "epoch": 0.42321481924111143,
      "grad_norm": 0.14317989349365234,
      "learning_rate": 0.00012415266275736636,
      "loss": 0.8741,
      "step": 2833
    },
    {
      "epoch": 0.4233642067523155,
      "grad_norm": 0.14560556411743164,
      "learning_rate": 0.00012410704992863792,
      "loss": 0.7669,
      "step": 2834
    },
    {
      "epoch": 0.4235135942635196,
      "grad_norm": 0.16069158911705017,
      "learning_rate": 0.00012406143177428033,
      "loss": 0.4259,
      "step": 2835
    },
    {
      "epoch": 0.4236629817747236,
      "grad_norm": 0.13197310268878937,
      "learning_rate": 0.00012401580830437135,
      "loss": 0.6307,
      "step": 2836
    },
    {
      "epoch": 0.4238123692859277,
      "grad_norm": 0.12921732664108276,
      "learning_rate": 0.00012397017952898992,
      "loss": 0.6366,
      "step": 2837
    },
    {
      "epoch": 0.4239617567971318,
      "grad_norm": 0.1424647867679596,
      "learning_rate": 0.0001239245454582162,
      "loss": 0.4977,
      "step": 2838
    },
    {
      "epoch": 0.4241111443083358,
      "grad_norm": 0.16429497301578522,
      "learning_rate": 0.0001238789061021314,
      "loss": 0.6552,
      "step": 2839
    },
    {
      "epoch": 0.4242605318195399,
      "grad_norm": 0.14486564695835114,
      "learning_rate": 0.000123833261470818,
      "loss": 1.0555,
      "step": 2840
    },
    {
      "epoch": 0.42440991933074396,
      "grad_norm": 0.41415485739707947,
      "learning_rate": 0.00012378761157435972,
      "loss": 1.4948,
      "step": 2841
    },
    {
      "epoch": 0.42455930684194804,
      "grad_norm": 0.12499793618917465,
      "learning_rate": 0.0001237419564228412,
      "loss": 0.7543,
      "step": 2842
    },
    {
      "epoch": 0.42470869435315206,
      "grad_norm": 0.14574459195137024,
      "learning_rate": 0.00012369629602634847,
      "loss": 0.7935,
      "step": 2843
    },
    {
      "epoch": 0.42485808186435614,
      "grad_norm": 0.13071438670158386,
      "learning_rate": 0.00012365063039496862,
      "loss": 0.6842,
      "step": 2844
    },
    {
      "epoch": 0.4250074693755602,
      "grad_norm": 0.16986867785453796,
      "learning_rate": 0.0001236049595387899,
      "loss": 1.0144,
      "step": 2845
    },
    {
      "epoch": 0.42515685688676425,
      "grad_norm": 0.13037452101707458,
      "learning_rate": 0.00012355928346790174,
      "loss": 1.0499,
      "step": 2846
    },
    {
      "epoch": 0.4253062443979683,
      "grad_norm": 0.24295803904533386,
      "learning_rate": 0.0001235136021923947,
      "loss": 0.502,
      "step": 2847
    },
    {
      "epoch": 0.4254556319091724,
      "grad_norm": 0.14616192877292633,
      "learning_rate": 0.0001234679157223605,
      "loss": 0.7216,
      "step": 2848
    },
    {
      "epoch": 0.42560501942037643,
      "grad_norm": 0.1372319906949997,
      "learning_rate": 0.00012342222406789205,
      "loss": 0.8509,
      "step": 2849
    },
    {
      "epoch": 0.4257544069315805,
      "grad_norm": 0.17448590695858002,
      "learning_rate": 0.00012337652723908325,
      "loss": 0.6277,
      "step": 2850
    },
    {
      "epoch": 0.4259037944427846,
      "grad_norm": 0.15089821815490723,
      "learning_rate": 0.00012333082524602936,
      "loss": 0.7324,
      "step": 2851
    },
    {
      "epoch": 0.42605318195398867,
      "grad_norm": 0.20889997482299805,
      "learning_rate": 0.0001232851180988266,
      "loss": 0.5825,
      "step": 2852
    },
    {
      "epoch": 0.4262025694651927,
      "grad_norm": 0.153483584523201,
      "learning_rate": 0.00012323940580757245,
      "loss": 0.7174,
      "step": 2853
    },
    {
      "epoch": 0.4263519569763968,
      "grad_norm": 0.19482477009296417,
      "learning_rate": 0.00012319368838236547,
      "loss": 0.6816,
      "step": 2854
    },
    {
      "epoch": 0.42650134448760085,
      "grad_norm": 0.14367447793483734,
      "learning_rate": 0.00012314796583330533,
      "loss": 0.7239,
      "step": 2855
    },
    {
      "epoch": 0.4266507319988049,
      "grad_norm": 0.13783380389213562,
      "learning_rate": 0.00012310223817049292,
      "loss": 0.6785,
      "step": 2856
    },
    {
      "epoch": 0.42680011951000896,
      "grad_norm": 0.12803930044174194,
      "learning_rate": 0.00012305650540403017,
      "loss": 0.5684,
      "step": 2857
    },
    {
      "epoch": 0.42694950702121304,
      "grad_norm": 0.209193617105484,
      "learning_rate": 0.00012301076754402018,
      "loss": 0.8854,
      "step": 2858
    },
    {
      "epoch": 0.4270988945324171,
      "grad_norm": 0.13880059123039246,
      "learning_rate": 0.0001229650246005672,
      "loss": 0.9132,
      "step": 2859
    },
    {
      "epoch": 0.42724828204362114,
      "grad_norm": 0.17912797629833221,
      "learning_rate": 0.00012291927658377648,
      "loss": 0.6801,
      "step": 2860
    },
    {
      "epoch": 0.4273976695548252,
      "grad_norm": 0.2727680504322052,
      "learning_rate": 0.00012287352350375457,
      "loss": 1.1291,
      "step": 2861
    },
    {
      "epoch": 0.4275470570660293,
      "grad_norm": 0.18817131221294403,
      "learning_rate": 0.00012282776537060903,
      "loss": 0.8583,
      "step": 2862
    },
    {
      "epoch": 0.4276964445772333,
      "grad_norm": 0.10555596649646759,
      "learning_rate": 0.00012278200219444853,
      "loss": 0.5435,
      "step": 2863
    },
    {
      "epoch": 0.4278458320884374,
      "grad_norm": 0.12090355902910233,
      "learning_rate": 0.0001227362339853829,
      "loss": 0.6824,
      "step": 2864
    },
    {
      "epoch": 0.4279952195996415,
      "grad_norm": 0.1797504723072052,
      "learning_rate": 0.00012269046075352304,
      "loss": 0.9718,
      "step": 2865
    },
    {
      "epoch": 0.4281446071108455,
      "grad_norm": 0.32817405462265015,
      "learning_rate": 0.000122644682508981,
      "loss": 0.8559,
      "step": 2866
    },
    {
      "epoch": 0.4282939946220496,
      "grad_norm": 0.33181795477867126,
      "learning_rate": 0.00012259889926187,
      "loss": 0.7284,
      "step": 2867
    },
    {
      "epoch": 0.42844338213325367,
      "grad_norm": 0.21278224885463715,
      "learning_rate": 0.0001225531110223041,
      "loss": 0.5379,
      "step": 2868
    },
    {
      "epoch": 0.42859276964445775,
      "grad_norm": 0.31142812967300415,
      "learning_rate": 0.00012250731780039881,
      "loss": 1.3163,
      "step": 2869
    },
    {
      "epoch": 0.4287421571556618,
      "grad_norm": 0.1409955471754074,
      "learning_rate": 0.00012246151960627053,
      "loss": 0.4663,
      "step": 2870
    },
    {
      "epoch": 0.42889154466686585,
      "grad_norm": 0.17104126513004303,
      "learning_rate": 0.0001224157164500368,
      "loss": 0.5052,
      "step": 2871
    },
    {
      "epoch": 0.42904093217806993,
      "grad_norm": 0.16933749616146088,
      "learning_rate": 0.0001223699083418162,
      "loss": 0.4681,
      "step": 2872
    },
    {
      "epoch": 0.42919031968927396,
      "grad_norm": 0.27251312136650085,
      "learning_rate": 0.00012232409529172857,
      "loss": 0.8397,
      "step": 2873
    },
    {
      "epoch": 0.42933970720047804,
      "grad_norm": 0.2431839406490326,
      "learning_rate": 0.00012227827730989466,
      "loss": 0.9524,
      "step": 2874
    },
    {
      "epoch": 0.4294890947116821,
      "grad_norm": 0.1225556880235672,
      "learning_rate": 0.00012223245440643647,
      "loss": 0.5091,
      "step": 2875
    },
    {
      "epoch": 0.42963848222288614,
      "grad_norm": 0.14125360548496246,
      "learning_rate": 0.00012218662659147693,
      "loss": 0.6381,
      "step": 2876
    },
    {
      "epoch": 0.4297878697340902,
      "grad_norm": 0.2318171113729477,
      "learning_rate": 0.00012214079387514018,
      "loss": 0.6855,
      "step": 2877
    },
    {
      "epoch": 0.4299372572452943,
      "grad_norm": 0.12671636044979095,
      "learning_rate": 0.00012209495626755134,
      "loss": 0.5804,
      "step": 2878
    },
    {
      "epoch": 0.4300866447564984,
      "grad_norm": 0.14059051871299744,
      "learning_rate": 0.0001220491137788367,
      "loss": 0.6767,
      "step": 2879
    },
    {
      "epoch": 0.4302360322677024,
      "grad_norm": 0.12409482151269913,
      "learning_rate": 0.00012200326641912361,
      "loss": 0.6482,
      "step": 2880
    },
    {
      "epoch": 0.4303854197789065,
      "grad_norm": 0.45754534006118774,
      "learning_rate": 0.00012195741419854043,
      "loss": 0.8891,
      "step": 2881
    },
    {
      "epoch": 0.43053480729011057,
      "grad_norm": 0.12788382172584534,
      "learning_rate": 0.00012191155712721667,
      "loss": 0.6527,
      "step": 2882
    },
    {
      "epoch": 0.4306841948013146,
      "grad_norm": 0.1149110347032547,
      "learning_rate": 0.00012186569521528291,
      "loss": 0.6415,
      "step": 2883
    },
    {
      "epoch": 0.43083358231251867,
      "grad_norm": 0.5283525586128235,
      "learning_rate": 0.0001218198284728707,
      "loss": 1.7794,
      "step": 2884
    },
    {
      "epoch": 0.43098296982372275,
      "grad_norm": 0.2018698751926422,
      "learning_rate": 0.0001217739569101128,
      "loss": 0.9464,
      "step": 2885
    },
    {
      "epoch": 0.4311323573349268,
      "grad_norm": 0.6236558556556702,
      "learning_rate": 0.00012172808053714292,
      "loss": 1.768,
      "step": 2886
    },
    {
      "epoch": 0.43128174484613085,
      "grad_norm": 0.17425435781478882,
      "learning_rate": 0.00012168219936409593,
      "loss": 0.5593,
      "step": 2887
    },
    {
      "epoch": 0.43143113235733493,
      "grad_norm": 0.1876472383737564,
      "learning_rate": 0.00012163631340110764,
      "loss": 0.636,
      "step": 2888
    },
    {
      "epoch": 0.431580519868539,
      "grad_norm": 0.1363982856273651,
      "learning_rate": 0.000121590422658315,
      "loss": 0.9004,
      "step": 2889
    },
    {
      "epoch": 0.43172990737974304,
      "grad_norm": 0.10356506705284119,
      "learning_rate": 0.00012154452714585605,
      "loss": 0.6772,
      "step": 2890
    },
    {
      "epoch": 0.4318792948909471,
      "grad_norm": 0.32284942269325256,
      "learning_rate": 0.0001214986268738698,
      "loss": 0.8568,
      "step": 2891
    },
    {
      "epoch": 0.4320286824021512,
      "grad_norm": 0.11978273093700409,
      "learning_rate": 0.00012145272185249634,
      "loss": 0.5998,
      "step": 2892
    },
    {
      "epoch": 0.4321780699133552,
      "grad_norm": 0.14554016292095184,
      "learning_rate": 0.00012140681209187685,
      "loss": 0.6076,
      "step": 2893
    },
    {
      "epoch": 0.4323274574245593,
      "grad_norm": 0.11474756896495819,
      "learning_rate": 0.0001213608976021535,
      "loss": 0.4925,
      "step": 2894
    },
    {
      "epoch": 0.4324768449357634,
      "grad_norm": 0.12136001884937286,
      "learning_rate": 0.0001213149783934695,
      "loss": 0.7951,
      "step": 2895
    },
    {
      "epoch": 0.4326262324469674,
      "grad_norm": 0.1416076272726059,
      "learning_rate": 0.00012126905447596921,
      "loss": 0.9939,
      "step": 2896
    },
    {
      "epoch": 0.4327756199581715,
      "grad_norm": 0.14951768517494202,
      "learning_rate": 0.00012122312585979787,
      "loss": 0.8104,
      "step": 2897
    },
    {
      "epoch": 0.43292500746937557,
      "grad_norm": 0.12565945088863373,
      "learning_rate": 0.00012117719255510188,
      "loss": 0.7277,
      "step": 2898
    },
    {
      "epoch": 0.43307439498057965,
      "grad_norm": 0.19931551814079285,
      "learning_rate": 0.00012113125457202861,
      "loss": 0.833,
      "step": 2899
    },
    {
      "epoch": 0.43322378249178367,
      "grad_norm": 0.1605071723461151,
      "learning_rate": 0.00012108531192072652,
      "loss": 0.8735,
      "step": 2900
    },
    {
      "epoch": 0.43337317000298775,
      "grad_norm": 0.1678197830915451,
      "learning_rate": 0.00012103936461134504,
      "loss": 0.4439,
      "step": 2901
    },
    {
      "epoch": 0.43352255751419183,
      "grad_norm": 0.2413042038679123,
      "learning_rate": 0.0001209934126540347,
      "loss": 0.9831,
      "step": 2902
    },
    {
      "epoch": 0.43367194502539586,
      "grad_norm": 0.21469725668430328,
      "learning_rate": 0.00012094745605894695,
      "loss": 0.8506,
      "step": 2903
    },
    {
      "epoch": 0.43382133253659994,
      "grad_norm": 0.1983805000782013,
      "learning_rate": 0.00012090149483623438,
      "loss": 0.8558,
      "step": 2904
    },
    {
      "epoch": 0.433970720047804,
      "grad_norm": 0.18417227268218994,
      "learning_rate": 0.00012085552899605053,
      "loss": 0.5094,
      "step": 2905
    },
    {
      "epoch": 0.43412010755900804,
      "grad_norm": 0.16300241649150848,
      "learning_rate": 0.00012080955854855002,
      "loss": 0.9974,
      "step": 2906
    },
    {
      "epoch": 0.4342694950702121,
      "grad_norm": 0.13425952196121216,
      "learning_rate": 0.00012076358350388839,
      "loss": 0.8635,
      "step": 2907
    },
    {
      "epoch": 0.4344188825814162,
      "grad_norm": 0.155387744307518,
      "learning_rate": 0.00012071760387222229,
      "loss": 0.7905,
      "step": 2908
    },
    {
      "epoch": 0.4345682700926203,
      "grad_norm": 0.13523848354816437,
      "learning_rate": 0.00012067161966370934,
      "loss": 0.7993,
      "step": 2909
    },
    {
      "epoch": 0.4347176576038243,
      "grad_norm": 0.17934578657150269,
      "learning_rate": 0.0001206256308885082,
      "loss": 0.7142,
      "step": 2910
    },
    {
      "epoch": 0.4348670451150284,
      "grad_norm": 0.13066963851451874,
      "learning_rate": 0.0001205796375567785,
      "loss": 0.7335,
      "step": 2911
    },
    {
      "epoch": 0.43501643262623246,
      "grad_norm": 0.11466148495674133,
      "learning_rate": 0.00012053363967868092,
      "loss": 0.7659,
      "step": 2912
    },
    {
      "epoch": 0.4351658201374365,
      "grad_norm": 0.12826809287071228,
      "learning_rate": 0.00012048763726437706,
      "loss": 0.8333,
      "step": 2913
    },
    {
      "epoch": 0.43531520764864057,
      "grad_norm": 0.20584125816822052,
      "learning_rate": 0.00012044163032402965,
      "loss": 0.8431,
      "step": 2914
    },
    {
      "epoch": 0.43546459515984465,
      "grad_norm": 0.13546201586723328,
      "learning_rate": 0.00012039561886780233,
      "loss": 0.7489,
      "step": 2915
    },
    {
      "epoch": 0.4356139826710487,
      "grad_norm": 0.13818135857582092,
      "learning_rate": 0.0001203496029058597,
      "loss": 0.8035,
      "step": 2916
    },
    {
      "epoch": 0.43576337018225275,
      "grad_norm": 0.14965300261974335,
      "learning_rate": 0.00012030358244836751,
      "loss": 0.7938,
      "step": 2917
    },
    {
      "epoch": 0.43591275769345683,
      "grad_norm": 0.14069749414920807,
      "learning_rate": 0.00012025755750549233,
      "loss": 0.4435,
      "step": 2918
    },
    {
      "epoch": 0.4360621452046609,
      "grad_norm": 0.10928347706794739,
      "learning_rate": 0.00012021152808740188,
      "loss": 0.5808,
      "step": 2919
    },
    {
      "epoch": 0.43621153271586494,
      "grad_norm": 0.18559522926807404,
      "learning_rate": 0.00012016549420426471,
      "loss": 0.8224,
      "step": 2920
    },
    {
      "epoch": 0.436360920227069,
      "grad_norm": 0.3212657868862152,
      "learning_rate": 0.00012011945586625049,
      "loss": 1.1248,
      "step": 2921
    },
    {
      "epoch": 0.4365103077382731,
      "grad_norm": 0.6345764994621277,
      "learning_rate": 0.00012007341308352977,
      "loss": 1.5043,
      "step": 2922
    },
    {
      "epoch": 0.4366596952494771,
      "grad_norm": 0.19030210375785828,
      "learning_rate": 0.00012002736586627416,
      "loss": 0.9212,
      "step": 2923
    },
    {
      "epoch": 0.4368090827606812,
      "grad_norm": 0.19158798456192017,
      "learning_rate": 0.00011998131422465621,
      "loss": 0.7028,
      "step": 2924
    },
    {
      "epoch": 0.4369584702718853,
      "grad_norm": 0.14270685613155365,
      "learning_rate": 0.00011993525816884947,
      "loss": 0.694,
      "step": 2925
    },
    {
      "epoch": 0.43710785778308936,
      "grad_norm": 0.17417439818382263,
      "learning_rate": 0.00011988919770902845,
      "loss": 0.7124,
      "step": 2926
    },
    {
      "epoch": 0.4372572452942934,
      "grad_norm": 0.18878856301307678,
      "learning_rate": 0.00011984313285536859,
      "loss": 0.4815,
      "step": 2927
    },
    {
      "epoch": 0.43740663280549746,
      "grad_norm": 0.2801174521446228,
      "learning_rate": 0.00011979706361804644,
      "loss": 0.7261,
      "step": 2928
    },
    {
      "epoch": 0.43755602031670154,
      "grad_norm": 0.11703645437955856,
      "learning_rate": 0.00011975099000723932,
      "loss": 0.7253,
      "step": 2929
    },
    {
      "epoch": 0.43770540782790557,
      "grad_norm": 0.11064983159303665,
      "learning_rate": 0.00011970491203312568,
      "loss": 0.6245,
      "step": 2930
    },
    {
      "epoch": 0.43785479533910965,
      "grad_norm": 0.14416848123073578,
      "learning_rate": 0.00011965882970588486,
      "loss": 0.6678,
      "step": 2931
    },
    {
      "epoch": 0.43800418285031373,
      "grad_norm": 0.14464090764522552,
      "learning_rate": 0.0001196127430356972,
      "loss": 0.6737,
      "step": 2932
    },
    {
      "epoch": 0.43815357036151775,
      "grad_norm": 0.24060434103012085,
      "learning_rate": 0.00011956665203274396,
      "loss": 0.7682,
      "step": 2933
    },
    {
      "epoch": 0.43830295787272183,
      "grad_norm": 0.260578453540802,
      "learning_rate": 0.00011952055670720732,
      "loss": 0.7488,
      "step": 2934
    },
    {
      "epoch": 0.4384523453839259,
      "grad_norm": 0.15550395846366882,
      "learning_rate": 0.00011947445706927054,
      "loss": 0.747,
      "step": 2935
    },
    {
      "epoch": 0.43860173289513,
      "grad_norm": 0.11961430311203003,
      "learning_rate": 0.00011942835312911773,
      "loss": 0.7161,
      "step": 2936
    },
    {
      "epoch": 0.438751120406334,
      "grad_norm": 0.2887740135192871,
      "learning_rate": 0.00011938224489693397,
      "loss": 0.7923,
      "step": 2937
    },
    {
      "epoch": 0.4389005079175381,
      "grad_norm": 0.7316411733627319,
      "learning_rate": 0.00011933613238290535,
      "loss": 1.3948,
      "step": 2938
    },
    {
      "epoch": 0.4390498954287422,
      "grad_norm": 0.12260571867227554,
      "learning_rate": 0.00011929001559721879,
      "loss": 0.7778,
      "step": 2939
    },
    {
      "epoch": 0.4391992829399462,
      "grad_norm": 0.19185659289360046,
      "learning_rate": 0.00011924389455006226,
      "loss": 0.6707,
      "step": 2940
    },
    {
      "epoch": 0.4393486704511503,
      "grad_norm": 0.17516249418258667,
      "learning_rate": 0.00011919776925162463,
      "loss": 0.6498,
      "step": 2941
    },
    {
      "epoch": 0.43949805796235436,
      "grad_norm": 0.22149495780467987,
      "learning_rate": 0.00011915163971209566,
      "loss": 1.0308,
      "step": 2942
    },
    {
      "epoch": 0.4396474454735584,
      "grad_norm": 0.15715281665325165,
      "learning_rate": 0.00011910550594166615,
      "loss": 0.6827,
      "step": 2943
    },
    {
      "epoch": 0.43979683298476246,
      "grad_norm": 0.17531903088092804,
      "learning_rate": 0.00011905936795052774,
      "loss": 0.5687,
      "step": 2944
    },
    {
      "epoch": 0.43994622049596654,
      "grad_norm": 0.1986943930387497,
      "learning_rate": 0.00011901322574887311,
      "loss": 1.0203,
      "step": 2945
    },
    {
      "epoch": 0.4400956080071706,
      "grad_norm": 0.17914053797721863,
      "learning_rate": 0.0001189670793468957,
      "loss": 0.8286,
      "step": 2946
    },
    {
      "epoch": 0.44024499551837465,
      "grad_norm": 0.2752475440502167,
      "learning_rate": 0.00011892092875479008,
      "loss": 1.1348,
      "step": 2947
    },
    {
      "epoch": 0.44039438302957873,
      "grad_norm": 0.1363374888896942,
      "learning_rate": 0.00011887477398275162,
      "loss": 0.6349,
      "step": 2948
    },
    {
      "epoch": 0.4405437705407828,
      "grad_norm": 0.13850156962871552,
      "learning_rate": 0.0001188286150409766,
      "loss": 0.741,
      "step": 2949
    },
    {
      "epoch": 0.44069315805198683,
      "grad_norm": 0.14029064774513245,
      "learning_rate": 0.00011878245193966229,
      "loss": 0.6352,
      "step": 2950
    },
    {
      "epoch": 0.4408425455631909,
      "grad_norm": 0.12867359817028046,
      "learning_rate": 0.00011873628468900686,
      "loss": 0.8405,
      "step": 2951
    },
    {
      "epoch": 0.440991933074395,
      "grad_norm": 0.15658962726593018,
      "learning_rate": 0.00011869011329920936,
      "loss": 0.6066,
      "step": 2952
    },
    {
      "epoch": 0.441141320585599,
      "grad_norm": 0.300100713968277,
      "learning_rate": 0.00011864393778046976,
      "loss": 1.0935,
      "step": 2953
    },
    {
      "epoch": 0.4412907080968031,
      "grad_norm": 0.16302046179771423,
      "learning_rate": 0.00011859775814298905,
      "loss": 0.7751,
      "step": 2954
    },
    {
      "epoch": 0.4414400956080072,
      "grad_norm": 0.13358524441719055,
      "learning_rate": 0.000118551574396969,
      "loss": 0.7147,
      "step": 2955
    },
    {
      "epoch": 0.44158948311921126,
      "grad_norm": 0.2991906702518463,
      "learning_rate": 0.0001185053865526123,
      "loss": 0.8664,
      "step": 2956
    },
    {
      "epoch": 0.4417388706304153,
      "grad_norm": 0.1485229730606079,
      "learning_rate": 0.00011845919462012263,
      "loss": 0.6917,
      "step": 2957
    },
    {
      "epoch": 0.44188825814161936,
      "grad_norm": 0.18354055285453796,
      "learning_rate": 0.00011841299860970445,
      "loss": 0.5544,
      "step": 2958
    },
    {
      "epoch": 0.44203764565282344,
      "grad_norm": 0.12863700091838837,
      "learning_rate": 0.00011836679853156325,
      "loss": 0.8128,
      "step": 2959
    },
    {
      "epoch": 0.44218703316402747,
      "grad_norm": 0.16558223962783813,
      "learning_rate": 0.00011832059439590533,
      "loss": 0.7463,
      "step": 2960
    },
    {
      "epoch": 0.44233642067523155,
      "grad_norm": 0.14082030951976776,
      "learning_rate": 0.00011827438621293793,
      "loss": 0.9492,
      "step": 2961
    },
    {
      "epoch": 0.4424858081864356,
      "grad_norm": 0.1319563090801239,
      "learning_rate": 0.00011822817399286916,
      "loss": 0.7779,
      "step": 2962
    },
    {
      "epoch": 0.4426351956976397,
      "grad_norm": 0.26467761397361755,
      "learning_rate": 0.00011818195774590804,
      "loss": 0.8064,
      "step": 2963
    },
    {
      "epoch": 0.44278458320884373,
      "grad_norm": 0.2120577096939087,
      "learning_rate": 0.00011813573748226447,
      "loss": 0.6801,
      "step": 2964
    },
    {
      "epoch": 0.4429339707200478,
      "grad_norm": 0.19719187915325165,
      "learning_rate": 0.00011808951321214925,
      "loss": 0.4558,
      "step": 2965
    },
    {
      "epoch": 0.4430833582312519,
      "grad_norm": 0.3748786449432373,
      "learning_rate": 0.00011804328494577402,
      "loss": 0.7635,
      "step": 2966
    },
    {
      "epoch": 0.4432327457424559,
      "grad_norm": 0.10475513339042664,
      "learning_rate": 0.00011799705269335135,
      "loss": 0.4616,
      "step": 2967
    },
    {
      "epoch": 0.44338213325366,
      "grad_norm": 0.2338329702615738,
      "learning_rate": 0.00011795081646509469,
      "loss": 1.2285,
      "step": 2968
    },
    {
      "epoch": 0.4435315207648641,
      "grad_norm": 0.13546501100063324,
      "learning_rate": 0.00011790457627121834,
      "loss": 0.6832,
      "step": 2969
    },
    {
      "epoch": 0.4436809082760681,
      "grad_norm": 0.20702789723873138,
      "learning_rate": 0.00011785833212193749,
      "loss": 0.8439,
      "step": 2970
    },
    {
      "epoch": 0.4438302957872722,
      "grad_norm": 0.13944704830646515,
      "learning_rate": 0.00011781208402746821,
      "loss": 0.6687,
      "step": 2971
    },
    {
      "epoch": 0.44397968329847626,
      "grad_norm": 0.1658911406993866,
      "learning_rate": 0.00011776583199802746,
      "loss": 0.6067,
      "step": 2972
    },
    {
      "epoch": 0.44412907080968034,
      "grad_norm": 0.11394371092319489,
      "learning_rate": 0.00011771957604383304,
      "loss": 0.5935,
      "step": 2973
    },
    {
      "epoch": 0.44427845832088436,
      "grad_norm": 0.12178372591733932,
      "learning_rate": 0.00011767331617510358,
      "loss": 0.7595,
      "step": 2974
    },
    {
      "epoch": 0.44442784583208844,
      "grad_norm": 0.2116834670305252,
      "learning_rate": 0.00011762705240205866,
      "loss": 1.1761,
      "step": 2975
    },
    {
      "epoch": 0.4445772333432925,
      "grad_norm": 0.14498242735862732,
      "learning_rate": 0.00011758078473491864,
      "loss": 0.8792,
      "step": 2976
    },
    {
      "epoch": 0.44472662085449655,
      "grad_norm": 0.19560860097408295,
      "learning_rate": 0.00011753451318390485,
      "loss": 1.0021,
      "step": 2977
    },
    {
      "epoch": 0.4448760083657006,
      "grad_norm": 0.1636829376220703,
      "learning_rate": 0.00011748823775923934,
      "loss": 0.6888,
      "step": 2978
    },
    {
      "epoch": 0.4450253958769047,
      "grad_norm": 0.15937615931034088,
      "learning_rate": 0.00011744195847114512,
      "loss": 0.5621,
      "step": 2979
    },
    {
      "epoch": 0.44517478338810873,
      "grad_norm": 0.1868770569562912,
      "learning_rate": 0.00011739567532984598,
      "loss": 0.7358,
      "step": 2980
    },
    {
      "epoch": 0.4453241708993128,
      "grad_norm": 0.2493693083524704,
      "learning_rate": 0.00011734938834556666,
      "loss": 0.7867,
      "step": 2981
    },
    {
      "epoch": 0.4454735584105169,
      "grad_norm": 0.14532269537448883,
      "learning_rate": 0.00011730309752853261,
      "loss": 0.7529,
      "step": 2982
    },
    {
      "epoch": 0.44562294592172097,
      "grad_norm": 0.17708668112754822,
      "learning_rate": 0.00011725680288897032,
      "loss": 0.6723,
      "step": 2983
    },
    {
      "epoch": 0.445772333432925,
      "grad_norm": 0.1584847867488861,
      "learning_rate": 0.00011721050443710688,
      "loss": 0.4894,
      "step": 2984
    },
    {
      "epoch": 0.4459217209441291,
      "grad_norm": 0.1430836170911789,
      "learning_rate": 0.00011716420218317044,
      "loss": 0.9075,
      "step": 2985
    },
    {
      "epoch": 0.44607110845533315,
      "grad_norm": 0.21276701986789703,
      "learning_rate": 0.00011711789613738986,
      "loss": 0.5817,
      "step": 2986
    },
    {
      "epoch": 0.4462204959665372,
      "grad_norm": 0.24936650693416595,
      "learning_rate": 0.00011707158630999489,
      "loss": 0.7923,
      "step": 2987
    },
    {
      "epoch": 0.44636988347774126,
      "grad_norm": 0.3508971929550171,
      "learning_rate": 0.00011702527271121609,
      "loss": 1.1471,
      "step": 2988
    },
    {
      "epoch": 0.44651927098894534,
      "grad_norm": 0.18281325697898865,
      "learning_rate": 0.00011697895535128489,
      "loss": 0.4447,
      "step": 2989
    },
    {
      "epoch": 0.44666865850014936,
      "grad_norm": 0.3060774803161621,
      "learning_rate": 0.00011693263424043353,
      "loss": 0.9235,
      "step": 2990
    },
    {
      "epoch": 0.44681804601135344,
      "grad_norm": 0.1618868112564087,
      "learning_rate": 0.00011688630938889504,
      "loss": 0.5214,
      "step": 2991
    },
    {
      "epoch": 0.4469674335225575,
      "grad_norm": 0.13401705026626587,
      "learning_rate": 0.00011683998080690334,
      "loss": 0.8924,
      "step": 2992
    },
    {
      "epoch": 0.4471168210337616,
      "grad_norm": 0.27304574847221375,
      "learning_rate": 0.00011679364850469322,
      "loss": 0.8616,
      "step": 2993
    },
    {
      "epoch": 0.4472662085449656,
      "grad_norm": 0.1593634933233261,
      "learning_rate": 0.00011674731249250008,
      "loss": 0.6231,
      "step": 2994
    },
    {
      "epoch": 0.4474155960561697,
      "grad_norm": 0.20717370510101318,
      "learning_rate": 0.00011670097278056036,
      "loss": 0.8969,
      "step": 2995
    },
    {
      "epoch": 0.4475649835673738,
      "grad_norm": 0.30716297030448914,
      "learning_rate": 0.00011665462937911124,
      "loss": 1.1953,
      "step": 2996
    },
    {
      "epoch": 0.4477143710785778,
      "grad_norm": 0.26136818528175354,
      "learning_rate": 0.00011660828229839067,
      "loss": 0.8488,
      "step": 2997
    },
    {
      "epoch": 0.4478637585897819,
      "grad_norm": 0.13850393891334534,
      "learning_rate": 0.00011656193154863749,
      "loss": 0.8695,
      "step": 2998
    },
    {
      "epoch": 0.44801314610098597,
      "grad_norm": 0.19375580549240112,
      "learning_rate": 0.00011651557714009131,
      "loss": 0.5681,
      "step": 2999
    },
    {
      "epoch": 0.44816253361219,
      "grad_norm": 0.18379487097263336,
      "learning_rate": 0.00011646921908299254,
      "loss": 0.9334,
      "step": 3000
    },
    {
      "epoch": 0.4483119211233941,
      "grad_norm": 0.115275077521801,
      "learning_rate": 0.0001164228573875824,
      "loss": 0.5079,
      "step": 3001
    },
    {
      "epoch": 0.44846130863459815,
      "grad_norm": 0.12357329577207565,
      "learning_rate": 0.00011637649206410298,
      "loss": 0.6684,
      "step": 3002
    },
    {
      "epoch": 0.44861069614580223,
      "grad_norm": 0.12659284472465515,
      "learning_rate": 0.00011633012312279703,
      "loss": 0.6701,
      "step": 3003
    },
    {
      "epoch": 0.44876008365700626,
      "grad_norm": 0.1477823257446289,
      "learning_rate": 0.00011628375057390824,
      "loss": 0.4636,
      "step": 3004
    },
    {
      "epoch": 0.44890947116821034,
      "grad_norm": 0.3900808095932007,
      "learning_rate": 0.00011623737442768103,
      "loss": 0.6313,
      "step": 3005
    },
    {
      "epoch": 0.4490588586794144,
      "grad_norm": 0.14636239409446716,
      "learning_rate": 0.00011619099469436061,
      "loss": 0.7859,
      "step": 3006
    },
    {
      "epoch": 0.44920824619061844,
      "grad_norm": 0.17527520656585693,
      "learning_rate": 0.00011614461138419303,
      "loss": 0.9349,
      "step": 3007
    },
    {
      "epoch": 0.4493576337018225,
      "grad_norm": 0.17213378846645355,
      "learning_rate": 0.00011609822450742507,
      "loss": 0.9978,
      "step": 3008
    },
    {
      "epoch": 0.4495070212130266,
      "grad_norm": 0.39235687255859375,
      "learning_rate": 0.00011605183407430434,
      "loss": 1.0454,
      "step": 3009
    },
    {
      "epoch": 0.4496564087242306,
      "grad_norm": 0.194938525557518,
      "learning_rate": 0.0001160054400950792,
      "loss": 0.5566,
      "step": 3010
    },
    {
      "epoch": 0.4498057962354347,
      "grad_norm": 0.2180972546339035,
      "learning_rate": 0.00011595904257999883,
      "loss": 0.8599,
      "step": 3011
    },
    {
      "epoch": 0.4499551837466388,
      "grad_norm": 0.18901172280311584,
      "learning_rate": 0.00011591264153931321,
      "loss": 0.957,
      "step": 3012
    },
    {
      "epoch": 0.45010457125784287,
      "grad_norm": 0.14183160662651062,
      "learning_rate": 0.00011586623698327302,
      "loss": 0.9539,
      "step": 3013
    },
    {
      "epoch": 0.4502539587690469,
      "grad_norm": 0.17253778874874115,
      "learning_rate": 0.00011581982892212975,
      "loss": 0.6894,
      "step": 3014
    },
    {
      "epoch": 0.45040334628025097,
      "grad_norm": 0.1741241216659546,
      "learning_rate": 0.00011577341736613569,
      "loss": 0.9838,
      "step": 3015
    },
    {
      "epoch": 0.45055273379145505,
      "grad_norm": 0.15802161395549774,
      "learning_rate": 0.0001157270023255439,
      "loss": 0.5658,
      "step": 3016
    },
    {
      "epoch": 0.4507021213026591,
      "grad_norm": 0.16704627871513367,
      "learning_rate": 0.0001156805838106082,
      "loss": 0.6154,
      "step": 3017
    },
    {
      "epoch": 0.45085150881386316,
      "grad_norm": 0.5187243223190308,
      "learning_rate": 0.00011563416183158318,
      "loss": 1.1316,
      "step": 3018
    },
    {
      "epoch": 0.45100089632506724,
      "grad_norm": 0.1419323831796646,
      "learning_rate": 0.0001155877363987242,
      "loss": 0.4808,
      "step": 3019
    },
    {
      "epoch": 0.4511502838362713,
      "grad_norm": 0.13114412128925323,
      "learning_rate": 0.00011554130752228731,
      "loss": 0.8154,
      "step": 3020
    },
    {
      "epoch": 0.45129967134747534,
      "grad_norm": 0.35912376642227173,
      "learning_rate": 0.00011549487521252945,
      "loss": 0.9483,
      "step": 3021
    },
    {
      "epoch": 0.4514490588586794,
      "grad_norm": 0.13670995831489563,
      "learning_rate": 0.00011544843947970822,
      "loss": 0.6175,
      "step": 3022
    },
    {
      "epoch": 0.4515984463698835,
      "grad_norm": 0.14213192462921143,
      "learning_rate": 0.00011540200033408202,
      "loss": 0.6474,
      "step": 3023
    },
    {
      "epoch": 0.4517478338810875,
      "grad_norm": 0.11808183789253235,
      "learning_rate": 0.00011535555778590999,
      "loss": 0.6347,
      "step": 3024
    },
    {
      "epoch": 0.4518972213922916,
      "grad_norm": 0.13648447394371033,
      "learning_rate": 0.00011530911184545201,
      "loss": 0.8915,
      "step": 3025
    },
    {
      "epoch": 0.4520466089034957,
      "grad_norm": 0.303817480802536,
      "learning_rate": 0.00011526266252296876,
      "loss": 0.6827,
      "step": 3026
    },
    {
      "epoch": 0.4521959964146997,
      "grad_norm": 0.1179194226861,
      "learning_rate": 0.0001152162098287216,
      "loss": 0.7514,
      "step": 3027
    },
    {
      "epoch": 0.4523453839259038,
      "grad_norm": 0.6338302493095398,
      "learning_rate": 0.0001151697537729727,
      "loss": 1.9988,
      "step": 3028
    },
    {
      "epoch": 0.45249477143710787,
      "grad_norm": 0.14428377151489258,
      "learning_rate": 0.0001151232943659849,
      "loss": 0.7328,
      "step": 3029
    },
    {
      "epoch": 0.45264415894831195,
      "grad_norm": 0.11680792272090912,
      "learning_rate": 0.00011507683161802184,
      "loss": 0.6931,
      "step": 3030
    },
    {
      "epoch": 0.45279354645951597,
      "grad_norm": 0.2534790635108948,
      "learning_rate": 0.00011503036553934783,
      "loss": 0.8214,
      "step": 3031
    },
    {
      "epoch": 0.45294293397072005,
      "grad_norm": 0.14233434200286865,
      "learning_rate": 0.00011498389614022807,
      "loss": 0.6874,
      "step": 3032
    },
    {
      "epoch": 0.45309232148192413,
      "grad_norm": 0.20028844475746155,
      "learning_rate": 0.00011493742343092829,
      "loss": 0.9863,
      "step": 3033
    },
    {
      "epoch": 0.45324170899312816,
      "grad_norm": 0.2064978927373886,
      "learning_rate": 0.00011489094742171502,
      "loss": 1.0185,
      "step": 3034
    },
    {
      "epoch": 0.45339109650433224,
      "grad_norm": 0.29884544014930725,
      "learning_rate": 0.00011484446812285569,
      "loss": 0.9631,
      "step": 3035
    },
    {
      "epoch": 0.4535404840155363,
      "grad_norm": 0.1820349097251892,
      "learning_rate": 0.00011479798554461818,
      "loss": 1.0093,
      "step": 3036
    },
    {
      "epoch": 0.45368987152674034,
      "grad_norm": 0.21576669812202454,
      "learning_rate": 0.0001147514996972713,
      "loss": 0.7707,
      "step": 3037
    },
    {
      "epoch": 0.4538392590379444,
      "grad_norm": 0.37682294845581055,
      "learning_rate": 0.0001147050105910845,
      "loss": 1.347,
      "step": 3038
    },
    {
      "epoch": 0.4539886465491485,
      "grad_norm": 0.14763346314430237,
      "learning_rate": 0.00011465851823632791,
      "loss": 0.7439,
      "step": 3039
    },
    {
      "epoch": 0.4541380340603526,
      "grad_norm": 0.1878838837146759,
      "learning_rate": 0.00011461202264327246,
      "loss": 0.7848,
      "step": 3040
    },
    {
      "epoch": 0.4542874215715566,
      "grad_norm": 0.15333043038845062,
      "learning_rate": 0.00011456552382218982,
      "loss": 0.7631,
      "step": 3041
    },
    {
      "epoch": 0.4544368090827607,
      "grad_norm": 0.13961176574230194,
      "learning_rate": 0.00011451902178335219,
      "loss": 0.5979,
      "step": 3042
    },
    {
      "epoch": 0.45458619659396476,
      "grad_norm": 0.22130531072616577,
      "learning_rate": 0.00011447251653703271,
      "loss": 0.9177,
      "step": 3043
    },
    {
      "epoch": 0.4547355841051688,
      "grad_norm": 0.1538708209991455,
      "learning_rate": 0.0001144260080935051,
      "loss": 0.7141,
      "step": 3044
    },
    {
      "epoch": 0.45488497161637287,
      "grad_norm": 0.13157567381858826,
      "learning_rate": 0.0001143794964630438,
      "loss": 0.5479,
      "step": 3045
    },
    {
      "epoch": 0.45503435912757695,
      "grad_norm": 0.14429813623428345,
      "learning_rate": 0.00011433298165592396,
      "loss": 0.804,
      "step": 3046
    },
    {
      "epoch": 0.455183746638781,
      "grad_norm": 0.1315287947654724,
      "learning_rate": 0.00011428646368242146,
      "loss": 0.7519,
      "step": 3047
    },
    {
      "epoch": 0.45533313414998505,
      "grad_norm": 0.11657236516475677,
      "learning_rate": 0.00011423994255281285,
      "loss": 0.5812,
      "step": 3048
    },
    {
      "epoch": 0.45548252166118913,
      "grad_norm": 0.16290625929832458,
      "learning_rate": 0.00011419341827737539,
      "loss": 0.688,
      "step": 3049
    },
    {
      "epoch": 0.4556319091723932,
      "grad_norm": 0.13318508863449097,
      "learning_rate": 0.000114146890866387,
      "loss": 0.8208,
      "step": 3050
    },
    {
      "epoch": 0.45578129668359724,
      "grad_norm": 0.1449301689863205,
      "learning_rate": 0.00011410036033012634,
      "loss": 0.547,
      "step": 3051
    },
    {
      "epoch": 0.4559306841948013,
      "grad_norm": 0.4237108528614044,
      "learning_rate": 0.00011405382667887276,
      "loss": 1.1981,
      "step": 3052
    },
    {
      "epoch": 0.4560800717060054,
      "grad_norm": 0.12480130791664124,
      "learning_rate": 0.00011400728992290626,
      "loss": 0.5806,
      "step": 3053
    },
    {
      "epoch": 0.4562294592172094,
      "grad_norm": 0.135677307844162,
      "learning_rate": 0.00011396075007250758,
      "loss": 0.8688,
      "step": 3054
    },
    {
      "epoch": 0.4563788467284135,
      "grad_norm": 0.1564306914806366,
      "learning_rate": 0.00011391420713795804,
      "loss": 0.9283,
      "step": 3055
    },
    {
      "epoch": 0.4565282342396176,
      "grad_norm": 0.1302371621131897,
      "learning_rate": 0.00011386766112953977,
      "loss": 0.6694,
      "step": 3056
    },
    {
      "epoch": 0.4566776217508216,
      "grad_norm": 0.16178357601165771,
      "learning_rate": 0.00011382111205753551,
      "loss": 0.6563,
      "step": 3057
    },
    {
      "epoch": 0.4568270092620257,
      "grad_norm": 0.1393168717622757,
      "learning_rate": 0.00011377455993222867,
      "loss": 0.5492,
      "step": 3058
    },
    {
      "epoch": 0.45697639677322976,
      "grad_norm": 0.15978658199310303,
      "learning_rate": 0.00011372800476390338,
      "loss": 0.6731,
      "step": 3059
    },
    {
      "epoch": 0.45712578428443384,
      "grad_norm": 0.4594833552837372,
      "learning_rate": 0.00011368144656284436,
      "loss": 0.9138,
      "step": 3060
    },
    {
      "epoch": 0.45727517179563787,
      "grad_norm": 0.15956401824951172,
      "learning_rate": 0.00011363488533933712,
      "loss": 0.5944,
      "step": 3061
    },
    {
      "epoch": 0.45742455930684195,
      "grad_norm": 0.14904318749904633,
      "learning_rate": 0.00011358832110366775,
      "loss": 0.9244,
      "step": 3062
    },
    {
      "epoch": 0.45757394681804603,
      "grad_norm": 0.16681870818138123,
      "learning_rate": 0.000113541753866123,
      "loss": 0.5344,
      "step": 3063
    },
    {
      "epoch": 0.45772333432925005,
      "grad_norm": 0.20832683145999908,
      "learning_rate": 0.00011349518363699036,
      "loss": 0.8546,
      "step": 3064
    },
    {
      "epoch": 0.45787272184045413,
      "grad_norm": 0.172454833984375,
      "learning_rate": 0.00011344861042655788,
      "loss": 0.7132,
      "step": 3065
    },
    {
      "epoch": 0.4580221093516582,
      "grad_norm": 0.1885954737663269,
      "learning_rate": 0.00011340203424511434,
      "loss": 0.6205,
      "step": 3066
    },
    {
      "epoch": 0.45817149686286224,
      "grad_norm": 0.13376373052597046,
      "learning_rate": 0.00011335545510294918,
      "loss": 0.5791,
      "step": 3067
    },
    {
      "epoch": 0.4583208843740663,
      "grad_norm": 0.19313010573387146,
      "learning_rate": 0.00011330887301035242,
      "loss": 0.5894,
      "step": 3068
    },
    {
      "epoch": 0.4584702718852704,
      "grad_norm": 0.164376363158226,
      "learning_rate": 0.00011326228797761482,
      "loss": 0.5936,
      "step": 3069
    },
    {
      "epoch": 0.4586196593964745,
      "grad_norm": 0.14847366511821747,
      "learning_rate": 0.00011321570001502775,
      "loss": 0.93,
      "step": 3070
    },
    {
      "epoch": 0.4587690469076785,
      "grad_norm": 0.13392731547355652,
      "learning_rate": 0.0001131691091328832,
      "loss": 0.7955,
      "step": 3071
    },
    {
      "epoch": 0.4589184344188826,
      "grad_norm": 0.15295733511447906,
      "learning_rate": 0.00011312251534147387,
      "loss": 0.9339,
      "step": 3072
    },
    {
      "epoch": 0.45906782193008666,
      "grad_norm": 0.16152796149253845,
      "learning_rate": 0.00011307591865109306,
      "loss": 0.6446,
      "step": 3073
    },
    {
      "epoch": 0.4592172094412907,
      "grad_norm": 0.13615615665912628,
      "learning_rate": 0.0001130293190720347,
      "loss": 0.6695,
      "step": 3074
    },
    {
      "epoch": 0.45936659695249477,
      "grad_norm": 0.16136105358600616,
      "learning_rate": 0.00011298271661459336,
      "loss": 0.6209,
      "step": 3075
    },
    {
      "epoch": 0.45951598446369885,
      "grad_norm": 0.3457396328449249,
      "learning_rate": 0.00011293611128906431,
      "loss": 0.8181,
      "step": 3076
    },
    {
      "epoch": 0.4596653719749029,
      "grad_norm": 0.12549446523189545,
      "learning_rate": 0.0001128895031057434,
      "loss": 0.7068,
      "step": 3077
    },
    {
      "epoch": 0.45981475948610695,
      "grad_norm": 0.13915947079658508,
      "learning_rate": 0.00011284289207492706,
      "loss": 0.706,
      "step": 3078
    },
    {
      "epoch": 0.45996414699731103,
      "grad_norm": 0.19965484738349915,
      "learning_rate": 0.00011279627820691243,
      "loss": 1.0998,
      "step": 3079
    },
    {
      "epoch": 0.4601135345085151,
      "grad_norm": 0.1360960751771927,
      "learning_rate": 0.00011274966151199731,
      "loss": 0.4705,
      "step": 3080
    },
    {
      "epoch": 0.46026292201971913,
      "grad_norm": 0.20444996654987335,
      "learning_rate": 0.00011270304200047997,
      "loss": 0.8962,
      "step": 3081
    },
    {
      "epoch": 0.4604123095309232,
      "grad_norm": 0.1538262665271759,
      "learning_rate": 0.00011265641968265945,
      "loss": 0.8265,
      "step": 3082
    },
    {
      "epoch": 0.4605616970421273,
      "grad_norm": 0.1473671942949295,
      "learning_rate": 0.0001126097945688354,
      "loss": 0.7334,
      "step": 3083
    },
    {
      "epoch": 0.4607110845533313,
      "grad_norm": 0.11609522253274918,
      "learning_rate": 0.00011256316666930798,
      "loss": 0.6119,
      "step": 3084
    },
    {
      "epoch": 0.4608604720645354,
      "grad_norm": 0.13394205272197723,
      "learning_rate": 0.00011251653599437801,
      "loss": 0.9461,
      "step": 3085
    },
    {
      "epoch": 0.4610098595757395,
      "grad_norm": 0.20881475508213043,
      "learning_rate": 0.00011246990255434704,
      "loss": 0.7784,
      "step": 3086
    },
    {
      "epoch": 0.46115924708694356,
      "grad_norm": 0.12687896192073822,
      "learning_rate": 0.00011242326635951705,
      "loss": 0.7877,
      "step": 3087
    },
    {
      "epoch": 0.4613086345981476,
      "grad_norm": 0.14070720970630646,
      "learning_rate": 0.00011237662742019075,
      "loss": 0.7313,
      "step": 3088
    },
    {
      "epoch": 0.46145802210935166,
      "grad_norm": 0.18513940274715424,
      "learning_rate": 0.00011232998574667142,
      "loss": 0.9903,
      "step": 3089
    },
    {
      "epoch": 0.46160740962055574,
      "grad_norm": 0.10740021616220474,
      "learning_rate": 0.00011228334134926297,
      "loss": 0.516,
      "step": 3090
    },
    {
      "epoch": 0.46175679713175977,
      "grad_norm": 0.25325068831443787,
      "learning_rate": 0.00011223669423826983,
      "loss": 0.9773,
      "step": 3091
    },
    {
      "epoch": 0.46190618464296385,
      "grad_norm": 0.4688616394996643,
      "learning_rate": 0.00011219004442399712,
      "loss": 1.349,
      "step": 3092
    },
    {
      "epoch": 0.4620555721541679,
      "grad_norm": 0.24754329025745392,
      "learning_rate": 0.00011214339191675054,
      "loss": 0.9312,
      "step": 3093
    },
    {
      "epoch": 0.46220495966537195,
      "grad_norm": 0.44255977869033813,
      "learning_rate": 0.00011209673672683632,
      "loss": 0.9714,
      "step": 3094
    },
    {
      "epoch": 0.46235434717657603,
      "grad_norm": 0.13128554821014404,
      "learning_rate": 0.00011205007886456134,
      "loss": 0.5836,
      "step": 3095
    },
    {
      "epoch": 0.4625037346877801,
      "grad_norm": 0.28436312079429626,
      "learning_rate": 0.00011200341834023309,
      "loss": 0.9439,
      "step": 3096
    },
    {
      "epoch": 0.4626531221989842,
      "grad_norm": 0.21179884672164917,
      "learning_rate": 0.0001119567551641596,
      "loss": 0.7634,
      "step": 3097
    },
    {
      "epoch": 0.4628025097101882,
      "grad_norm": 0.471744567155838,
      "learning_rate": 0.00011191008934664951,
      "loss": 1.1989,
      "step": 3098
    },
    {
      "epoch": 0.4629518972213923,
      "grad_norm": 0.1297721564769745,
      "learning_rate": 0.00011186342089801205,
      "loss": 0.8909,
      "step": 3099
    },
    {
      "epoch": 0.4631012847325964,
      "grad_norm": 0.12586760520935059,
      "learning_rate": 0.000111816749828557,
      "loss": 0.6949,
      "step": 3100
    },
    {
      "epoch": 0.4632506722438004,
      "grad_norm": 0.1491735577583313,
      "learning_rate": 0.00011177007614859474,
      "loss": 0.9893,
      "step": 3101
    },
    {
      "epoch": 0.4634000597550045,
      "grad_norm": 0.1652022749185562,
      "learning_rate": 0.00011172339986843626,
      "loss": 0.7671,
      "step": 3102
    },
    {
      "epoch": 0.46354944726620856,
      "grad_norm": 0.349630206823349,
      "learning_rate": 0.00011167672099839301,
      "loss": 1.1454,
      "step": 3103
    },
    {
      "epoch": 0.4636988347774126,
      "grad_norm": 0.19648872315883636,
      "learning_rate": 0.00011163003954877718,
      "loss": 0.4813,
      "step": 3104
    },
    {
      "epoch": 0.46384822228861666,
      "grad_norm": 0.2013506442308426,
      "learning_rate": 0.00011158335552990138,
      "loss": 0.891,
      "step": 3105
    },
    {
      "epoch": 0.46399760979982074,
      "grad_norm": 0.14355367422103882,
      "learning_rate": 0.00011153666895207885,
      "loss": 0.795,
      "step": 3106
    },
    {
      "epoch": 0.4641469973110248,
      "grad_norm": 0.23531007766723633,
      "learning_rate": 0.00011148997982562346,
      "loss": 0.8204,
      "step": 3107
    },
    {
      "epoch": 0.46429638482222885,
      "grad_norm": 0.14198468625545502,
      "learning_rate": 0.00011144328816084952,
      "loss": 0.8012,
      "step": 3108
    },
    {
      "epoch": 0.4644457723334329,
      "grad_norm": 0.13423845171928406,
      "learning_rate": 0.00011139659396807198,
      "loss": 0.8991,
      "step": 3109
    },
    {
      "epoch": 0.464595159844637,
      "grad_norm": 0.2894625663757324,
      "learning_rate": 0.00011134989725760632,
      "loss": 1.1126,
      "step": 3110
    },
    {
      "epoch": 0.46474454735584103,
      "grad_norm": 0.17335911095142365,
      "learning_rate": 0.00011130319803976859,
      "loss": 0.5031,
      "step": 3111
    },
    {
      "epoch": 0.4648939348670451,
      "grad_norm": 0.24626342952251434,
      "learning_rate": 0.00011125649632487538,
      "loss": 0.8004,
      "step": 3112
    },
    {
      "epoch": 0.4650433223782492,
      "grad_norm": 0.21964651346206665,
      "learning_rate": 0.00011120979212324383,
      "loss": 0.9791,
      "step": 3113
    },
    {
      "epoch": 0.4651927098894532,
      "grad_norm": 0.2148008495569229,
      "learning_rate": 0.00011116308544519163,
      "loss": 0.8094,
      "step": 3114
    },
    {
      "epoch": 0.4653420974006573,
      "grad_norm": 0.16167204082012177,
      "learning_rate": 0.00011111637630103704,
      "loss": 0.763,
      "step": 3115
    },
    {
      "epoch": 0.4654914849118614,
      "grad_norm": 0.10906770825386047,
      "learning_rate": 0.00011106966470109888,
      "loss": 0.577,
      "step": 3116
    },
    {
      "epoch": 0.46564087242306545,
      "grad_norm": 0.13580769300460815,
      "learning_rate": 0.00011102295065569641,
      "loss": 0.8811,
      "step": 3117
    },
    {
      "epoch": 0.4657902599342695,
      "grad_norm": 0.12718717753887177,
      "learning_rate": 0.00011097623417514957,
      "loss": 0.6157,
      "step": 3118
    },
    {
      "epoch": 0.46593964744547356,
      "grad_norm": 0.18473593890666962,
      "learning_rate": 0.00011092951526977877,
      "loss": 0.6046,
      "step": 3119
    },
    {
      "epoch": 0.46608903495667764,
      "grad_norm": 0.14006124436855316,
      "learning_rate": 0.00011088279394990491,
      "loss": 0.931,
      "step": 3120
    },
    {
      "epoch": 0.46623842246788166,
      "grad_norm": 0.10441367328166962,
      "learning_rate": 0.00011083607022584948,
      "loss": 0.4056,
      "step": 3121
    },
    {
      "epoch": 0.46638780997908574,
      "grad_norm": 0.1470928192138672,
      "learning_rate": 0.00011078934410793453,
      "loss": 0.6346,
      "step": 3122
    },
    {
      "epoch": 0.4665371974902898,
      "grad_norm": 0.6779512763023376,
      "learning_rate": 0.00011074261560648253,
      "loss": 1.1182,
      "step": 3123
    },
    {
      "epoch": 0.46668658500149385,
      "grad_norm": 0.14141033589839935,
      "learning_rate": 0.00011069588473181663,
      "loss": 0.5169,
      "step": 3124
    },
    {
      "epoch": 0.4668359725126979,
      "grad_norm": 0.2803492248058319,
      "learning_rate": 0.00011064915149426038,
      "loss": 1.0145,
      "step": 3125
    },
    {
      "epoch": 0.466985360023902,
      "grad_norm": 0.39002007246017456,
      "learning_rate": 0.00011060241590413787,
      "loss": 0.8957,
      "step": 3126
    },
    {
      "epoch": 0.4671347475351061,
      "grad_norm": 0.14184287190437317,
      "learning_rate": 0.00011055567797177375,
      "loss": 0.4883,
      "step": 3127
    },
    {
      "epoch": 0.4672841350463101,
      "grad_norm": 0.11846556514501572,
      "learning_rate": 0.0001105089377074932,
      "loss": 0.7406,
      "step": 3128
    },
    {
      "epoch": 0.4674335225575142,
      "grad_norm": 0.1450822800397873,
      "learning_rate": 0.00011046219512162184,
      "loss": 0.8389,
      "step": 3129
    },
    {
      "epoch": 0.46758291006871827,
      "grad_norm": 0.1421011984348297,
      "learning_rate": 0.00011041545022448585,
      "loss": 0.8591,
      "step": 3130
    },
    {
      "epoch": 0.4677322975799223,
      "grad_norm": 0.15280935168266296,
      "learning_rate": 0.00011036870302641196,
      "loss": 0.5392,
      "step": 3131
    },
    {
      "epoch": 0.4678816850911264,
      "grad_norm": 0.15633803606033325,
      "learning_rate": 0.00011032195353772732,
      "loss": 0.5298,
      "step": 3132
    },
    {
      "epoch": 0.46803107260233046,
      "grad_norm": 0.09848080575466156,
      "learning_rate": 0.00011027520176875966,
      "loss": 0.4363,
      "step": 3133
    },
    {
      "epoch": 0.46818046011353454,
      "grad_norm": 0.17978063225746155,
      "learning_rate": 0.00011022844772983716,
      "loss": 0.5409,
      "step": 3134
    },
    {
      "epoch": 0.46832984762473856,
      "grad_norm": 0.13446225225925446,
      "learning_rate": 0.00011018169143128857,
      "loss": 0.784,
      "step": 3135
    },
    {
      "epoch": 0.46847923513594264,
      "grad_norm": 0.10704005509614944,
      "learning_rate": 0.00011013493288344307,
      "loss": 0.5516,
      "step": 3136
    },
    {
      "epoch": 0.4686286226471467,
      "grad_norm": 0.24544815719127655,
      "learning_rate": 0.00011008817209663036,
      "loss": 0.9212,
      "step": 3137
    },
    {
      "epoch": 0.46877801015835074,
      "grad_norm": 0.15715865790843964,
      "learning_rate": 0.00011004140908118069,
      "loss": 0.657,
      "step": 3138
    },
    {
      "epoch": 0.4689273976695548,
      "grad_norm": 0.16576170921325684,
      "learning_rate": 0.00010999464384742464,
      "loss": 0.5796,
      "step": 3139
    },
    {
      "epoch": 0.4690767851807589,
      "grad_norm": 0.15970630943775177,
      "learning_rate": 0.00010994787640569348,
      "loss": 0.5201,
      "step": 3140
    },
    {
      "epoch": 0.46922617269196293,
      "grad_norm": 0.15541622042655945,
      "learning_rate": 0.00010990110676631887,
      "loss": 0.8497,
      "step": 3141
    },
    {
      "epoch": 0.469375560203167,
      "grad_norm": 0.21690452098846436,
      "learning_rate": 0.00010985433493963294,
      "loss": 0.8471,
      "step": 3142
    },
    {
      "epoch": 0.4695249477143711,
      "grad_norm": 0.12569521367549896,
      "learning_rate": 0.00010980756093596832,
      "loss": 0.9512,
      "step": 3143
    },
    {
      "epoch": 0.46967433522557517,
      "grad_norm": 0.2046610713005066,
      "learning_rate": 0.00010976078476565818,
      "loss": 0.7953,
      "step": 3144
    },
    {
      "epoch": 0.4698237227367792,
      "grad_norm": 0.14827166497707367,
      "learning_rate": 0.00010971400643903607,
      "loss": 0.6808,
      "step": 3145
    },
    {
      "epoch": 0.4699731102479833,
      "grad_norm": 0.13954679667949677,
      "learning_rate": 0.00010966722596643607,
      "loss": 0.7184,
      "step": 3146
    },
    {
      "epoch": 0.47012249775918735,
      "grad_norm": 0.24878309667110443,
      "learning_rate": 0.0001096204433581927,
      "loss": 0.7849,
      "step": 3147
    },
    {
      "epoch": 0.4702718852703914,
      "grad_norm": 0.14201480150222778,
      "learning_rate": 0.00010957365862464106,
      "loss": 0.4273,
      "step": 3148
    },
    {
      "epoch": 0.47042127278159546,
      "grad_norm": 0.14987538754940033,
      "learning_rate": 0.00010952687177611658,
      "loss": 0.6978,
      "step": 3149
    },
    {
      "epoch": 0.47057066029279954,
      "grad_norm": 0.17707224190235138,
      "learning_rate": 0.00010948008282295523,
      "loss": 0.9736,
      "step": 3150
    },
    {
      "epoch": 0.47072004780400356,
      "grad_norm": 0.2077748030424118,
      "learning_rate": 0.00010943329177549342,
      "loss": 0.7377,
      "step": 3151
    },
    {
      "epoch": 0.47086943531520764,
      "grad_norm": 0.24994945526123047,
      "learning_rate": 0.00010938649864406803,
      "loss": 0.7098,
      "step": 3152
    },
    {
      "epoch": 0.4710188228264117,
      "grad_norm": 0.16622072458267212,
      "learning_rate": 0.00010933970343901643,
      "loss": 0.5084,
      "step": 3153
    },
    {
      "epoch": 0.4711682103376158,
      "grad_norm": 0.1731005162000656,
      "learning_rate": 0.0001092929061706764,
      "loss": 0.902,
      "step": 3154
    },
    {
      "epoch": 0.4713175978488198,
      "grad_norm": 0.13618820905685425,
      "learning_rate": 0.00010924610684938621,
      "loss": 0.9018,
      "step": 3155
    },
    {
      "epoch": 0.4714669853600239,
      "grad_norm": 0.16348271071910858,
      "learning_rate": 0.00010919930548548456,
      "loss": 0.6081,
      "step": 3156
    },
    {
      "epoch": 0.471616372871228,
      "grad_norm": 0.1639767438173294,
      "learning_rate": 0.00010915250208931065,
      "loss": 0.5547,
      "step": 3157
    },
    {
      "epoch": 0.471765760382432,
      "grad_norm": 0.13406722247600555,
      "learning_rate": 0.00010910569667120402,
      "loss": 0.8522,
      "step": 3158
    },
    {
      "epoch": 0.4719151478936361,
      "grad_norm": 0.2892661690711975,
      "learning_rate": 0.00010905888924150478,
      "loss": 0.9023,
      "step": 3159
    },
    {
      "epoch": 0.47206453540484017,
      "grad_norm": 0.13335159420967102,
      "learning_rate": 0.0001090120798105534,
      "loss": 0.5776,
      "step": 3160
    },
    {
      "epoch": 0.4722139229160442,
      "grad_norm": 0.20557230710983276,
      "learning_rate": 0.00010896526838869088,
      "loss": 0.9214,
      "step": 3161
    },
    {
      "epoch": 0.4723633104272483,
      "grad_norm": 0.3257283568382263,
      "learning_rate": 0.00010891845498625857,
      "loss": 1.1597,
      "step": 3162
    },
    {
      "epoch": 0.47251269793845235,
      "grad_norm": 0.13302114605903625,
      "learning_rate": 0.00010887163961359827,
      "loss": 0.6987,
      "step": 3163
    },
    {
      "epoch": 0.47266208544965643,
      "grad_norm": 0.21071402728557587,
      "learning_rate": 0.00010882482228105229,
      "loss": 0.9251,
      "step": 3164
    },
    {
      "epoch": 0.47281147296086046,
      "grad_norm": 0.2068517506122589,
      "learning_rate": 0.00010877800299896325,
      "loss": 0.7064,
      "step": 3165
    },
    {
      "epoch": 0.47296086047206454,
      "grad_norm": 0.20497195422649384,
      "learning_rate": 0.00010873118177767433,
      "loss": 0.782,
      "step": 3166
    },
    {
      "epoch": 0.4731102479832686,
      "grad_norm": 0.21684594452381134,
      "learning_rate": 0.0001086843586275291,
      "loss": 0.9252,
      "step": 3167
    },
    {
      "epoch": 0.47325963549447264,
      "grad_norm": 0.15851536393165588,
      "learning_rate": 0.00010863753355887143,
      "loss": 1.0218,
      "step": 3168
    },
    {
      "epoch": 0.4734090230056767,
      "grad_norm": 0.16891556978225708,
      "learning_rate": 0.00010859070658204582,
      "loss": 0.7818,
      "step": 3169
    },
    {
      "epoch": 0.4735584105168808,
      "grad_norm": 0.14669287204742432,
      "learning_rate": 0.00010854387770739707,
      "loss": 0.6811,
      "step": 3170
    },
    {
      "epoch": 0.4737077980280848,
      "grad_norm": 0.4893437922000885,
      "learning_rate": 0.00010849704694527039,
      "loss": 0.7328,
      "step": 3171
    },
    {
      "epoch": 0.4738571855392889,
      "grad_norm": 0.1580810248851776,
      "learning_rate": 0.00010845021430601143,
      "loss": 0.7486,
      "step": 3172
    },
    {
      "epoch": 0.474006573050493,
      "grad_norm": 0.3402783274650574,
      "learning_rate": 0.00010840337979996634,
      "loss": 0.937,
      "step": 3173
    },
    {
      "epoch": 0.47415596056169707,
      "grad_norm": 0.2276431769132614,
      "learning_rate": 0.00010835654343748149,
      "loss": 0.7072,
      "step": 3174
    },
    {
      "epoch": 0.4743053480729011,
      "grad_norm": 0.1206081211566925,
      "learning_rate": 0.00010830970522890386,
      "loss": 0.4204,
      "step": 3175
    },
    {
      "epoch": 0.47445473558410517,
      "grad_norm": 0.12862072885036469,
      "learning_rate": 0.00010826286518458073,
      "loss": 0.6255,
      "step": 3176
    },
    {
      "epoch": 0.47460412309530925,
      "grad_norm": 0.1384020745754242,
      "learning_rate": 0.0001082160233148598,
      "loss": 0.507,
      "step": 3177
    },
    {
      "epoch": 0.4747535106065133,
      "grad_norm": 0.16803063452243805,
      "learning_rate": 0.00010816917963008916,
      "loss": 0.8284,
      "step": 3178
    },
    {
      "epoch": 0.47490289811771735,
      "grad_norm": 0.2736947238445282,
      "learning_rate": 0.00010812233414061739,
      "loss": 0.7256,
      "step": 3179
    },
    {
      "epoch": 0.47505228562892143,
      "grad_norm": 0.16186657547950745,
      "learning_rate": 0.00010807548685679334,
      "loss": 0.8353,
      "step": 3180
    },
    {
      "epoch": 0.47520167314012546,
      "grad_norm": 0.23407797515392303,
      "learning_rate": 0.00010802863778896635,
      "loss": 0.8564,
      "step": 3181
    },
    {
      "epoch": 0.47535106065132954,
      "grad_norm": 0.30235782265663147,
      "learning_rate": 0.00010798178694748607,
      "loss": 0.6587,
      "step": 3182
    },
    {
      "epoch": 0.4755004481625336,
      "grad_norm": 0.16188107430934906,
      "learning_rate": 0.00010793493434270268,
      "loss": 0.5329,
      "step": 3183
    },
    {
      "epoch": 0.4756498356737377,
      "grad_norm": 0.23610438406467438,
      "learning_rate": 0.00010788807998496655,
      "loss": 0.8462,
      "step": 3184
    },
    {
      "epoch": 0.4757992231849417,
      "grad_norm": 0.11139927804470062,
      "learning_rate": 0.00010784122388462865,
      "loss": 0.6147,
      "step": 3185
    },
    {
      "epoch": 0.4759486106961458,
      "grad_norm": 0.14932647347450256,
      "learning_rate": 0.00010779436605204017,
      "loss": 0.6692,
      "step": 3186
    },
    {
      "epoch": 0.4760979982073499,
      "grad_norm": 0.13887833058834076,
      "learning_rate": 0.00010774750649755279,
      "loss": 0.7901,
      "step": 3187
    },
    {
      "epoch": 0.4762473857185539,
      "grad_norm": 0.38351479172706604,
      "learning_rate": 0.0001077006452315185,
      "loss": 1.2271,
      "step": 3188
    },
    {
      "epoch": 0.476396773229758,
      "grad_norm": 0.12766525149345398,
      "learning_rate": 0.0001076537822642897,
      "loss": 0.8614,
      "step": 3189
    },
    {
      "epoch": 0.47654616074096207,
      "grad_norm": 0.21229498088359833,
      "learning_rate": 0.00010760691760621921,
      "loss": 0.984,
      "step": 3190
    },
    {
      "epoch": 0.47669554825216615,
      "grad_norm": 0.21585416793823242,
      "learning_rate": 0.00010756005126766009,
      "loss": 0.7831,
      "step": 3191
    },
    {
      "epoch": 0.47684493576337017,
      "grad_norm": 0.22231648862361908,
      "learning_rate": 0.00010751318325896592,
      "loss": 1.0953,
      "step": 3192
    },
    {
      "epoch": 0.47699432327457425,
      "grad_norm": 0.14937400817871094,
      "learning_rate": 0.00010746631359049057,
      "loss": 0.5668,
      "step": 3193
    },
    {
      "epoch": 0.47714371078577833,
      "grad_norm": 0.2658711373806,
      "learning_rate": 0.00010741944227258827,
      "loss": 1.1531,
      "step": 3194
    },
    {
      "epoch": 0.47729309829698235,
      "grad_norm": 0.12763851881027222,
      "learning_rate": 0.00010737256931561366,
      "loss": 0.7568,
      "step": 3195
    },
    {
      "epoch": 0.47744248580818643,
      "grad_norm": 0.12254296988248825,
      "learning_rate": 0.00010732569472992171,
      "loss": 0.4182,
      "step": 3196
    },
    {
      "epoch": 0.4775918733193905,
      "grad_norm": 0.1597006469964981,
      "learning_rate": 0.00010727881852586776,
      "loss": 0.6588,
      "step": 3197
    },
    {
      "epoch": 0.47774126083059454,
      "grad_norm": 0.17875833809375763,
      "learning_rate": 0.00010723194071380751,
      "loss": 0.8988,
      "step": 3198
    },
    {
      "epoch": 0.4778906483417986,
      "grad_norm": 0.14364966750144958,
      "learning_rate": 0.00010718506130409702,
      "loss": 0.9703,
      "step": 3199
    },
    {
      "epoch": 0.4780400358530027,
      "grad_norm": 0.12277444452047348,
      "learning_rate": 0.00010713818030709268,
      "loss": 0.6312,
      "step": 3200
    },
    {
      "epoch": 0.4781894233642068,
      "grad_norm": 0.16490934789180756,
      "learning_rate": 0.00010709129773315123,
      "loss": 0.9029,
      "step": 3201
    },
    {
      "epoch": 0.4783388108754108,
      "grad_norm": 0.2033427357673645,
      "learning_rate": 0.00010704441359262982,
      "loss": 1.0855,
      "step": 3202
    },
    {
      "epoch": 0.4784881983866149,
      "grad_norm": 0.11926289647817612,
      "learning_rate": 0.00010699752789588586,
      "loss": 0.5048,
      "step": 3203
    },
    {
      "epoch": 0.47863758589781896,
      "grad_norm": 0.303581178188324,
      "learning_rate": 0.00010695064065327712,
      "loss": 0.9602,
      "step": 3204
    },
    {
      "epoch": 0.478786973409023,
      "grad_norm": 0.1519826352596283,
      "learning_rate": 0.00010690375187516177,
      "loss": 0.6955,
      "step": 3205
    },
    {
      "epoch": 0.47893636092022707,
      "grad_norm": 0.2958853840827942,
      "learning_rate": 0.00010685686157189832,
      "loss": 0.6829,
      "step": 3206
    },
    {
      "epoch": 0.47908574843143115,
      "grad_norm": 0.16560699045658112,
      "learning_rate": 0.00010680996975384552,
      "loss": 0.7845,
      "step": 3207
    },
    {
      "epoch": 0.47923513594263517,
      "grad_norm": 0.16155114769935608,
      "learning_rate": 0.00010676307643136254,
      "loss": 0.7715,
      "step": 3208
    },
    {
      "epoch": 0.47938452345383925,
      "grad_norm": 0.2867703437805176,
      "learning_rate": 0.00010671618161480887,
      "loss": 0.9398,
      "step": 3209
    },
    {
      "epoch": 0.47953391096504333,
      "grad_norm": 0.15584948658943176,
      "learning_rate": 0.00010666928531454428,
      "loss": 0.7899,
      "step": 3210
    },
    {
      "epoch": 0.4796832984762474,
      "grad_norm": 0.16091302037239075,
      "learning_rate": 0.00010662238754092892,
      "loss": 0.5302,
      "step": 3211
    },
    {
      "epoch": 0.47983268598745143,
      "grad_norm": 0.2147711217403412,
      "learning_rate": 0.00010657548830432329,
      "loss": 0.7911,
      "step": 3212
    },
    {
      "epoch": 0.4799820734986555,
      "grad_norm": 0.253465861082077,
      "learning_rate": 0.00010652858761508813,
      "loss": 0.8169,
      "step": 3213
    },
    {
      "epoch": 0.4801314610098596,
      "grad_norm": 0.1429709792137146,
      "learning_rate": 0.00010648168548358455,
      "loss": 0.7799,
      "step": 3214
    },
    {
      "epoch": 0.4802808485210636,
      "grad_norm": 0.1140756905078888,
      "learning_rate": 0.000106434781920174,
      "loss": 0.6006,
      "step": 3215
    },
    {
      "epoch": 0.4804302360322677,
      "grad_norm": 0.16223487257957458,
      "learning_rate": 0.00010638787693521819,
      "loss": 0.4977,
      "step": 3216
    },
    {
      "epoch": 0.4805796235434718,
      "grad_norm": 0.12336257845163345,
      "learning_rate": 0.00010634097053907916,
      "loss": 0.5733,
      "step": 3217
    },
    {
      "epoch": 0.4807290110546758,
      "grad_norm": 0.17670951783657074,
      "learning_rate": 0.00010629406274211934,
      "loss": 0.9,
      "step": 3218
    },
    {
      "epoch": 0.4808783985658799,
      "grad_norm": 0.12306135892868042,
      "learning_rate": 0.00010624715355470136,
      "loss": 0.7489,
      "step": 3219
    },
    {
      "epoch": 0.48102778607708396,
      "grad_norm": 0.18291476368904114,
      "learning_rate": 0.00010620024298718822,
      "loss": 0.5989,
      "step": 3220
    },
    {
      "epoch": 0.48117717358828804,
      "grad_norm": 0.13022394478321075,
      "learning_rate": 0.00010615333104994318,
      "loss": 0.6436,
      "step": 3221
    },
    {
      "epoch": 0.48132656109949207,
      "grad_norm": 0.2752431333065033,
      "learning_rate": 0.00010610641775332983,
      "loss": 0.8934,
      "step": 3222
    },
    {
      "epoch": 0.48147594861069615,
      "grad_norm": 0.18069930374622345,
      "learning_rate": 0.00010605950310771212,
      "loss": 0.8267,
      "step": 3223
    },
    {
      "epoch": 0.4816253361219002,
      "grad_norm": 0.22545437514781952,
      "learning_rate": 0.00010601258712345414,
      "loss": 0.8824,
      "step": 3224
    },
    {
      "epoch": 0.48177472363310425,
      "grad_norm": 0.1509438008069992,
      "learning_rate": 0.00010596566981092049,
      "loss": 0.7042,
      "step": 3225
    },
    {
      "epoch": 0.48192411114430833,
      "grad_norm": 0.13593561947345734,
      "learning_rate": 0.00010591875118047588,
      "loss": 0.9899,
      "step": 3226
    },
    {
      "epoch": 0.4820734986555124,
      "grad_norm": 0.15942330658435822,
      "learning_rate": 0.00010587183124248534,
      "loss": 0.9158,
      "step": 3227
    },
    {
      "epoch": 0.48222288616671644,
      "grad_norm": 0.13079741597175598,
      "learning_rate": 0.00010582491000731432,
      "loss": 0.746,
      "step": 3228
    },
    {
      "epoch": 0.4823722736779205,
      "grad_norm": 0.1051473543047905,
      "learning_rate": 0.00010577798748532841,
      "loss": 0.4886,
      "step": 3229
    },
    {
      "epoch": 0.4825216611891246,
      "grad_norm": 0.2743512988090515,
      "learning_rate": 0.00010573106368689352,
      "loss": 0.6236,
      "step": 3230
    },
    {
      "epoch": 0.4826710487003287,
      "grad_norm": 0.13678042590618134,
      "learning_rate": 0.0001056841386223759,
      "loss": 0.8298,
      "step": 3231
    },
    {
      "epoch": 0.4828204362115327,
      "grad_norm": 0.1541474610567093,
      "learning_rate": 0.00010563721230214203,
      "loss": 0.6308,
      "step": 3232
    },
    {
      "epoch": 0.4829698237227368,
      "grad_norm": 0.16513827443122864,
      "learning_rate": 0.00010559028473655866,
      "loss": 0.9199,
      "step": 3233
    },
    {
      "epoch": 0.48311921123394086,
      "grad_norm": 0.21207313239574432,
      "learning_rate": 0.00010554335593599285,
      "loss": 0.6461,
      "step": 3234
    },
    {
      "epoch": 0.4832685987451449,
      "grad_norm": 0.18756775557994843,
      "learning_rate": 0.00010549642591081194,
      "loss": 0.9897,
      "step": 3235
    },
    {
      "epoch": 0.48341798625634896,
      "grad_norm": 0.1757032871246338,
      "learning_rate": 0.00010544949467138346,
      "loss": 0.5895,
      "step": 3236
    },
    {
      "epoch": 0.48356737376755304,
      "grad_norm": 0.23568972945213318,
      "learning_rate": 0.0001054025622280753,
      "loss": 1.0592,
      "step": 3237
    },
    {
      "epoch": 0.4837167612787571,
      "grad_norm": 0.2145964503288269,
      "learning_rate": 0.00010535562859125558,
      "loss": 0.5977,
      "step": 3238
    },
    {
      "epoch": 0.48386614878996115,
      "grad_norm": 0.26484590768814087,
      "learning_rate": 0.00010530869377129267,
      "loss": 0.8678,
      "step": 3239
    },
    {
      "epoch": 0.48401553630116523,
      "grad_norm": 0.15472498536109924,
      "learning_rate": 0.0001052617577785552,
      "loss": 0.5646,
      "step": 3240
    },
    {
      "epoch": 0.4841649238123693,
      "grad_norm": 0.16602066159248352,
      "learning_rate": 0.0001052148206234121,
      "loss": 0.404,
      "step": 3241
    },
    {
      "epoch": 0.48431431132357333,
      "grad_norm": 0.1432691365480423,
      "learning_rate": 0.00010516788231623253,
      "loss": 0.6054,
      "step": 3242
    },
    {
      "epoch": 0.4844636988347774,
      "grad_norm": 0.1686859279870987,
      "learning_rate": 0.00010512094286738592,
      "loss": 0.7051,
      "step": 3243
    },
    {
      "epoch": 0.4846130863459815,
      "grad_norm": 0.161126509308815,
      "learning_rate": 0.00010507400228724192,
      "loss": 0.6088,
      "step": 3244
    },
    {
      "epoch": 0.4847624738571855,
      "grad_norm": 0.18919625878334045,
      "learning_rate": 0.00010502706058617042,
      "loss": 0.9567,
      "step": 3245
    },
    {
      "epoch": 0.4849118613683896,
      "grad_norm": 0.13224321603775024,
      "learning_rate": 0.00010498011777454163,
      "loss": 0.8173,
      "step": 3246
    },
    {
      "epoch": 0.4850612488795937,
      "grad_norm": 0.18885143101215363,
      "learning_rate": 0.00010493317386272593,
      "loss": 0.6827,
      "step": 3247
    },
    {
      "epoch": 0.48521063639079776,
      "grad_norm": 0.16263827681541443,
      "learning_rate": 0.000104886228861094,
      "loss": 0.9218,
      "step": 3248
    },
    {
      "epoch": 0.4853600239020018,
      "grad_norm": 0.18450219929218292,
      "learning_rate": 0.00010483928278001669,
      "loss": 0.4582,
      "step": 3249
    },
    {
      "epoch": 0.48550941141320586,
      "grad_norm": 0.18349440395832062,
      "learning_rate": 0.00010479233562986519,
      "loss": 0.7357,
      "step": 3250
    },
    {
      "epoch": 0.48565879892440994,
      "grad_norm": 0.1312660127878189,
      "learning_rate": 0.00010474538742101085,
      "loss": 0.7591,
      "step": 3251
    },
    {
      "epoch": 0.48580818643561396,
      "grad_norm": 0.14447490870952606,
      "learning_rate": 0.00010469843816382526,
      "loss": 0.4734,
      "step": 3252
    },
    {
      "epoch": 0.48595757394681804,
      "grad_norm": 0.13847716152668,
      "learning_rate": 0.00010465148786868024,
      "loss": 0.8431,
      "step": 3253
    },
    {
      "epoch": 0.4861069614580221,
      "grad_norm": 0.20934633910655975,
      "learning_rate": 0.0001046045365459479,
      "loss": 0.6299,
      "step": 3254
    },
    {
      "epoch": 0.48625634896922615,
      "grad_norm": 0.19869495928287506,
      "learning_rate": 0.0001045575842060005,
      "loss": 0.4894,
      "step": 3255
    },
    {
      "epoch": 0.48640573648043023,
      "grad_norm": 0.14602303504943848,
      "learning_rate": 0.00010451063085921056,
      "loss": 0.6349,
      "step": 3256
    },
    {
      "epoch": 0.4865551239916343,
      "grad_norm": 0.13285601139068604,
      "learning_rate": 0.00010446367651595081,
      "loss": 0.8707,
      "step": 3257
    },
    {
      "epoch": 0.4867045115028384,
      "grad_norm": 0.1567457616329193,
      "learning_rate": 0.00010441672118659422,
      "loss": 0.8926,
      "step": 3258
    },
    {
      "epoch": 0.4868538990140424,
      "grad_norm": 0.14940892159938812,
      "learning_rate": 0.00010436976488151395,
      "loss": 0.594,
      "step": 3259
    },
    {
      "epoch": 0.4870032865252465,
      "grad_norm": 0.1392555981874466,
      "learning_rate": 0.00010432280761108342,
      "loss": 0.5553,
      "step": 3260
    },
    {
      "epoch": 0.4871526740364506,
      "grad_norm": 0.1326533704996109,
      "learning_rate": 0.00010427584938567623,
      "loss": 0.391,
      "step": 3261
    },
    {
      "epoch": 0.4873020615476546,
      "grad_norm": 0.1550893783569336,
      "learning_rate": 0.00010422889021566618,
      "loss": 0.7032,
      "step": 3262
    },
    {
      "epoch": 0.4874514490588587,
      "grad_norm": 0.1566886305809021,
      "learning_rate": 0.00010418193011142727,
      "loss": 0.6163,
      "step": 3263
    },
    {
      "epoch": 0.48760083657006276,
      "grad_norm": 0.1285940259695053,
      "learning_rate": 0.0001041349690833338,
      "loss": 0.6255,
      "step": 3264
    },
    {
      "epoch": 0.4877502240812668,
      "grad_norm": 0.14291048049926758,
      "learning_rate": 0.00010408800714176015,
      "loss": 0.8496,
      "step": 3265
    },
    {
      "epoch": 0.48789961159247086,
      "grad_norm": 0.19264325499534607,
      "learning_rate": 0.00010404104429708097,
      "loss": 0.8868,
      "step": 3266
    },
    {
      "epoch": 0.48804899910367494,
      "grad_norm": 0.34244269132614136,
      "learning_rate": 0.00010399408055967108,
      "loss": 1.0975,
      "step": 3267
    },
    {
      "epoch": 0.488198386614879,
      "grad_norm": 0.17104557156562805,
      "learning_rate": 0.00010394711593990554,
      "loss": 0.5284,
      "step": 3268
    },
    {
      "epoch": 0.48834777412608305,
      "grad_norm": 0.16921040415763855,
      "learning_rate": 0.00010390015044815958,
      "loss": 0.6267,
      "step": 3269
    },
    {
      "epoch": 0.4884971616372871,
      "grad_norm": 0.12005407363176346,
      "learning_rate": 0.00010385318409480862,
      "loss": 0.7159,
      "step": 3270
    },
    {
      "epoch": 0.4886465491484912,
      "grad_norm": 0.7021434903144836,
      "learning_rate": 0.00010380621689022825,
      "loss": 1.5941,
      "step": 3271
    },
    {
      "epoch": 0.48879593665969523,
      "grad_norm": 0.26370319724082947,
      "learning_rate": 0.00010375924884479427,
      "loss": 0.9023,
      "step": 3272
    },
    {
      "epoch": 0.4889453241708993,
      "grad_norm": 0.11657514423131943,
      "learning_rate": 0.00010371227996888271,
      "loss": 0.5922,
      "step": 3273
    },
    {
      "epoch": 0.4890947116821034,
      "grad_norm": 0.2122240513563156,
      "learning_rate": 0.00010366531027286967,
      "loss": 0.468,
      "step": 3274
    },
    {
      "epoch": 0.4892440991933074,
      "grad_norm": 0.1361604481935501,
      "learning_rate": 0.00010361833976713155,
      "loss": 0.858,
      "step": 3275
    },
    {
      "epoch": 0.4893934867045115,
      "grad_norm": 0.13386155664920807,
      "learning_rate": 0.00010357136846204487,
      "loss": 0.4652,
      "step": 3276
    },
    {
      "epoch": 0.4895428742157156,
      "grad_norm": 0.12180904299020767,
      "learning_rate": 0.00010352439636798632,
      "loss": 0.4927,
      "step": 3277
    },
    {
      "epoch": 0.48969226172691965,
      "grad_norm": 0.16701848804950714,
      "learning_rate": 0.00010347742349533278,
      "loss": 0.8225,
      "step": 3278
    },
    {
      "epoch": 0.4898416492381237,
      "grad_norm": 0.17934517562389374,
      "learning_rate": 0.0001034304498544613,
      "loss": 1.0569,
      "step": 3279
    },
    {
      "epoch": 0.48999103674932776,
      "grad_norm": 0.11470920592546463,
      "learning_rate": 0.00010338347545574916,
      "loss": 0.6879,
      "step": 3280
    },
    {
      "epoch": 0.49014042426053184,
      "grad_norm": 0.14119862020015717,
      "learning_rate": 0.00010333650030957366,
      "loss": 0.7557,
      "step": 3281
    },
    {
      "epoch": 0.49028981177173586,
      "grad_norm": 0.1528460532426834,
      "learning_rate": 0.00010328952442631241,
      "loss": 0.8541,
      "step": 3282
    },
    {
      "epoch": 0.49043919928293994,
      "grad_norm": 0.13389696180820465,
      "learning_rate": 0.00010324254781634312,
      "loss": 0.8753,
      "step": 3283
    },
    {
      "epoch": 0.490588586794144,
      "grad_norm": 0.2853409945964813,
      "learning_rate": 0.00010319557049004365,
      "loss": 0.6021,
      "step": 3284
    },
    {
      "epoch": 0.49073797430534805,
      "grad_norm": 0.14676959812641144,
      "learning_rate": 0.00010314859245779202,
      "loss": 0.7133,
      "step": 3285
    },
    {
      "epoch": 0.4908873618165521,
      "grad_norm": 0.17747355997562408,
      "learning_rate": 0.00010310161372996648,
      "loss": 0.7222,
      "step": 3286
    },
    {
      "epoch": 0.4910367493277562,
      "grad_norm": 0.28462037444114685,
      "learning_rate": 0.0001030546343169453,
      "loss": 0.8593,
      "step": 3287
    },
    {
      "epoch": 0.4911861368389603,
      "grad_norm": 0.28177785873413086,
      "learning_rate": 0.00010300765422910706,
      "loss": 0.9987,
      "step": 3288
    },
    {
      "epoch": 0.4913355243501643,
      "grad_norm": 0.11662929505109787,
      "learning_rate": 0.00010296067347683034,
      "loss": 0.4956,
      "step": 3289
    },
    {
      "epoch": 0.4914849118613684,
      "grad_norm": 0.19040469825267792,
      "learning_rate": 0.00010291369207049397,
      "loss": 0.6833,
      "step": 3290
    },
    {
      "epoch": 0.49163429937257247,
      "grad_norm": 0.18772046267986298,
      "learning_rate": 0.00010286671002047685,
      "loss": 0.3936,
      "step": 3291
    },
    {
      "epoch": 0.4917836868837765,
      "grad_norm": 0.1391916424036026,
      "learning_rate": 0.00010281972733715808,
      "loss": 0.5701,
      "step": 3292
    },
    {
      "epoch": 0.4919330743949806,
      "grad_norm": 0.3673042058944702,
      "learning_rate": 0.0001027727440309169,
      "loss": 0.8775,
      "step": 3293
    },
    {
      "epoch": 0.49208246190618465,
      "grad_norm": 0.15252585709095,
      "learning_rate": 0.00010272576011213262,
      "loss": 0.8585,
      "step": 3294
    },
    {
      "epoch": 0.49223184941738873,
      "grad_norm": 0.2030603438615799,
      "learning_rate": 0.00010267877559118478,
      "loss": 0.7116,
      "step": 3295
    },
    {
      "epoch": 0.49238123692859276,
      "grad_norm": 0.3568076193332672,
      "learning_rate": 0.00010263179047845297,
      "loss": 0.9962,
      "step": 3296
    },
    {
      "epoch": 0.49253062443979684,
      "grad_norm": 0.11111945658922195,
      "learning_rate": 0.00010258480478431695,
      "loss": 0.5714,
      "step": 3297
    },
    {
      "epoch": 0.4926800119510009,
      "grad_norm": 0.18714840710163116,
      "learning_rate": 0.00010253781851915663,
      "loss": 0.9194,
      "step": 3298
    },
    {
      "epoch": 0.49282939946220494,
      "grad_norm": 0.19076299667358398,
      "learning_rate": 0.00010249083169335201,
      "loss": 0.5261,
      "step": 3299
    },
    {
      "epoch": 0.492978786973409,
      "grad_norm": 0.1914680004119873,
      "learning_rate": 0.0001024438443172832,
      "loss": 0.7332,
      "step": 3300
    },
    {
      "epoch": 0.4931281744846131,
      "grad_norm": 0.12699691951274872,
      "learning_rate": 0.00010239685640133045,
      "loss": 0.7384,
      "step": 3301
    },
    {
      "epoch": 0.4932775619958171,
      "grad_norm": 0.18444816768169403,
      "learning_rate": 0.00010234986795587418,
      "loss": 0.5359,
      "step": 3302
    },
    {
      "epoch": 0.4934269495070212,
      "grad_norm": 0.23525670170783997,
      "learning_rate": 0.00010230287899129487,
      "loss": 1.0311,
      "step": 3303
    },
    {
      "epoch": 0.4935763370182253,
      "grad_norm": 0.16428425908088684,
      "learning_rate": 0.00010225588951797309,
      "loss": 0.786,
      "step": 3304
    },
    {
      "epoch": 0.49372572452942937,
      "grad_norm": 0.12360762804746628,
      "learning_rate": 0.00010220889954628961,
      "loss": 0.7882,
      "step": 3305
    },
    {
      "epoch": 0.4938751120406334,
      "grad_norm": 0.23710285127162933,
      "learning_rate": 0.00010216190908662522,
      "loss": 0.9197,
      "step": 3306
    },
    {
      "epoch": 0.49402449955183747,
      "grad_norm": 0.33589887619018555,
      "learning_rate": 0.00010211491814936087,
      "loss": 0.9662,
      "step": 3307
    },
    {
      "epoch": 0.49417388706304155,
      "grad_norm": 0.1265798807144165,
      "learning_rate": 0.0001020679267448776,
      "loss": 0.8177,
      "step": 3308
    },
    {
      "epoch": 0.4943232745742456,
      "grad_norm": 0.12427424639463425,
      "learning_rate": 0.00010202093488355657,
      "loss": 0.6632,
      "step": 3309
    },
    {
      "epoch": 0.49447266208544965,
      "grad_norm": 0.1983823925256729,
      "learning_rate": 0.00010197394257577902,
      "loss": 0.9116,
      "step": 3310
    },
    {
      "epoch": 0.49462204959665373,
      "grad_norm": 0.11778527498245239,
      "learning_rate": 0.00010192694983192626,
      "loss": 0.8034,
      "step": 3311
    },
    {
      "epoch": 0.49477143710785776,
      "grad_norm": 0.17188487946987152,
      "learning_rate": 0.00010187995666237977,
      "loss": 0.9057,
      "step": 3312
    },
    {
      "epoch": 0.49492082461906184,
      "grad_norm": 0.12710382044315338,
      "learning_rate": 0.00010183296307752108,
      "loss": 0.5904,
      "step": 3313
    },
    {
      "epoch": 0.4950702121302659,
      "grad_norm": 0.1679123491048813,
      "learning_rate": 0.00010178596908773179,
      "loss": 0.7233,
      "step": 3314
    },
    {
      "epoch": 0.49521959964147,
      "grad_norm": 0.11891725659370422,
      "learning_rate": 0.00010173897470339369,
      "loss": 0.6146,
      "step": 3315
    },
    {
      "epoch": 0.495368987152674,
      "grad_norm": 0.14667171239852905,
      "learning_rate": 0.00010169197993488851,
      "loss": 0.7802,
      "step": 3316
    },
    {
      "epoch": 0.4955183746638781,
      "grad_norm": 0.1279142200946808,
      "learning_rate": 0.00010164498479259817,
      "loss": 0.798,
      "step": 3317
    },
    {
      "epoch": 0.4956677621750822,
      "grad_norm": 0.13514754176139832,
      "learning_rate": 0.0001015979892869046,
      "loss": 0.5317,
      "step": 3318
    },
    {
      "epoch": 0.4958171496862862,
      "grad_norm": 0.1267642378807068,
      "learning_rate": 0.00010155099342818995,
      "loss": 0.6149,
      "step": 3319
    },
    {
      "epoch": 0.4959665371974903,
      "grad_norm": 0.1201544776558876,
      "learning_rate": 0.00010150399722683623,
      "loss": 0.72,
      "step": 3320
    },
    {
      "epoch": 0.49611592470869437,
      "grad_norm": 0.11963790655136108,
      "learning_rate": 0.00010145700069322572,
      "loss": 0.726,
      "step": 3321
    },
    {
      "epoch": 0.4962653122198984,
      "grad_norm": 0.15055450797080994,
      "learning_rate": 0.00010141000383774067,
      "loss": 0.9086,
      "step": 3322
    },
    {
      "epoch": 0.49641469973110247,
      "grad_norm": 0.13218680024147034,
      "learning_rate": 0.00010136300667076344,
      "loss": 0.5336,
      "step": 3323
    },
    {
      "epoch": 0.49656408724230655,
      "grad_norm": 0.14864204823970795,
      "learning_rate": 0.00010131600920267645,
      "loss": 0.3602,
      "step": 3324
    },
    {
      "epoch": 0.49671347475351063,
      "grad_norm": 0.12407877296209335,
      "learning_rate": 0.00010126901144386221,
      "loss": 0.7472,
      "step": 3325
    },
    {
      "epoch": 0.49686286226471466,
      "grad_norm": 0.1972523182630539,
      "learning_rate": 0.00010122201340470321,
      "loss": 0.8393,
      "step": 3326
    },
    {
      "epoch": 0.49701224977591874,
      "grad_norm": 0.46555835008621216,
      "learning_rate": 0.0001011750150955821,
      "loss": 1.1339,
      "step": 3327
    },
    {
      "epoch": 0.4971616372871228,
      "grad_norm": 0.15617221593856812,
      "learning_rate": 0.00010112801652688155,
      "loss": 0.7224,
      "step": 3328
    },
    {
      "epoch": 0.49731102479832684,
      "grad_norm": 0.11712933331727982,
      "learning_rate": 0.00010108101770898428,
      "loss": 0.6659,
      "step": 3329
    },
    {
      "epoch": 0.4974604123095309,
      "grad_norm": 0.12405367940664291,
      "learning_rate": 0.00010103401865227304,
      "loss": 0.8377,
      "step": 3330
    },
    {
      "epoch": 0.497609799820735,
      "grad_norm": 0.11650660634040833,
      "learning_rate": 0.00010098701936713073,
      "loss": 0.7012,
      "step": 3331
    },
    {
      "epoch": 0.497759187331939,
      "grad_norm": 0.14338290691375732,
      "learning_rate": 0.00010094001986394024,
      "loss": 1.0737,
      "step": 3332
    },
    {
      "epoch": 0.4979085748431431,
      "grad_norm": 0.17379824817180634,
      "learning_rate": 0.00010089302015308444,
      "loss": 0.8722,
      "step": 3333
    },
    {
      "epoch": 0.4980579623543472,
      "grad_norm": 0.1527802050113678,
      "learning_rate": 0.00010084602024494633,
      "loss": 0.6456,
      "step": 3334
    },
    {
      "epoch": 0.49820734986555126,
      "grad_norm": 0.1445491760969162,
      "learning_rate": 0.00010079902014990899,
      "loss": 0.6007,
      "step": 3335
    },
    {
      "epoch": 0.4983567373767553,
      "grad_norm": 0.11192949116230011,
      "learning_rate": 0.0001007520198783554,
      "loss": 0.6109,
      "step": 3336
    },
    {
      "epoch": 0.49850612488795937,
      "grad_norm": 0.28887754678726196,
      "learning_rate": 0.00010070501944066875,
      "loss": 1.452,
      "step": 3337
    },
    {
      "epoch": 0.49865551239916345,
      "grad_norm": 0.14257770776748657,
      "learning_rate": 0.0001006580188472321,
      "loss": 0.8282,
      "step": 3338
    },
    {
      "epoch": 0.49880489991036747,
      "grad_norm": 0.19091013073921204,
      "learning_rate": 0.00010061101810842867,
      "loss": 1.036,
      "step": 3339
    },
    {
      "epoch": 0.49895428742157155,
      "grad_norm": 0.13388298451900482,
      "learning_rate": 0.00010056401723464166,
      "loss": 0.6776,
      "step": 3340
    },
    {
      "epoch": 0.49910367493277563,
      "grad_norm": 0.14166954159736633,
      "learning_rate": 0.00010051701623625435,
      "loss": 0.7404,
      "step": 3341
    },
    {
      "epoch": 0.49925306244397966,
      "grad_norm": 0.1938195824623108,
      "learning_rate": 0.00010047001512364992,
      "loss": 1.2493,
      "step": 3342
    },
    {
      "epoch": 0.49940244995518374,
      "grad_norm": 0.13798008859157562,
      "learning_rate": 0.00010042301390721171,
      "loss": 0.7332,
      "step": 3343
    },
    {
      "epoch": 0.4995518374663878,
      "grad_norm": 0.15777172148227692,
      "learning_rate": 0.00010037601259732308,
      "loss": 0.7001,
      "step": 3344
    },
    {
      "epoch": 0.4997012249775919,
      "grad_norm": 0.1827012151479721,
      "learning_rate": 0.00010032901120436726,
      "loss": 0.6007,
      "step": 3345
    },
    {
      "epoch": 0.4998506124887959,
      "grad_norm": 0.17342330515384674,
      "learning_rate": 0.00010028200973872766,
      "loss": 0.8919,
      "step": 3346
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.12758640944957733,
      "learning_rate": 0.00010023500821078765,
      "loss": 0.7974,
      "step": 3347
    },
    {
      "epoch": 0.5001493875112041,
      "grad_norm": 0.1501816064119339,
      "learning_rate": 0.00010018800663093057,
      "loss": 0.7668,
      "step": 3348
    },
    {
      "epoch": 0.5002987750224082,
      "grad_norm": 0.15742836892604828,
      "learning_rate": 0.00010014100500953989,
      "loss": 0.8888,
      "step": 3349
    },
    {
      "epoch": 0.5004481625336122,
      "grad_norm": 0.1827993243932724,
      "learning_rate": 0.00010009400335699894,
      "loss": 0.8371,
      "step": 3350
    },
    {
      "epoch": 0.5005975500448162,
      "grad_norm": 0.12702827155590057,
      "learning_rate": 0.00010004700168369119,
      "loss": 0.5703,
      "step": 3351
    },
    {
      "epoch": 0.5007469375560203,
      "grad_norm": 0.1451091766357422,
      "learning_rate": 0.0001,
      "loss": 0.9624,
      "step": 3352
    },
    {
      "epoch": 0.5008963250672244,
      "grad_norm": 0.195283904671669,
      "learning_rate": 9.995299831630882e-05,
      "loss": 0.8291,
      "step": 3353
    },
    {
      "epoch": 0.5010457125784284,
      "grad_norm": 0.1774168759584427,
      "learning_rate": 9.990599664300105e-05,
      "loss": 0.7494,
      "step": 3354
    },
    {
      "epoch": 0.5011951000896325,
      "grad_norm": 0.1609886884689331,
      "learning_rate": 9.985899499046012e-05,
      "loss": 0.6764,
      "step": 3355
    },
    {
      "epoch": 0.5013444876008366,
      "grad_norm": 0.10945264250040054,
      "learning_rate": 9.981199336906944e-05,
      "loss": 0.5207,
      "step": 3356
    },
    {
      "epoch": 0.5014938751120407,
      "grad_norm": 0.14048771560192108,
      "learning_rate": 9.976499178921239e-05,
      "loss": 0.4921,
      "step": 3357
    },
    {
      "epoch": 0.5016432626232447,
      "grad_norm": 0.16898074746131897,
      "learning_rate": 9.971799026127236e-05,
      "loss": 0.6298,
      "step": 3358
    },
    {
      "epoch": 0.5017926501344487,
      "grad_norm": 0.11817678064107895,
      "learning_rate": 9.967098879563277e-05,
      "loss": 0.6325,
      "step": 3359
    },
    {
      "epoch": 0.5019420376456528,
      "grad_norm": 0.1395687311887741,
      "learning_rate": 9.962398740267696e-05,
      "loss": 0.6824,
      "step": 3360
    },
    {
      "epoch": 0.5020914251568569,
      "grad_norm": 0.14531132578849792,
      "learning_rate": 9.957698609278827e-05,
      "loss": 0.6157,
      "step": 3361
    },
    {
      "epoch": 0.502240812668061,
      "grad_norm": 0.12905453145503998,
      "learning_rate": 9.952998487635011e-05,
      "loss": 0.6195,
      "step": 3362
    },
    {
      "epoch": 0.5023902001792651,
      "grad_norm": 0.11037211865186691,
      "learning_rate": 9.94829837637457e-05,
      "loss": 0.5779,
      "step": 3363
    },
    {
      "epoch": 0.502539587690469,
      "grad_norm": 0.18889397382736206,
      "learning_rate": 9.943598276535835e-05,
      "loss": 0.477,
      "step": 3364
    },
    {
      "epoch": 0.5026889752016731,
      "grad_norm": 0.24633678793907166,
      "learning_rate": 9.938898189157135e-05,
      "loss": 0.75,
      "step": 3365
    },
    {
      "epoch": 0.5028383627128772,
      "grad_norm": 0.14785300195217133,
      "learning_rate": 9.934198115276793e-05,
      "loss": 0.5637,
      "step": 3366
    },
    {
      "epoch": 0.5029877502240813,
      "grad_norm": 0.6878348588943481,
      "learning_rate": 9.929498055933126e-05,
      "loss": 1.0777,
      "step": 3367
    },
    {
      "epoch": 0.5031371377352853,
      "grad_norm": 0.17142246663570404,
      "learning_rate": 9.924798012164459e-05,
      "loss": 0.6047,
      "step": 3368
    },
    {
      "epoch": 0.5032865252464894,
      "grad_norm": 0.17519189417362213,
      "learning_rate": 9.920097985009105e-05,
      "loss": 0.6441,
      "step": 3369
    },
    {
      "epoch": 0.5034359127576935,
      "grad_norm": 0.22075048089027405,
      "learning_rate": 9.915397975505369e-05,
      "loss": 0.7218,
      "step": 3370
    },
    {
      "epoch": 0.5035853002688975,
      "grad_norm": 0.12722843885421753,
      "learning_rate": 9.91069798469156e-05,
      "loss": 0.6702,
      "step": 3371
    },
    {
      "epoch": 0.5037346877801016,
      "grad_norm": 0.219132661819458,
      "learning_rate": 9.90599801360598e-05,
      "loss": 0.8335,
      "step": 3372
    },
    {
      "epoch": 0.5038840752913056,
      "grad_norm": 0.16751305758953094,
      "learning_rate": 9.901298063286927e-05,
      "loss": 0.3965,
      "step": 3373
    },
    {
      "epoch": 0.5040334628025097,
      "grad_norm": 0.2040434628725052,
      "learning_rate": 9.896598134772697e-05,
      "loss": 0.9113,
      "step": 3374
    },
    {
      "epoch": 0.5041828503137138,
      "grad_norm": 0.23715925216674805,
      "learning_rate": 9.891898229101577e-05,
      "loss": 1.1276,
      "step": 3375
    },
    {
      "epoch": 0.5043322378249179,
      "grad_norm": 0.14151549339294434,
      "learning_rate": 9.887198347311849e-05,
      "loss": 0.6685,
      "step": 3376
    },
    {
      "epoch": 0.504481625336122,
      "grad_norm": 0.19641682505607605,
      "learning_rate": 9.882498490441794e-05,
      "loss": 0.6833,
      "step": 3377
    },
    {
      "epoch": 0.5046310128473259,
      "grad_norm": 0.14036989212036133,
      "learning_rate": 9.877798659529683e-05,
      "loss": 0.9383,
      "step": 3378
    },
    {
      "epoch": 0.50478040035853,
      "grad_norm": 0.24824388325214386,
      "learning_rate": 9.873098855613783e-05,
      "loss": 0.7151,
      "step": 3379
    },
    {
      "epoch": 0.5049297878697341,
      "grad_norm": 0.22324711084365845,
      "learning_rate": 9.868399079732356e-05,
      "loss": 0.3287,
      "step": 3380
    },
    {
      "epoch": 0.5050791753809382,
      "grad_norm": 0.14349500834941864,
      "learning_rate": 9.863699332923656e-05,
      "loss": 0.486,
      "step": 3381
    },
    {
      "epoch": 0.5052285628921422,
      "grad_norm": 0.1973506659269333,
      "learning_rate": 9.858999616225939e-05,
      "loss": 0.6452,
      "step": 3382
    },
    {
      "epoch": 0.5053779504033463,
      "grad_norm": 0.11889201402664185,
      "learning_rate": 9.854299930677431e-05,
      "loss": 0.6023,
      "step": 3383
    },
    {
      "epoch": 0.5055273379145503,
      "grad_norm": 0.1579374223947525,
      "learning_rate": 9.849600277316379e-05,
      "loss": 0.7812,
      "step": 3384
    },
    {
      "epoch": 0.5056767254257544,
      "grad_norm": 0.14273232221603394,
      "learning_rate": 9.844900657181009e-05,
      "loss": 0.9102,
      "step": 3385
    },
    {
      "epoch": 0.5058261129369584,
      "grad_norm": 0.14018802344799042,
      "learning_rate": 9.840201071309539e-05,
      "loss": 0.8549,
      "step": 3386
    },
    {
      "epoch": 0.5059755004481625,
      "grad_norm": 0.1488267481327057,
      "learning_rate": 9.835501520740184e-05,
      "loss": 0.507,
      "step": 3387
    },
    {
      "epoch": 0.5061248879593666,
      "grad_norm": 0.19517706334590912,
      "learning_rate": 9.830802006511154e-05,
      "loss": 0.7193,
      "step": 3388
    },
    {
      "epoch": 0.5062742754705707,
      "grad_norm": 0.1316169947385788,
      "learning_rate": 9.826102529660635e-05,
      "loss": 0.7118,
      "step": 3389
    },
    {
      "epoch": 0.5064236629817748,
      "grad_norm": 0.22407954931259155,
      "learning_rate": 9.821403091226822e-05,
      "loss": 0.7368,
      "step": 3390
    },
    {
      "epoch": 0.5065730504929787,
      "grad_norm": 0.14248794317245483,
      "learning_rate": 9.816703692247895e-05,
      "loss": 0.8912,
      "step": 3391
    },
    {
      "epoch": 0.5067224380041828,
      "grad_norm": 0.1354566514492035,
      "learning_rate": 9.812004333762027e-05,
      "loss": 0.7897,
      "step": 3392
    },
    {
      "epoch": 0.5068718255153869,
      "grad_norm": 0.20816969871520996,
      "learning_rate": 9.807305016807376e-05,
      "loss": 0.5609,
      "step": 3393
    },
    {
      "epoch": 0.507021213026591,
      "grad_norm": 0.7150976061820984,
      "learning_rate": 9.802605742422104e-05,
      "loss": 1.335,
      "step": 3394
    },
    {
      "epoch": 0.5071706005377951,
      "grad_norm": 0.12511125206947327,
      "learning_rate": 9.797906511644347e-05,
      "loss": 0.769,
      "step": 3395
    },
    {
      "epoch": 0.5073199880489991,
      "grad_norm": 0.1519971340894699,
      "learning_rate": 9.793207325512242e-05,
      "loss": 0.6708,
      "step": 3396
    },
    {
      "epoch": 0.5074693755602032,
      "grad_norm": 0.138926699757576,
      "learning_rate": 9.788508185063914e-05,
      "loss": 0.6253,
      "step": 3397
    },
    {
      "epoch": 0.5076187630714072,
      "grad_norm": 0.19386403262615204,
      "learning_rate": 9.78380909133748e-05,
      "loss": 0.7845,
      "step": 3398
    },
    {
      "epoch": 0.5077681505826113,
      "grad_norm": 0.15685401856899261,
      "learning_rate": 9.77911004537104e-05,
      "loss": 0.6553,
      "step": 3399
    },
    {
      "epoch": 0.5079175380938153,
      "grad_norm": 0.1295422613620758,
      "learning_rate": 9.77441104820269e-05,
      "loss": 0.6084,
      "step": 3400
    },
    {
      "epoch": 0.5080669256050194,
      "grad_norm": 0.1537609100341797,
      "learning_rate": 9.769712100870516e-05,
      "loss": 0.8397,
      "step": 3401
    },
    {
      "epoch": 0.5082163131162235,
      "grad_norm": 0.13319097459316254,
      "learning_rate": 9.765013204412583e-05,
      "loss": 0.6549,
      "step": 3402
    },
    {
      "epoch": 0.5083657006274276,
      "grad_norm": 0.21536122262477875,
      "learning_rate": 9.760314359866956e-05,
      "loss": 0.8137,
      "step": 3403
    },
    {
      "epoch": 0.5085150881386317,
      "grad_norm": 0.21349015831947327,
      "learning_rate": 9.755615568271683e-05,
      "loss": 0.7613,
      "step": 3404
    },
    {
      "epoch": 0.5086644756498356,
      "grad_norm": 0.15359088778495789,
      "learning_rate": 9.750916830664801e-05,
      "loss": 0.697,
      "step": 3405
    },
    {
      "epoch": 0.5088138631610397,
      "grad_norm": 0.23274457454681396,
      "learning_rate": 9.746218148084337e-05,
      "loss": 0.9161,
      "step": 3406
    },
    {
      "epoch": 0.5089632506722438,
      "grad_norm": 0.261106938123703,
      "learning_rate": 9.741519521568308e-05,
      "loss": 1.1296,
      "step": 3407
    },
    {
      "epoch": 0.5091126381834479,
      "grad_norm": 0.14916419982910156,
      "learning_rate": 9.736820952154706e-05,
      "loss": 0.5589,
      "step": 3408
    },
    {
      "epoch": 0.509262025694652,
      "grad_norm": 0.1515217423439026,
      "learning_rate": 9.732122440881526e-05,
      "loss": 0.8718,
      "step": 3409
    },
    {
      "epoch": 0.509411413205856,
      "grad_norm": 0.2686207890510559,
      "learning_rate": 9.72742398878674e-05,
      "loss": 0.6577,
      "step": 3410
    },
    {
      "epoch": 0.50956080071706,
      "grad_norm": 0.18468278646469116,
      "learning_rate": 9.722725596908311e-05,
      "loss": 1.1371,
      "step": 3411
    },
    {
      "epoch": 0.5097101882282641,
      "grad_norm": 0.13466569781303406,
      "learning_rate": 9.718027266284192e-05,
      "loss": 0.9602,
      "step": 3412
    },
    {
      "epoch": 0.5098595757394682,
      "grad_norm": 0.12578530609607697,
      "learning_rate": 9.713328997952315e-05,
      "loss": 0.3997,
      "step": 3413
    },
    {
      "epoch": 0.5100089632506722,
      "grad_norm": 0.20510593056678772,
      "learning_rate": 9.708630792950608e-05,
      "loss": 0.8048,
      "step": 3414
    },
    {
      "epoch": 0.5101583507618763,
      "grad_norm": 0.128560408949852,
      "learning_rate": 9.70393265231697e-05,
      "loss": 0.8289,
      "step": 3415
    },
    {
      "epoch": 0.5103077382730804,
      "grad_norm": 0.12908673286437988,
      "learning_rate": 9.699234577089297e-05,
      "loss": 0.8511,
      "step": 3416
    },
    {
      "epoch": 0.5104571257842845,
      "grad_norm": 0.15049892663955688,
      "learning_rate": 9.694536568305471e-05,
      "loss": 0.5444,
      "step": 3417
    },
    {
      "epoch": 0.5106065132954885,
      "grad_norm": 0.11548695713281631,
      "learning_rate": 9.689838627003354e-05,
      "loss": 0.5771,
      "step": 3418
    },
    {
      "epoch": 0.5107559008066925,
      "grad_norm": 0.28927114605903625,
      "learning_rate": 9.685140754220799e-05,
      "loss": 0.7775,
      "step": 3419
    },
    {
      "epoch": 0.5109052883178966,
      "grad_norm": 0.13904966413974762,
      "learning_rate": 9.68044295099564e-05,
      "loss": 0.8469,
      "step": 3420
    },
    {
      "epoch": 0.5110546758291007,
      "grad_norm": 0.13271227478981018,
      "learning_rate": 9.675745218365691e-05,
      "loss": 0.8021,
      "step": 3421
    },
    {
      "epoch": 0.5112040633403048,
      "grad_norm": 0.1266804188489914,
      "learning_rate": 9.671047557368761e-05,
      "loss": 0.8026,
      "step": 3422
    },
    {
      "epoch": 0.5113534508515089,
      "grad_norm": 0.14836454391479492,
      "learning_rate": 9.666349969042634e-05,
      "loss": 0.4041,
      "step": 3423
    },
    {
      "epoch": 0.5115028383627129,
      "grad_norm": 0.15794409811496735,
      "learning_rate": 9.661652454425086e-05,
      "loss": 0.7716,
      "step": 3424
    },
    {
      "epoch": 0.5116522258739169,
      "grad_norm": 0.12600180506706238,
      "learning_rate": 9.656955014553869e-05,
      "loss": 0.5385,
      "step": 3425
    },
    {
      "epoch": 0.511801613385121,
      "grad_norm": 0.12617719173431396,
      "learning_rate": 9.652257650466723e-05,
      "loss": 0.936,
      "step": 3426
    },
    {
      "epoch": 0.5119510008963251,
      "grad_norm": 0.14457567036151886,
      "learning_rate": 9.64756036320137e-05,
      "loss": 0.7614,
      "step": 3427
    },
    {
      "epoch": 0.5121003884075291,
      "grad_norm": 0.19280950725078583,
      "learning_rate": 9.642863153795516e-05,
      "loss": 0.8501,
      "step": 3428
    },
    {
      "epoch": 0.5122497759187332,
      "grad_norm": 0.19472643733024597,
      "learning_rate": 9.638166023286847e-05,
      "loss": 0.5745,
      "step": 3429
    },
    {
      "epoch": 0.5123991634299373,
      "grad_norm": 0.1436043083667755,
      "learning_rate": 9.633468972713034e-05,
      "loss": 0.8974,
      "step": 3430
    },
    {
      "epoch": 0.5125485509411413,
      "grad_norm": 0.31997206807136536,
      "learning_rate": 9.628772003111731e-05,
      "loss": 0.8759,
      "step": 3431
    },
    {
      "epoch": 0.5126979384523453,
      "grad_norm": 0.13290756940841675,
      "learning_rate": 9.624075115520572e-05,
      "loss": 0.5431,
      "step": 3432
    },
    {
      "epoch": 0.5128473259635494,
      "grad_norm": 0.20605328679084778,
      "learning_rate": 9.619378310977179e-05,
      "loss": 0.8757,
      "step": 3433
    },
    {
      "epoch": 0.5129967134747535,
      "grad_norm": 0.14956097304821014,
      "learning_rate": 9.614681590519143e-05,
      "loss": 0.6055,
      "step": 3434
    },
    {
      "epoch": 0.5131461009859576,
      "grad_norm": 0.14163655042648315,
      "learning_rate": 9.609984955184045e-05,
      "loss": 0.5154,
      "step": 3435
    },
    {
      "epoch": 0.5132954884971617,
      "grad_norm": 0.3266400992870331,
      "learning_rate": 9.605288406009447e-05,
      "loss": 1.1511,
      "step": 3436
    },
    {
      "epoch": 0.5134448760083657,
      "grad_norm": 0.1402955800294876,
      "learning_rate": 9.600591944032894e-05,
      "loss": 0.6846,
      "step": 3437
    },
    {
      "epoch": 0.5135942635195697,
      "grad_norm": 0.1279054433107376,
      "learning_rate": 9.595895570291906e-05,
      "loss": 0.7921,
      "step": 3438
    },
    {
      "epoch": 0.5137436510307738,
      "grad_norm": 0.27574923634529114,
      "learning_rate": 9.591199285823987e-05,
      "loss": 1.1381,
      "step": 3439
    },
    {
      "epoch": 0.5138930385419779,
      "grad_norm": 0.1963779628276825,
      "learning_rate": 9.586503091666623e-05,
      "loss": 0.9703,
      "step": 3440
    },
    {
      "epoch": 0.514042426053182,
      "grad_norm": 0.1390615552663803,
      "learning_rate": 9.581806988857275e-05,
      "loss": 0.7046,
      "step": 3441
    },
    {
      "epoch": 0.514191813564386,
      "grad_norm": 0.14069336652755737,
      "learning_rate": 9.577110978433385e-05,
      "loss": 0.4768,
      "step": 3442
    },
    {
      "epoch": 0.5143412010755901,
      "grad_norm": 0.1340341866016388,
      "learning_rate": 9.572415061432378e-05,
      "loss": 0.9329,
      "step": 3443
    },
    {
      "epoch": 0.5144905885867942,
      "grad_norm": 0.20776212215423584,
      "learning_rate": 9.567719238891658e-05,
      "loss": 0.6952,
      "step": 3444
    },
    {
      "epoch": 0.5146399760979982,
      "grad_norm": 0.19183500111103058,
      "learning_rate": 9.563023511848605e-05,
      "loss": 0.8624,
      "step": 3445
    },
    {
      "epoch": 0.5147893636092022,
      "grad_norm": 0.1788901388645172,
      "learning_rate": 9.55832788134058e-05,
      "loss": 0.8414,
      "step": 3446
    },
    {
      "epoch": 0.5149387511204063,
      "grad_norm": 0.12668971717357635,
      "learning_rate": 9.553632348404921e-05,
      "loss": 0.713,
      "step": 3447
    },
    {
      "epoch": 0.5150881386316104,
      "grad_norm": 0.14039155840873718,
      "learning_rate": 9.548936914078946e-05,
      "loss": 0.8789,
      "step": 3448
    },
    {
      "epoch": 0.5152375261428145,
      "grad_norm": 0.1316952407360077,
      "learning_rate": 9.544241579399952e-05,
      "loss": 0.5363,
      "step": 3449
    },
    {
      "epoch": 0.5153869136540186,
      "grad_norm": 0.23554357886314392,
      "learning_rate": 9.53954634540521e-05,
      "loss": 0.8788,
      "step": 3450
    },
    {
      "epoch": 0.5155363011652225,
      "grad_norm": 0.14937858283519745,
      "learning_rate": 9.534851213131974e-05,
      "loss": 0.8544,
      "step": 3451
    },
    {
      "epoch": 0.5156856886764266,
      "grad_norm": 0.18684762716293335,
      "learning_rate": 9.530156183617475e-05,
      "loss": 0.7534,
      "step": 3452
    },
    {
      "epoch": 0.5158350761876307,
      "grad_norm": 0.16386356949806213,
      "learning_rate": 9.525461257898918e-05,
      "loss": 0.6531,
      "step": 3453
    },
    {
      "epoch": 0.5159844636988348,
      "grad_norm": 0.12716203927993774,
      "learning_rate": 9.520766437013483e-05,
      "loss": 0.5557,
      "step": 3454
    },
    {
      "epoch": 0.5161338512100389,
      "grad_norm": 0.17026858031749725,
      "learning_rate": 9.516071721998333e-05,
      "loss": 0.738,
      "step": 3455
    },
    {
      "epoch": 0.5162832387212429,
      "grad_norm": 0.1371774971485138,
      "learning_rate": 9.511377113890602e-05,
      "loss": 0.7938,
      "step": 3456
    },
    {
      "epoch": 0.516432626232447,
      "grad_norm": 0.12389857321977615,
      "learning_rate": 9.506682613727408e-05,
      "loss": 0.8081,
      "step": 3457
    },
    {
      "epoch": 0.516582013743651,
      "grad_norm": 0.16528381407260895,
      "learning_rate": 9.501988222545838e-05,
      "loss": 0.6689,
      "step": 3458
    },
    {
      "epoch": 0.5167314012548551,
      "grad_norm": 0.18804830312728882,
      "learning_rate": 9.497293941382962e-05,
      "loss": 0.8123,
      "step": 3459
    },
    {
      "epoch": 0.5168807887660591,
      "grad_norm": 0.18107783794403076,
      "learning_rate": 9.492599771275813e-05,
      "loss": 0.8787,
      "step": 3460
    },
    {
      "epoch": 0.5170301762772632,
      "grad_norm": 0.12641404569149017,
      "learning_rate": 9.487905713261411e-05,
      "loss": 0.7972,
      "step": 3461
    },
    {
      "epoch": 0.5171795637884673,
      "grad_norm": 0.15088582038879395,
      "learning_rate": 9.483211768376749e-05,
      "loss": 0.5159,
      "step": 3462
    },
    {
      "epoch": 0.5173289512996714,
      "grad_norm": 0.17809826135635376,
      "learning_rate": 9.478517937658792e-05,
      "loss": 0.9898,
      "step": 3463
    },
    {
      "epoch": 0.5174783388108755,
      "grad_norm": 0.14919377863407135,
      "learning_rate": 9.473824222144483e-05,
      "loss": 0.8064,
      "step": 3464
    },
    {
      "epoch": 0.5176277263220794,
      "grad_norm": 0.12197237461805344,
      "learning_rate": 9.469130622870738e-05,
      "loss": 0.647,
      "step": 3465
    },
    {
      "epoch": 0.5177771138332835,
      "grad_norm": 0.22007903456687927,
      "learning_rate": 9.464437140874447e-05,
      "loss": 0.597,
      "step": 3466
    },
    {
      "epoch": 0.5179265013444876,
      "grad_norm": 0.15934868156909943,
      "learning_rate": 9.459743777192474e-05,
      "loss": 0.6567,
      "step": 3467
    },
    {
      "epoch": 0.5180758888556917,
      "grad_norm": 0.19033685326576233,
      "learning_rate": 9.455050532861656e-05,
      "loss": 0.9218,
      "step": 3468
    },
    {
      "epoch": 0.5182252763668957,
      "grad_norm": 0.23369066417217255,
      "learning_rate": 9.450357408918808e-05,
      "loss": 0.9239,
      "step": 3469
    },
    {
      "epoch": 0.5183746638780998,
      "grad_norm": 0.2063974291086197,
      "learning_rate": 9.445664406400716e-05,
      "loss": 0.706,
      "step": 3470
    },
    {
      "epoch": 0.5185240513893039,
      "grad_norm": 0.17821954190731049,
      "learning_rate": 9.440971526344134e-05,
      "loss": 0.7256,
      "step": 3471
    },
    {
      "epoch": 0.5186734389005079,
      "grad_norm": 0.19735683500766754,
      "learning_rate": 9.4362787697858e-05,
      "loss": 0.9632,
      "step": 3472
    },
    {
      "epoch": 0.518822826411712,
      "grad_norm": 0.17005017399787903,
      "learning_rate": 9.431586137762413e-05,
      "loss": 0.5516,
      "step": 3473
    },
    {
      "epoch": 0.518972213922916,
      "grad_norm": 0.17612014710903168,
      "learning_rate": 9.42689363131065e-05,
      "loss": 0.8091,
      "step": 3474
    },
    {
      "epoch": 0.5191216014341201,
      "grad_norm": 0.16084341704845428,
      "learning_rate": 9.422201251467162e-05,
      "loss": 0.5677,
      "step": 3475
    },
    {
      "epoch": 0.5192709889453242,
      "grad_norm": 0.19567766785621643,
      "learning_rate": 9.417508999268569e-05,
      "loss": 0.7471,
      "step": 3476
    },
    {
      "epoch": 0.5194203764565283,
      "grad_norm": 0.11310239136219025,
      "learning_rate": 9.412816875751465e-05,
      "loss": 0.5878,
      "step": 3477
    },
    {
      "epoch": 0.5195697639677322,
      "grad_norm": 0.1666627824306488,
      "learning_rate": 9.408124881952418e-05,
      "loss": 0.6466,
      "step": 3478
    },
    {
      "epoch": 0.5197191514789363,
      "grad_norm": 0.18175387382507324,
      "learning_rate": 9.403433018907954e-05,
      "loss": 0.904,
      "step": 3479
    },
    {
      "epoch": 0.5198685389901404,
      "grad_norm": 0.24765436351299286,
      "learning_rate": 9.398741287654587e-05,
      "loss": 1.2862,
      "step": 3480
    },
    {
      "epoch": 0.5200179265013445,
      "grad_norm": 0.14062359929084778,
      "learning_rate": 9.394049689228792e-05,
      "loss": 0.6961,
      "step": 3481
    },
    {
      "epoch": 0.5201673140125486,
      "grad_norm": 0.1280830204486847,
      "learning_rate": 9.389358224667019e-05,
      "loss": 0.6593,
      "step": 3482
    },
    {
      "epoch": 0.5203167015237526,
      "grad_norm": 0.12441227585077286,
      "learning_rate": 9.384666895005684e-05,
      "loss": 0.5456,
      "step": 3483
    },
    {
      "epoch": 0.5204660890349567,
      "grad_norm": 0.16274914145469666,
      "learning_rate": 9.379975701281181e-05,
      "loss": 0.6615,
      "step": 3484
    },
    {
      "epoch": 0.5206154765461607,
      "grad_norm": 0.15335504710674286,
      "learning_rate": 9.375284644529866e-05,
      "loss": 0.6249,
      "step": 3485
    },
    {
      "epoch": 0.5207648640573648,
      "grad_norm": 0.20155827701091766,
      "learning_rate": 9.370593725788068e-05,
      "loss": 0.9714,
      "step": 3486
    },
    {
      "epoch": 0.5209142515685689,
      "grad_norm": 0.14104795455932617,
      "learning_rate": 9.365902946092085e-05,
      "loss": 0.8276,
      "step": 3487
    },
    {
      "epoch": 0.5210636390797729,
      "grad_norm": 0.21122927963733673,
      "learning_rate": 9.361212306478185e-05,
      "loss": 0.8191,
      "step": 3488
    },
    {
      "epoch": 0.521213026590977,
      "grad_norm": 0.13260582089424133,
      "learning_rate": 9.356521807982602e-05,
      "loss": 0.578,
      "step": 3489
    },
    {
      "epoch": 0.5213624141021811,
      "grad_norm": 0.12246285378932953,
      "learning_rate": 9.351831451641546e-05,
      "loss": 0.7701,
      "step": 3490
    },
    {
      "epoch": 0.5215118016133852,
      "grad_norm": 0.2386673092842102,
      "learning_rate": 9.34714123849119e-05,
      "loss": 0.8315,
      "step": 3491
    },
    {
      "epoch": 0.5216611891245891,
      "grad_norm": 0.1858374923467636,
      "learning_rate": 9.342451169567675e-05,
      "loss": 0.8544,
      "step": 3492
    },
    {
      "epoch": 0.5218105766357932,
      "grad_norm": 0.16115090250968933,
      "learning_rate": 9.337761245907109e-05,
      "loss": 0.6658,
      "step": 3493
    },
    {
      "epoch": 0.5219599641469973,
      "grad_norm": 0.20373351871967316,
      "learning_rate": 9.333071468545573e-05,
      "loss": 1.0894,
      "step": 3494
    },
    {
      "epoch": 0.5221093516582014,
      "grad_norm": 0.12027011811733246,
      "learning_rate": 9.328381838519115e-05,
      "loss": 0.7684,
      "step": 3495
    },
    {
      "epoch": 0.5222587391694055,
      "grad_norm": 0.16143304109573364,
      "learning_rate": 9.323692356863746e-05,
      "loss": 0.5501,
      "step": 3496
    },
    {
      "epoch": 0.5224081266806095,
      "grad_norm": 0.1744130253791809,
      "learning_rate": 9.319003024615448e-05,
      "loss": 0.7193,
      "step": 3497
    },
    {
      "epoch": 0.5225575141918135,
      "grad_norm": 0.3834339380264282,
      "learning_rate": 9.314313842810172e-05,
      "loss": 0.9314,
      "step": 3498
    },
    {
      "epoch": 0.5227069017030176,
      "grad_norm": 0.18307146430015564,
      "learning_rate": 9.309624812483825e-05,
      "loss": 0.6509,
      "step": 3499
    },
    {
      "epoch": 0.5228562892142217,
      "grad_norm": 0.17570190131664276,
      "learning_rate": 9.30493593467229e-05,
      "loss": 0.8402,
      "step": 3500
    },
    {
      "epoch": 0.5230056767254257,
      "grad_norm": 0.11995959281921387,
      "learning_rate": 9.300247210411417e-05,
      "loss": 0.4742,
      "step": 3501
    },
    {
      "epoch": 0.5231550642366298,
      "grad_norm": 0.12027498334646225,
      "learning_rate": 9.295558640737019e-05,
      "loss": 0.4314,
      "step": 3502
    },
    {
      "epoch": 0.5233044517478339,
      "grad_norm": 0.17835885286331177,
      "learning_rate": 9.290870226684876e-05,
      "loss": 0.6171,
      "step": 3503
    },
    {
      "epoch": 0.523453839259038,
      "grad_norm": 0.1909136325120926,
      "learning_rate": 9.286181969290736e-05,
      "loss": 1.0337,
      "step": 3504
    },
    {
      "epoch": 0.523603226770242,
      "grad_norm": 0.2401609718799591,
      "learning_rate": 9.281493869590301e-05,
      "loss": 0.694,
      "step": 3505
    },
    {
      "epoch": 0.523752614281446,
      "grad_norm": 0.18573708832263947,
      "learning_rate": 9.276805928619251e-05,
      "loss": 0.6847,
      "step": 3506
    },
    {
      "epoch": 0.5239020017926501,
      "grad_norm": 0.5800057649612427,
      "learning_rate": 9.272118147413225e-05,
      "loss": 1.159,
      "step": 3507
    },
    {
      "epoch": 0.5240513893038542,
      "grad_norm": 0.1667032092809677,
      "learning_rate": 9.267430527007831e-05,
      "loss": 0.9169,
      "step": 3508
    },
    {
      "epoch": 0.5242007768150583,
      "grad_norm": 0.35562795400619507,
      "learning_rate": 9.262743068438637e-05,
      "loss": 0.6762,
      "step": 3509
    },
    {
      "epoch": 0.5243501643262624,
      "grad_norm": 0.12914936244487762,
      "learning_rate": 9.258055772741174e-05,
      "loss": 0.9183,
      "step": 3510
    },
    {
      "epoch": 0.5244995518374664,
      "grad_norm": 0.16052919626235962,
      "learning_rate": 9.253368640950946e-05,
      "loss": 0.9188,
      "step": 3511
    },
    {
      "epoch": 0.5246489393486704,
      "grad_norm": 0.20360180735588074,
      "learning_rate": 9.24868167410341e-05,
      "loss": 0.9673,
      "step": 3512
    },
    {
      "epoch": 0.5247983268598745,
      "grad_norm": 0.14888262748718262,
      "learning_rate": 9.243994873233992e-05,
      "loss": 0.6196,
      "step": 3513
    },
    {
      "epoch": 0.5249477143710786,
      "grad_norm": 0.20588485896587372,
      "learning_rate": 9.239308239378081e-05,
      "loss": 0.7824,
      "step": 3514
    },
    {
      "epoch": 0.5250971018822826,
      "grad_norm": 0.11617185920476913,
      "learning_rate": 9.23462177357103e-05,
      "loss": 0.5336,
      "step": 3515
    },
    {
      "epoch": 0.5252464893934867,
      "grad_norm": 0.16770832240581512,
      "learning_rate": 9.229935476848151e-05,
      "loss": 0.8198,
      "step": 3516
    },
    {
      "epoch": 0.5253958769046908,
      "grad_norm": 0.14460767805576324,
      "learning_rate": 9.225249350244724e-05,
      "loss": 0.8462,
      "step": 3517
    },
    {
      "epoch": 0.5255452644158949,
      "grad_norm": 0.11499165743589401,
      "learning_rate": 9.220563394795984e-05,
      "loss": 0.618,
      "step": 3518
    },
    {
      "epoch": 0.5256946519270989,
      "grad_norm": 0.19859512150287628,
      "learning_rate": 9.215877611537139e-05,
      "loss": 0.8818,
      "step": 3519
    },
    {
      "epoch": 0.5258440394383029,
      "grad_norm": 0.30726444721221924,
      "learning_rate": 9.211192001503346e-05,
      "loss": 0.976,
      "step": 3520
    },
    {
      "epoch": 0.525993426949507,
      "grad_norm": 0.15608680248260498,
      "learning_rate": 9.206506565729736e-05,
      "loss": 0.6371,
      "step": 3521
    },
    {
      "epoch": 0.5261428144607111,
      "grad_norm": 0.23737439513206482,
      "learning_rate": 9.201821305251393e-05,
      "loss": 0.9394,
      "step": 3522
    },
    {
      "epoch": 0.5262922019719152,
      "grad_norm": 0.1427346169948578,
      "learning_rate": 9.197136221103371e-05,
      "loss": 0.856,
      "step": 3523
    },
    {
      "epoch": 0.5264415894831193,
      "grad_norm": 0.1283576786518097,
      "learning_rate": 9.192451314320669e-05,
      "loss": 0.6503,
      "step": 3524
    },
    {
      "epoch": 0.5265909769943232,
      "grad_norm": 0.12818120419979095,
      "learning_rate": 9.187766585938263e-05,
      "loss": 0.8402,
      "step": 3525
    },
    {
      "epoch": 0.5267403645055273,
      "grad_norm": 0.24110840260982513,
      "learning_rate": 9.183082036991084e-05,
      "loss": 0.7696,
      "step": 3526
    },
    {
      "epoch": 0.5268897520167314,
      "grad_norm": 0.17753493785858154,
      "learning_rate": 9.178397668514023e-05,
      "loss": 0.8054,
      "step": 3527
    },
    {
      "epoch": 0.5270391395279355,
      "grad_norm": 0.12951089441776276,
      "learning_rate": 9.173713481541929e-05,
      "loss": 0.7367,
      "step": 3528
    },
    {
      "epoch": 0.5271885270391395,
      "grad_norm": 0.1222463846206665,
      "learning_rate": 9.169029477109613e-05,
      "loss": 0.5205,
      "step": 3529
    },
    {
      "epoch": 0.5273379145503436,
      "grad_norm": 0.27653244137763977,
      "learning_rate": 9.164345656251853e-05,
      "loss": 0.9928,
      "step": 3530
    },
    {
      "epoch": 0.5274873020615477,
      "grad_norm": 0.2122201919555664,
      "learning_rate": 9.159662020003371e-05,
      "loss": 0.999,
      "step": 3531
    },
    {
      "epoch": 0.5276366895727517,
      "grad_norm": 0.24955527484416962,
      "learning_rate": 9.154978569398859e-05,
      "loss": 0.6282,
      "step": 3532
    },
    {
      "epoch": 0.5277860770839558,
      "grad_norm": 0.17099502682685852,
      "learning_rate": 9.150295305472965e-05,
      "loss": 0.7037,
      "step": 3533
    },
    {
      "epoch": 0.5279354645951598,
      "grad_norm": 0.14562469720840454,
      "learning_rate": 9.145612229260295e-05,
      "loss": 0.5879,
      "step": 3534
    },
    {
      "epoch": 0.5280848521063639,
      "grad_norm": 0.2532983720302582,
      "learning_rate": 9.140929341795419e-05,
      "loss": 0.7544,
      "step": 3535
    },
    {
      "epoch": 0.528234239617568,
      "grad_norm": 0.16739071905612946,
      "learning_rate": 9.13624664411286e-05,
      "loss": 0.7832,
      "step": 3536
    },
    {
      "epoch": 0.5283836271287721,
      "grad_norm": 0.15692570805549622,
      "learning_rate": 9.131564137247094e-05,
      "loss": 0.5763,
      "step": 3537
    },
    {
      "epoch": 0.5285330146399762,
      "grad_norm": 0.1547144204378128,
      "learning_rate": 9.126881822232568e-05,
      "loss": 0.7527,
      "step": 3538
    },
    {
      "epoch": 0.5286824021511801,
      "grad_norm": 0.16494418680667877,
      "learning_rate": 9.122199700103675e-05,
      "loss": 0.8635,
      "step": 3539
    },
    {
      "epoch": 0.5288317896623842,
      "grad_norm": 0.15851503610610962,
      "learning_rate": 9.117517771894773e-05,
      "loss": 0.422,
      "step": 3540
    },
    {
      "epoch": 0.5289811771735883,
      "grad_norm": 0.11251464486122131,
      "learning_rate": 9.112836038640174e-05,
      "loss": 0.6462,
      "step": 3541
    },
    {
      "epoch": 0.5291305646847924,
      "grad_norm": 0.2646692097187042,
      "learning_rate": 9.108154501374143e-05,
      "loss": 1.2354,
      "step": 3542
    },
    {
      "epoch": 0.5292799521959964,
      "grad_norm": 0.14749304950237274,
      "learning_rate": 9.103473161130916e-05,
      "loss": 0.4317,
      "step": 3543
    },
    {
      "epoch": 0.5294293397072005,
      "grad_norm": 0.12720715999603271,
      "learning_rate": 9.098792018944661e-05,
      "loss": 0.7812,
      "step": 3544
    },
    {
      "epoch": 0.5295787272184045,
      "grad_norm": 0.10938086360692978,
      "learning_rate": 9.094111075849524e-05,
      "loss": 0.6789,
      "step": 3545
    },
    {
      "epoch": 0.5297281147296086,
      "grad_norm": 0.16336116194725037,
      "learning_rate": 9.089430332879599e-05,
      "loss": 0.7394,
      "step": 3546
    },
    {
      "epoch": 0.5298775022408126,
      "grad_norm": 0.17027264833450317,
      "learning_rate": 9.084749791068937e-05,
      "loss": 0.6733,
      "step": 3547
    },
    {
      "epoch": 0.5300268897520167,
      "grad_norm": 0.24851484596729279,
      "learning_rate": 9.080069451451544e-05,
      "loss": 1.041,
      "step": 3548
    },
    {
      "epoch": 0.5301762772632208,
      "grad_norm": 0.18901553750038147,
      "learning_rate": 9.075389315061383e-05,
      "loss": 0.7664,
      "step": 3549
    },
    {
      "epoch": 0.5303256647744249,
      "grad_norm": 0.2146105170249939,
      "learning_rate": 9.070709382932363e-05,
      "loss": 0.7107,
      "step": 3550
    },
    {
      "epoch": 0.530475052285629,
      "grad_norm": 0.2346753180027008,
      "learning_rate": 9.06602965609836e-05,
      "loss": 0.8637,
      "step": 3551
    },
    {
      "epoch": 0.5306244397968329,
      "grad_norm": 0.13710002601146698,
      "learning_rate": 9.0613501355932e-05,
      "loss": 0.5344,
      "step": 3552
    },
    {
      "epoch": 0.530773827308037,
      "grad_norm": 0.2677435576915741,
      "learning_rate": 9.056670822450661e-05,
      "loss": 0.7381,
      "step": 3553
    },
    {
      "epoch": 0.5309232148192411,
      "grad_norm": 0.21562586724758148,
      "learning_rate": 9.05199171770448e-05,
      "loss": 0.5619,
      "step": 3554
    },
    {
      "epoch": 0.5310726023304452,
      "grad_norm": 0.21366289258003235,
      "learning_rate": 9.047312822388344e-05,
      "loss": 0.6723,
      "step": 3555
    },
    {
      "epoch": 0.5312219898416493,
      "grad_norm": 0.23769651353359222,
      "learning_rate": 9.042634137535898e-05,
      "loss": 1.3054,
      "step": 3556
    },
    {
      "epoch": 0.5313713773528533,
      "grad_norm": 0.1455286294221878,
      "learning_rate": 9.037955664180731e-05,
      "loss": 0.7199,
      "step": 3557
    },
    {
      "epoch": 0.5315207648640574,
      "grad_norm": 0.12551367282867432,
      "learning_rate": 9.033277403356397e-05,
      "loss": 0.6881,
      "step": 3558
    },
    {
      "epoch": 0.5316701523752614,
      "grad_norm": 0.14695721864700317,
      "learning_rate": 9.028599356096396e-05,
      "loss": 0.7444,
      "step": 3559
    },
    {
      "epoch": 0.5318195398864655,
      "grad_norm": 0.153561532497406,
      "learning_rate": 9.023921523434184e-05,
      "loss": 0.6989,
      "step": 3560
    },
    {
      "epoch": 0.5319689273976695,
      "grad_norm": 0.2385473996400833,
      "learning_rate": 9.019243906403168e-05,
      "loss": 0.8096,
      "step": 3561
    },
    {
      "epoch": 0.5321183149088736,
      "grad_norm": 0.18886831402778625,
      "learning_rate": 9.01456650603671e-05,
      "loss": 1.0587,
      "step": 3562
    },
    {
      "epoch": 0.5322677024200777,
      "grad_norm": 0.4791150391101837,
      "learning_rate": 9.009889323368116e-05,
      "loss": 1.0353,
      "step": 3563
    },
    {
      "epoch": 0.5324170899312818,
      "grad_norm": 0.15345995128154755,
      "learning_rate": 9.005212359430654e-05,
      "loss": 0.6176,
      "step": 3564
    },
    {
      "epoch": 0.5325664774424858,
      "grad_norm": 0.2088991105556488,
      "learning_rate": 9.000535615257537e-05,
      "loss": 0.7367,
      "step": 3565
    },
    {
      "epoch": 0.5327158649536898,
      "grad_norm": 0.22386036813259125,
      "learning_rate": 8.995859091881935e-05,
      "loss": 0.8632,
      "step": 3566
    },
    {
      "epoch": 0.5328652524648939,
      "grad_norm": 0.19668370485305786,
      "learning_rate": 8.991182790336963e-05,
      "loss": 0.53,
      "step": 3567
    },
    {
      "epoch": 0.533014639976098,
      "grad_norm": 0.22110828757286072,
      "learning_rate": 8.986506711655692e-05,
      "loss": 1.0423,
      "step": 3568
    },
    {
      "epoch": 0.5331640274873021,
      "grad_norm": 0.19590145349502563,
      "learning_rate": 8.981830856871146e-05,
      "loss": 0.826,
      "step": 3569
    },
    {
      "epoch": 0.5333134149985062,
      "grad_norm": 0.2215142846107483,
      "learning_rate": 8.977155227016286e-05,
      "loss": 0.5382,
      "step": 3570
    },
    {
      "epoch": 0.5334628025097102,
      "grad_norm": 0.18073944747447968,
      "learning_rate": 8.972479823124037e-05,
      "loss": 0.8275,
      "step": 3571
    },
    {
      "epoch": 0.5336121900209142,
      "grad_norm": 0.11467383056879044,
      "learning_rate": 8.967804646227271e-05,
      "loss": 0.7464,
      "step": 3572
    },
    {
      "epoch": 0.5337615775321183,
      "grad_norm": 0.17253506183624268,
      "learning_rate": 8.963129697358808e-05,
      "loss": 0.6436,
      "step": 3573
    },
    {
      "epoch": 0.5339109650433224,
      "grad_norm": 0.5612257122993469,
      "learning_rate": 8.958454977551414e-05,
      "loss": 1.5051,
      "step": 3574
    },
    {
      "epoch": 0.5340603525545264,
      "grad_norm": 0.2774202227592468,
      "learning_rate": 8.95378048783782e-05,
      "loss": 0.8806,
      "step": 3575
    },
    {
      "epoch": 0.5342097400657305,
      "grad_norm": 0.24162757396697998,
      "learning_rate": 8.949106229250685e-05,
      "loss": 0.8341,
      "step": 3576
    },
    {
      "epoch": 0.5343591275769346,
      "grad_norm": 0.10684224963188171,
      "learning_rate": 8.944432202822626e-05,
      "loss": 0.58,
      "step": 3577
    },
    {
      "epoch": 0.5345085150881387,
      "grad_norm": 0.24626106023788452,
      "learning_rate": 8.939758409586216e-05,
      "loss": 0.8556,
      "step": 3578
    },
    {
      "epoch": 0.5346579025993426,
      "grad_norm": 0.2329065352678299,
      "learning_rate": 8.935084850573965e-05,
      "loss": 0.9332,
      "step": 3579
    },
    {
      "epoch": 0.5348072901105467,
      "grad_norm": 0.1440507024526596,
      "learning_rate": 8.930411526818337e-05,
      "loss": 0.5916,
      "step": 3580
    },
    {
      "epoch": 0.5349566776217508,
      "grad_norm": 0.3491341471672058,
      "learning_rate": 8.925738439351747e-05,
      "loss": 1.0247,
      "step": 3581
    },
    {
      "epoch": 0.5351060651329549,
      "grad_norm": 0.15624402463436127,
      "learning_rate": 8.92106558920655e-05,
      "loss": 0.8734,
      "step": 3582
    },
    {
      "epoch": 0.535255452644159,
      "grad_norm": 0.20890812575817108,
      "learning_rate": 8.916392977415053e-05,
      "loss": 0.6214,
      "step": 3583
    },
    {
      "epoch": 0.535404840155363,
      "grad_norm": 0.2275872379541397,
      "learning_rate": 8.911720605009511e-05,
      "loss": 0.6415,
      "step": 3584
    },
    {
      "epoch": 0.5355542276665671,
      "grad_norm": 0.19394375383853912,
      "learning_rate": 8.907048473022125e-05,
      "loss": 0.347,
      "step": 3585
    },
    {
      "epoch": 0.5357036151777711,
      "grad_norm": 0.4771156311035156,
      "learning_rate": 8.902376582485043e-05,
      "loss": 0.9013,
      "step": 3586
    },
    {
      "epoch": 0.5358530026889752,
      "grad_norm": 0.22501102089881897,
      "learning_rate": 8.897704934430358e-05,
      "loss": 0.5511,
      "step": 3587
    },
    {
      "epoch": 0.5360023902001793,
      "grad_norm": 0.30070528388023376,
      "learning_rate": 8.893033529890118e-05,
      "loss": 1.0615,
      "step": 3588
    },
    {
      "epoch": 0.5361517777113833,
      "grad_norm": 0.13151995837688446,
      "learning_rate": 8.888362369896301e-05,
      "loss": 0.8115,
      "step": 3589
    },
    {
      "epoch": 0.5363011652225874,
      "grad_norm": 0.1782151162624359,
      "learning_rate": 8.883691455480839e-05,
      "loss": 0.5749,
      "step": 3590
    },
    {
      "epoch": 0.5364505527337915,
      "grad_norm": 0.26052501797676086,
      "learning_rate": 8.879020787675621e-05,
      "loss": 0.9839,
      "step": 3591
    },
    {
      "epoch": 0.5365999402449955,
      "grad_norm": 0.1126941591501236,
      "learning_rate": 8.874350367512465e-05,
      "loss": 0.6061,
      "step": 3592
    },
    {
      "epoch": 0.5367493277561995,
      "grad_norm": 0.16189591586589813,
      "learning_rate": 8.869680196023142e-05,
      "loss": 0.6432,
      "step": 3593
    },
    {
      "epoch": 0.5368987152674036,
      "grad_norm": 0.19355341792106628,
      "learning_rate": 8.865010274239372e-05,
      "loss": 1.0303,
      "step": 3594
    },
    {
      "epoch": 0.5370481027786077,
      "grad_norm": 0.6486219763755798,
      "learning_rate": 8.860340603192806e-05,
      "loss": 1.2962,
      "step": 3595
    },
    {
      "epoch": 0.5371974902898118,
      "grad_norm": 0.13742980360984802,
      "learning_rate": 8.85567118391505e-05,
      "loss": 0.7937,
      "step": 3596
    },
    {
      "epoch": 0.5373468778010159,
      "grad_norm": 0.16776223480701447,
      "learning_rate": 8.851002017437656e-05,
      "loss": 0.4064,
      "step": 3597
    },
    {
      "epoch": 0.53749626531222,
      "grad_norm": 0.2089015543460846,
      "learning_rate": 8.846333104792116e-05,
      "loss": 0.5693,
      "step": 3598
    },
    {
      "epoch": 0.5376456528234239,
      "grad_norm": 0.2664184868335724,
      "learning_rate": 8.841664447009865e-05,
      "loss": 1.0731,
      "step": 3599
    },
    {
      "epoch": 0.537795040334628,
      "grad_norm": 0.13741925358772278,
      "learning_rate": 8.836996045122286e-05,
      "loss": 0.8823,
      "step": 3600
    },
    {
      "epoch": 0.5379444278458321,
      "grad_norm": 0.3124346137046814,
      "learning_rate": 8.832327900160703e-05,
      "loss": 1.0104,
      "step": 3601
    },
    {
      "epoch": 0.5380938153570362,
      "grad_norm": 0.19726713001728058,
      "learning_rate": 8.827660013156381e-05,
      "loss": 1.2038,
      "step": 3602
    },
    {
      "epoch": 0.5382432028682402,
      "grad_norm": 0.2321999967098236,
      "learning_rate": 8.822992385140529e-05,
      "loss": 0.9607,
      "step": 3603
    },
    {
      "epoch": 0.5383925903794443,
      "grad_norm": 0.15645650029182434,
      "learning_rate": 8.818325017144302e-05,
      "loss": 0.8551,
      "step": 3604
    },
    {
      "epoch": 0.5385419778906484,
      "grad_norm": 0.18321920931339264,
      "learning_rate": 8.813657910198797e-05,
      "loss": 0.6633,
      "step": 3605
    },
    {
      "epoch": 0.5386913654018524,
      "grad_norm": 0.11348626762628555,
      "learning_rate": 8.808991065335049e-05,
      "loss": 0.6334,
      "step": 3606
    },
    {
      "epoch": 0.5388407529130564,
      "grad_norm": 0.16270765662193298,
      "learning_rate": 8.804324483584043e-05,
      "loss": 0.7803,
      "step": 3607
    },
    {
      "epoch": 0.5389901404242605,
      "grad_norm": 0.2062917798757553,
      "learning_rate": 8.799658165976694e-05,
      "loss": 0.8379,
      "step": 3608
    },
    {
      "epoch": 0.5391395279354646,
      "grad_norm": 0.16289086639881134,
      "learning_rate": 8.794992113543868e-05,
      "loss": 0.5756,
      "step": 3609
    },
    {
      "epoch": 0.5392889154466687,
      "grad_norm": 0.17799049615859985,
      "learning_rate": 8.790326327316372e-05,
      "loss": 0.5123,
      "step": 3610
    },
    {
      "epoch": 0.5394383029578728,
      "grad_norm": 0.19015629589557648,
      "learning_rate": 8.785660808324949e-05,
      "loss": 0.8316,
      "step": 3611
    },
    {
      "epoch": 0.5395876904690767,
      "grad_norm": 0.12746602296829224,
      "learning_rate": 8.780995557600287e-05,
      "loss": 0.7487,
      "step": 3612
    },
    {
      "epoch": 0.5397370779802808,
      "grad_norm": 0.14309895038604736,
      "learning_rate": 8.776330576173016e-05,
      "loss": 0.5486,
      "step": 3613
    },
    {
      "epoch": 0.5398864654914849,
      "grad_norm": 0.1327267587184906,
      "learning_rate": 8.771665865073707e-05,
      "loss": 0.8603,
      "step": 3614
    },
    {
      "epoch": 0.540035853002689,
      "grad_norm": 0.1330852210521698,
      "learning_rate": 8.76700142533286e-05,
      "loss": 0.8199,
      "step": 3615
    },
    {
      "epoch": 0.540185240513893,
      "grad_norm": 0.1176176369190216,
      "learning_rate": 8.762337257980927e-05,
      "loss": 0.7801,
      "step": 3616
    },
    {
      "epoch": 0.5403346280250971,
      "grad_norm": 0.16695699095726013,
      "learning_rate": 8.757673364048298e-05,
      "loss": 0.5073,
      "step": 3617
    },
    {
      "epoch": 0.5404840155363012,
      "grad_norm": 0.17452150583267212,
      "learning_rate": 8.753009744565297e-05,
      "loss": 0.8845,
      "step": 3618
    },
    {
      "epoch": 0.5406334030475052,
      "grad_norm": 0.4428778886795044,
      "learning_rate": 8.748346400562197e-05,
      "loss": 1.1007,
      "step": 3619
    },
    {
      "epoch": 0.5407827905587093,
      "grad_norm": 0.1329747885465622,
      "learning_rate": 8.743683333069208e-05,
      "loss": 0.7978,
      "step": 3620
    },
    {
      "epoch": 0.5409321780699133,
      "grad_norm": 0.15354689955711365,
      "learning_rate": 8.739020543116465e-05,
      "loss": 0.9022,
      "step": 3621
    },
    {
      "epoch": 0.5410815655811174,
      "grad_norm": 0.12778078019618988,
      "learning_rate": 8.734358031734056e-05,
      "loss": 0.8688,
      "step": 3622
    },
    {
      "epoch": 0.5412309530923215,
      "grad_norm": 0.14656540751457214,
      "learning_rate": 8.729695799952004e-05,
      "loss": 0.8617,
      "step": 3623
    },
    {
      "epoch": 0.5413803406035256,
      "grad_norm": 0.12861032783985138,
      "learning_rate": 8.725033848800273e-05,
      "loss": 0.745,
      "step": 3624
    },
    {
      "epoch": 0.5415297281147297,
      "grad_norm": 0.14260074496269226,
      "learning_rate": 8.720372179308758e-05,
      "loss": 1.0235,
      "step": 3625
    },
    {
      "epoch": 0.5416791156259336,
      "grad_norm": 0.13216426968574524,
      "learning_rate": 8.715710792507295e-05,
      "loss": 0.4453,
      "step": 3626
    },
    {
      "epoch": 0.5418285031371377,
      "grad_norm": 0.211138516664505,
      "learning_rate": 8.711049689425664e-05,
      "loss": 0.9692,
      "step": 3627
    },
    {
      "epoch": 0.5419778906483418,
      "grad_norm": 0.15708252787590027,
      "learning_rate": 8.706388871093571e-05,
      "loss": 0.6211,
      "step": 3628
    },
    {
      "epoch": 0.5421272781595459,
      "grad_norm": 0.21318966150283813,
      "learning_rate": 8.701728338540665e-05,
      "loss": 0.6928,
      "step": 3629
    },
    {
      "epoch": 0.54227666567075,
      "grad_norm": 0.12893186509609222,
      "learning_rate": 8.697068092796531e-05,
      "loss": 0.5829,
      "step": 3630
    },
    {
      "epoch": 0.542426053181954,
      "grad_norm": 0.3459821045398712,
      "learning_rate": 8.692408134890696e-05,
      "loss": 1.0125,
      "step": 3631
    },
    {
      "epoch": 0.5425754406931581,
      "grad_norm": 0.20835642516613007,
      "learning_rate": 8.687748465852614e-05,
      "loss": 0.6886,
      "step": 3632
    },
    {
      "epoch": 0.5427248282043621,
      "grad_norm": 0.10729233920574188,
      "learning_rate": 8.683089086711685e-05,
      "loss": 0.6428,
      "step": 3633
    },
    {
      "epoch": 0.5428742157155662,
      "grad_norm": 0.1466486006975174,
      "learning_rate": 8.678429998497229e-05,
      "loss": 0.6247,
      "step": 3634
    },
    {
      "epoch": 0.5430236032267702,
      "grad_norm": 0.23065689206123352,
      "learning_rate": 8.67377120223852e-05,
      "loss": 0.9108,
      "step": 3635
    },
    {
      "epoch": 0.5431729907379743,
      "grad_norm": 0.3025874197483063,
      "learning_rate": 8.66911269896476e-05,
      "loss": 1.0507,
      "step": 3636
    },
    {
      "epoch": 0.5433223782491784,
      "grad_norm": 0.3378618359565735,
      "learning_rate": 8.664454489705085e-05,
      "loss": 1.0229,
      "step": 3637
    },
    {
      "epoch": 0.5434717657603825,
      "grad_norm": 0.12850575149059296,
      "learning_rate": 8.659796575488566e-05,
      "loss": 0.6146,
      "step": 3638
    },
    {
      "epoch": 0.5436211532715864,
      "grad_norm": 0.12255378812551498,
      "learning_rate": 8.655138957344211e-05,
      "loss": 0.6055,
      "step": 3639
    },
    {
      "epoch": 0.5437705407827905,
      "grad_norm": 0.14635306596755981,
      "learning_rate": 8.650481636300969e-05,
      "loss": 0.6397,
      "step": 3640
    },
    {
      "epoch": 0.5439199282939946,
      "grad_norm": 0.13539202511310577,
      "learning_rate": 8.645824613387702e-05,
      "loss": 0.656,
      "step": 3641
    },
    {
      "epoch": 0.5440693158051987,
      "grad_norm": 0.11104149371385574,
      "learning_rate": 8.641167889633228e-05,
      "loss": 0.6469,
      "step": 3642
    },
    {
      "epoch": 0.5442187033164028,
      "grad_norm": 0.17433156073093414,
      "learning_rate": 8.63651146606629e-05,
      "loss": 0.6093,
      "step": 3643
    },
    {
      "epoch": 0.5443680908276068,
      "grad_norm": 0.1628594696521759,
      "learning_rate": 8.631855343715565e-05,
      "loss": 0.98,
      "step": 3644
    },
    {
      "epoch": 0.5445174783388109,
      "grad_norm": 0.3815864026546478,
      "learning_rate": 8.627199523609665e-05,
      "loss": 0.9103,
      "step": 3645
    },
    {
      "epoch": 0.5446668658500149,
      "grad_norm": 0.23896418511867523,
      "learning_rate": 8.622544006777136e-05,
      "loss": 0.6625,
      "step": 3646
    },
    {
      "epoch": 0.544816253361219,
      "grad_norm": 0.1372939795255661,
      "learning_rate": 8.617888794246452e-05,
      "loss": 0.6004,
      "step": 3647
    },
    {
      "epoch": 0.544965640872423,
      "grad_norm": 0.2376316338777542,
      "learning_rate": 8.613233887046027e-05,
      "loss": 0.784,
      "step": 3648
    },
    {
      "epoch": 0.5451150283836271,
      "grad_norm": 0.5201293230056763,
      "learning_rate": 8.608579286204198e-05,
      "loss": 1.0979,
      "step": 3649
    },
    {
      "epoch": 0.5452644158948312,
      "grad_norm": 0.21554164588451385,
      "learning_rate": 8.603924992749245e-05,
      "loss": 0.877,
      "step": 3650
    },
    {
      "epoch": 0.5454138034060353,
      "grad_norm": 0.16923366487026215,
      "learning_rate": 8.599271007709374e-05,
      "loss": 0.7741,
      "step": 3651
    },
    {
      "epoch": 0.5455631909172394,
      "grad_norm": 0.11953236162662506,
      "learning_rate": 8.594617332112725e-05,
      "loss": 0.5996,
      "step": 3652
    },
    {
      "epoch": 0.5457125784284433,
      "grad_norm": 0.16639801859855652,
      "learning_rate": 8.589963966987368e-05,
      "loss": 0.9027,
      "step": 3653
    },
    {
      "epoch": 0.5458619659396474,
      "grad_norm": 0.3524928390979767,
      "learning_rate": 8.585310913361301e-05,
      "loss": 1.1005,
      "step": 3654
    },
    {
      "epoch": 0.5460113534508515,
      "grad_norm": 0.13823704421520233,
      "learning_rate": 8.580658172262463e-05,
      "loss": 0.8559,
      "step": 3655
    },
    {
      "epoch": 0.5461607409620556,
      "grad_norm": 0.24794632196426392,
      "learning_rate": 8.576005744718716e-05,
      "loss": 1.0205,
      "step": 3656
    },
    {
      "epoch": 0.5463101284732597,
      "grad_norm": 0.1253996342420578,
      "learning_rate": 8.571353631757855e-05,
      "loss": 0.8237,
      "step": 3657
    },
    {
      "epoch": 0.5464595159844637,
      "grad_norm": 0.14612741768360138,
      "learning_rate": 8.566701834407605e-05,
      "loss": 0.5167,
      "step": 3658
    },
    {
      "epoch": 0.5466089034956677,
      "grad_norm": 0.20115165412425995,
      "learning_rate": 8.562050353695625e-05,
      "loss": 0.6148,
      "step": 3659
    },
    {
      "epoch": 0.5467582910068718,
      "grad_norm": 0.1406886726617813,
      "learning_rate": 8.557399190649496e-05,
      "loss": 0.6425,
      "step": 3660
    },
    {
      "epoch": 0.5469076785180759,
      "grad_norm": 0.12534284591674805,
      "learning_rate": 8.552748346296733e-05,
      "loss": 0.7016,
      "step": 3661
    },
    {
      "epoch": 0.54705706602928,
      "grad_norm": 0.19271613657474518,
      "learning_rate": 8.548097821664785e-05,
      "loss": 0.5789,
      "step": 3662
    },
    {
      "epoch": 0.547206453540484,
      "grad_norm": 0.1629400998353958,
      "learning_rate": 8.543447617781022e-05,
      "loss": 0.663,
      "step": 3663
    },
    {
      "epoch": 0.5473558410516881,
      "grad_norm": 0.11532813310623169,
      "learning_rate": 8.538797735672753e-05,
      "loss": 0.4221,
      "step": 3664
    },
    {
      "epoch": 0.5475052285628922,
      "grad_norm": 0.5268710851669312,
      "learning_rate": 8.534148176367214e-05,
      "loss": 1.3714,
      "step": 3665
    },
    {
      "epoch": 0.5476546160740962,
      "grad_norm": 0.1772603690624237,
      "learning_rate": 8.529498940891554e-05,
      "loss": 0.6513,
      "step": 3666
    },
    {
      "epoch": 0.5478040035853002,
      "grad_norm": 0.12787871062755585,
      "learning_rate": 8.524850030272872e-05,
      "loss": 0.888,
      "step": 3667
    },
    {
      "epoch": 0.5479533910965043,
      "grad_norm": 0.36109769344329834,
      "learning_rate": 8.520201445538183e-05,
      "loss": 0.9002,
      "step": 3668
    },
    {
      "epoch": 0.5481027786077084,
      "grad_norm": 0.43515944480895996,
      "learning_rate": 8.515553187714433e-05,
      "loss": 0.8664,
      "step": 3669
    },
    {
      "epoch": 0.5482521661189125,
      "grad_norm": 0.15171855688095093,
      "learning_rate": 8.510905257828496e-05,
      "loss": 0.9539,
      "step": 3670
    },
    {
      "epoch": 0.5484015536301166,
      "grad_norm": 0.1639348864555359,
      "learning_rate": 8.506257656907175e-05,
      "loss": 0.4373,
      "step": 3671
    },
    {
      "epoch": 0.5485509411413206,
      "grad_norm": 0.12094151973724365,
      "learning_rate": 8.501610385977198e-05,
      "loss": 0.4464,
      "step": 3672
    },
    {
      "epoch": 0.5487003286525246,
      "grad_norm": 0.1904463917016983,
      "learning_rate": 8.496963446065218e-05,
      "loss": 1.0771,
      "step": 3673
    },
    {
      "epoch": 0.5488497161637287,
      "grad_norm": 0.1426345705986023,
      "learning_rate": 8.49231683819782e-05,
      "loss": 0.8451,
      "step": 3674
    },
    {
      "epoch": 0.5489991036749328,
      "grad_norm": 0.1682104766368866,
      "learning_rate": 8.487670563401513e-05,
      "loss": 0.9096,
      "step": 3675
    },
    {
      "epoch": 0.5491484911861368,
      "grad_norm": 0.1341891884803772,
      "learning_rate": 8.483024622702732e-05,
      "loss": 0.7758,
      "step": 3676
    },
    {
      "epoch": 0.5492978786973409,
      "grad_norm": 0.15437579154968262,
      "learning_rate": 8.47837901712784e-05,
      "loss": 0.7638,
      "step": 3677
    },
    {
      "epoch": 0.549447266208545,
      "grad_norm": 0.13084208965301514,
      "learning_rate": 8.473733747703129e-05,
      "loss": 0.8906,
      "step": 3678
    },
    {
      "epoch": 0.5495966537197491,
      "grad_norm": 0.11895976215600967,
      "learning_rate": 8.4690888154548e-05,
      "loss": 0.4241,
      "step": 3679
    },
    {
      "epoch": 0.549746041230953,
      "grad_norm": 0.15303203463554382,
      "learning_rate": 8.464444221409004e-05,
      "loss": 0.8025,
      "step": 3680
    },
    {
      "epoch": 0.5498954287421571,
      "grad_norm": 0.16749686002731323,
      "learning_rate": 8.4597999665918e-05,
      "loss": 0.5818,
      "step": 3681
    },
    {
      "epoch": 0.5500448162533612,
      "grad_norm": 0.1736450046300888,
      "learning_rate": 8.45515605202918e-05,
      "loss": 0.5353,
      "step": 3682
    },
    {
      "epoch": 0.5501942037645653,
      "grad_norm": 0.11968281865119934,
      "learning_rate": 8.450512478747056e-05,
      "loss": 0.6829,
      "step": 3683
    },
    {
      "epoch": 0.5503435912757694,
      "grad_norm": 0.3091181516647339,
      "learning_rate": 8.44586924777127e-05,
      "loss": 1.0903,
      "step": 3684
    },
    {
      "epoch": 0.5504929787869735,
      "grad_norm": 0.10731890052556992,
      "learning_rate": 8.441226360127585e-05,
      "loss": 0.4676,
      "step": 3685
    },
    {
      "epoch": 0.5506423662981774,
      "grad_norm": 0.2691383361816406,
      "learning_rate": 8.436583816841684e-05,
      "loss": 1.0552,
      "step": 3686
    },
    {
      "epoch": 0.5507917538093815,
      "grad_norm": 0.14232249557971954,
      "learning_rate": 8.43194161893918e-05,
      "loss": 0.5211,
      "step": 3687
    },
    {
      "epoch": 0.5509411413205856,
      "grad_norm": 0.22509650886058807,
      "learning_rate": 8.42729976744561e-05,
      "loss": 0.713,
      "step": 3688
    },
    {
      "epoch": 0.5510905288317897,
      "grad_norm": 0.12247148156166077,
      "learning_rate": 8.422658263386433e-05,
      "loss": 0.6514,
      "step": 3689
    },
    {
      "epoch": 0.5512399163429937,
      "grad_norm": 0.11414561420679092,
      "learning_rate": 8.418017107787028e-05,
      "loss": 0.5761,
      "step": 3690
    },
    {
      "epoch": 0.5513893038541978,
      "grad_norm": 0.14869552850723267,
      "learning_rate": 8.413376301672705e-05,
      "loss": 0.7923,
      "step": 3691
    },
    {
      "epoch": 0.5515386913654019,
      "grad_norm": 0.3078554570674896,
      "learning_rate": 8.408735846068683e-05,
      "loss": 1.3196,
      "step": 3692
    },
    {
      "epoch": 0.5516880788766059,
      "grad_norm": 0.1308305412530899,
      "learning_rate": 8.404095742000118e-05,
      "loss": 0.734,
      "step": 3693
    },
    {
      "epoch": 0.55183746638781,
      "grad_norm": 0.3791923224925995,
      "learning_rate": 8.399455990492082e-05,
      "loss": 0.9965,
      "step": 3694
    },
    {
      "epoch": 0.551986853899014,
      "grad_norm": 0.164544016122818,
      "learning_rate": 8.394816592569567e-05,
      "loss": 0.7069,
      "step": 3695
    },
    {
      "epoch": 0.5521362414102181,
      "grad_norm": 0.13011978566646576,
      "learning_rate": 8.390177549257494e-05,
      "loss": 0.4846,
      "step": 3696
    },
    {
      "epoch": 0.5522856289214222,
      "grad_norm": 0.1407751441001892,
      "learning_rate": 8.385538861580698e-05,
      "loss": 0.4587,
      "step": 3697
    },
    {
      "epoch": 0.5524350164326263,
      "grad_norm": 0.11466528475284576,
      "learning_rate": 8.38090053056394e-05,
      "loss": 0.5049,
      "step": 3698
    },
    {
      "epoch": 0.5525844039438303,
      "grad_norm": 0.2135232388973236,
      "learning_rate": 8.376262557231899e-05,
      "loss": 0.9208,
      "step": 3699
    },
    {
      "epoch": 0.5527337914550343,
      "grad_norm": 0.2697911262512207,
      "learning_rate": 8.371624942609177e-05,
      "loss": 0.9598,
      "step": 3700
    },
    {
      "epoch": 0.5528831789662384,
      "grad_norm": 0.12328141927719116,
      "learning_rate": 8.366987687720298e-05,
      "loss": 0.8135,
      "step": 3701
    },
    {
      "epoch": 0.5530325664774425,
      "grad_norm": 0.15307845175266266,
      "learning_rate": 8.362350793589705e-05,
      "loss": 1.1103,
      "step": 3702
    },
    {
      "epoch": 0.5531819539886466,
      "grad_norm": 0.12340089678764343,
      "learning_rate": 8.35771426124176e-05,
      "loss": 0.7008,
      "step": 3703
    },
    {
      "epoch": 0.5533313414998506,
      "grad_norm": 0.15964895486831665,
      "learning_rate": 8.353078091700751e-05,
      "loss": 0.575,
      "step": 3704
    },
    {
      "epoch": 0.5534807290110547,
      "grad_norm": 0.16260680556297302,
      "learning_rate": 8.348442285990874e-05,
      "loss": 0.3583,
      "step": 3705
    },
    {
      "epoch": 0.5536301165222587,
      "grad_norm": 0.3263389766216278,
      "learning_rate": 8.343806845136255e-05,
      "loss": 1.032,
      "step": 3706
    },
    {
      "epoch": 0.5537795040334628,
      "grad_norm": 0.3308280408382416,
      "learning_rate": 8.339171770160934e-05,
      "loss": 1.0473,
      "step": 3707
    },
    {
      "epoch": 0.5539288915446668,
      "grad_norm": 0.14974111318588257,
      "learning_rate": 8.334537062088878e-05,
      "loss": 0.4333,
      "step": 3708
    },
    {
      "epoch": 0.5540782790558709,
      "grad_norm": 0.184838205575943,
      "learning_rate": 8.329902721943964e-05,
      "loss": 0.4349,
      "step": 3709
    },
    {
      "epoch": 0.554227666567075,
      "grad_norm": 0.1789599359035492,
      "learning_rate": 8.325268750749991e-05,
      "loss": 0.5858,
      "step": 3710
    },
    {
      "epoch": 0.5543770540782791,
      "grad_norm": 0.1843758225440979,
      "learning_rate": 8.320635149530684e-05,
      "loss": 0.631,
      "step": 3711
    },
    {
      "epoch": 0.5545264415894832,
      "grad_norm": 0.21946020424365997,
      "learning_rate": 8.316001919309667e-05,
      "loss": 0.7438,
      "step": 3712
    },
    {
      "epoch": 0.5546758291006871,
      "grad_norm": 0.17891810834407806,
      "learning_rate": 8.311369061110498e-05,
      "loss": 0.4694,
      "step": 3713
    },
    {
      "epoch": 0.5548252166118912,
      "grad_norm": 0.14040721952915192,
      "learning_rate": 8.306736575956651e-05,
      "loss": 0.33,
      "step": 3714
    },
    {
      "epoch": 0.5549746041230953,
      "grad_norm": 0.4606151878833771,
      "learning_rate": 8.302104464871513e-05,
      "loss": 1.1912,
      "step": 3715
    },
    {
      "epoch": 0.5551239916342994,
      "grad_norm": 0.1793205738067627,
      "learning_rate": 8.297472728878392e-05,
      "loss": 0.5323,
      "step": 3716
    },
    {
      "epoch": 0.5552733791455035,
      "grad_norm": 0.10938454419374466,
      "learning_rate": 8.292841369000516e-05,
      "loss": 0.6334,
      "step": 3717
    },
    {
      "epoch": 0.5554227666567075,
      "grad_norm": 0.6650310158729553,
      "learning_rate": 8.288210386261019e-05,
      "loss": 1.0934,
      "step": 3718
    },
    {
      "epoch": 0.5555721541679116,
      "grad_norm": 0.3494865298271179,
      "learning_rate": 8.28357978168296e-05,
      "loss": 0.7037,
      "step": 3719
    },
    {
      "epoch": 0.5557215416791156,
      "grad_norm": 0.17872025072574615,
      "learning_rate": 8.278949556289314e-05,
      "loss": 0.7943,
      "step": 3720
    },
    {
      "epoch": 0.5558709291903197,
      "grad_norm": 0.11630931496620178,
      "learning_rate": 8.274319711102972e-05,
      "loss": 0.6952,
      "step": 3721
    },
    {
      "epoch": 0.5560203167015237,
      "grad_norm": 0.15169143676757812,
      "learning_rate": 8.269690247146737e-05,
      "loss": 0.4521,
      "step": 3722
    },
    {
      "epoch": 0.5561697042127278,
      "grad_norm": 0.1566866636276245,
      "learning_rate": 8.265061165443338e-05,
      "loss": 0.9718,
      "step": 3723
    },
    {
      "epoch": 0.5563190917239319,
      "grad_norm": 0.16685472428798676,
      "learning_rate": 8.260432467015403e-05,
      "loss": 0.7881,
      "step": 3724
    },
    {
      "epoch": 0.556468479235136,
      "grad_norm": 0.1447632908821106,
      "learning_rate": 8.255804152885492e-05,
      "loss": 0.5852,
      "step": 3725
    },
    {
      "epoch": 0.55661786674634,
      "grad_norm": 0.1420188546180725,
      "learning_rate": 8.251176224076067e-05,
      "loss": 0.9457,
      "step": 3726
    },
    {
      "epoch": 0.556767254257544,
      "grad_norm": 0.1290731579065323,
      "learning_rate": 8.246548681609516e-05,
      "loss": 0.6357,
      "step": 3727
    },
    {
      "epoch": 0.5569166417687481,
      "grad_norm": 0.1289801448583603,
      "learning_rate": 8.241921526508135e-05,
      "loss": 0.7124,
      "step": 3728
    },
    {
      "epoch": 0.5570660292799522,
      "grad_norm": 0.18616311252117157,
      "learning_rate": 8.237294759794135e-05,
      "loss": 0.7453,
      "step": 3729
    },
    {
      "epoch": 0.5572154167911563,
      "grad_norm": 0.13330665230751038,
      "learning_rate": 8.232668382489646e-05,
      "loss": 0.7464,
      "step": 3730
    },
    {
      "epoch": 0.5573648043023604,
      "grad_norm": 0.11477735638618469,
      "learning_rate": 8.2280423956167e-05,
      "loss": 0.6108,
      "step": 3731
    },
    {
      "epoch": 0.5575141918135644,
      "grad_norm": 0.19604364037513733,
      "learning_rate": 8.223416800197256e-05,
      "loss": 0.6504,
      "step": 3732
    },
    {
      "epoch": 0.5576635793247684,
      "grad_norm": 0.16029389202594757,
      "learning_rate": 8.21879159725318e-05,
      "loss": 0.5263,
      "step": 3733
    },
    {
      "epoch": 0.5578129668359725,
      "grad_norm": 0.15514934062957764,
      "learning_rate": 8.214166787806252e-05,
      "loss": 0.6435,
      "step": 3734
    },
    {
      "epoch": 0.5579623543471766,
      "grad_norm": 0.18050578236579895,
      "learning_rate": 8.209542372878168e-05,
      "loss": 1.011,
      "step": 3735
    },
    {
      "epoch": 0.5581117418583806,
      "grad_norm": 0.15090398490428925,
      "learning_rate": 8.204918353490535e-05,
      "loss": 0.5785,
      "step": 3736
    },
    {
      "epoch": 0.5582611293695847,
      "grad_norm": 0.13529077172279358,
      "learning_rate": 8.200294730664868e-05,
      "loss": 0.8035,
      "step": 3737
    },
    {
      "epoch": 0.5584105168807888,
      "grad_norm": 0.1745627075433731,
      "learning_rate": 8.195671505422602e-05,
      "loss": 0.8333,
      "step": 3738
    },
    {
      "epoch": 0.5585599043919929,
      "grad_norm": 0.18329249322414398,
      "learning_rate": 8.191048678785079e-05,
      "loss": 0.9066,
      "step": 3739
    },
    {
      "epoch": 0.5587092919031968,
      "grad_norm": 0.2413608878850937,
      "learning_rate": 8.186426251773554e-05,
      "loss": 0.6514,
      "step": 3740
    },
    {
      "epoch": 0.5588586794144009,
      "grad_norm": 0.18537455797195435,
      "learning_rate": 8.181804225409197e-05,
      "loss": 0.6992,
      "step": 3741
    },
    {
      "epoch": 0.559008066925605,
      "grad_norm": 0.12640683352947235,
      "learning_rate": 8.177182600713084e-05,
      "loss": 0.6906,
      "step": 3742
    },
    {
      "epoch": 0.5591574544368091,
      "grad_norm": 0.3122878670692444,
      "learning_rate": 8.172561378706209e-05,
      "loss": 1.0028,
      "step": 3743
    },
    {
      "epoch": 0.5593068419480132,
      "grad_norm": 0.16115644574165344,
      "learning_rate": 8.167940560409469e-05,
      "loss": 0.4202,
      "step": 3744
    },
    {
      "epoch": 0.5594562294592172,
      "grad_norm": 0.1910959929227829,
      "learning_rate": 8.163320146843676e-05,
      "loss": 0.6292,
      "step": 3745
    },
    {
      "epoch": 0.5596056169704213,
      "grad_norm": 0.28552910685539246,
      "learning_rate": 8.158700139029557e-05,
      "loss": 0.8864,
      "step": 3746
    },
    {
      "epoch": 0.5597550044816253,
      "grad_norm": 0.1834760308265686,
      "learning_rate": 8.154080537987741e-05,
      "loss": 0.6172,
      "step": 3747
    },
    {
      "epoch": 0.5599043919928294,
      "grad_norm": 0.19498953223228455,
      "learning_rate": 8.14946134473877e-05,
      "loss": 0.7101,
      "step": 3748
    },
    {
      "epoch": 0.5600537795040335,
      "grad_norm": 0.2991584539413452,
      "learning_rate": 8.144842560303106e-05,
      "loss": 1.2332,
      "step": 3749
    },
    {
      "epoch": 0.5602031670152375,
      "grad_norm": 0.1680474579334259,
      "learning_rate": 8.140224185701097e-05,
      "loss": 0.4453,
      "step": 3750
    },
    {
      "epoch": 0.5603525545264416,
      "grad_norm": 0.15305602550506592,
      "learning_rate": 8.135606221953025e-05,
      "loss": 0.8243,
      "step": 3751
    },
    {
      "epoch": 0.5605019420376457,
      "grad_norm": 0.2010365128517151,
      "learning_rate": 8.130988670079068e-05,
      "loss": 0.649,
      "step": 3752
    },
    {
      "epoch": 0.5606513295488497,
      "grad_norm": 0.5384266376495361,
      "learning_rate": 8.126371531099317e-05,
      "loss": 0.99,
      "step": 3753
    },
    {
      "epoch": 0.5608007170600537,
      "grad_norm": 0.14072974026203156,
      "learning_rate": 8.121754806033772e-05,
      "loss": 0.6765,
      "step": 3754
    },
    {
      "epoch": 0.5609501045712578,
      "grad_norm": 0.20828408002853394,
      "learning_rate": 8.117138495902341e-05,
      "loss": 0.6038,
      "step": 3755
    },
    {
      "epoch": 0.5610994920824619,
      "grad_norm": 0.12348310649394989,
      "learning_rate": 8.112522601724844e-05,
      "loss": 0.5557,
      "step": 3756
    },
    {
      "epoch": 0.561248879593666,
      "grad_norm": 0.14315848052501678,
      "learning_rate": 8.107907124520994e-05,
      "loss": 0.5327,
      "step": 3757
    },
    {
      "epoch": 0.5613982671048701,
      "grad_norm": 0.22999818623065948,
      "learning_rate": 8.103292065310431e-05,
      "loss": 1.1618,
      "step": 3758
    },
    {
      "epoch": 0.5615476546160741,
      "grad_norm": 0.12501800060272217,
      "learning_rate": 8.098677425112693e-05,
      "loss": 0.8068,
      "step": 3759
    },
    {
      "epoch": 0.5616970421272781,
      "grad_norm": 0.2536140978336334,
      "learning_rate": 8.094063204947227e-05,
      "loss": 1.0101,
      "step": 3760
    },
    {
      "epoch": 0.5618464296384822,
      "grad_norm": 0.19433334469795227,
      "learning_rate": 8.089449405833386e-05,
      "loss": 0.7679,
      "step": 3761
    },
    {
      "epoch": 0.5619958171496863,
      "grad_norm": 0.10527056455612183,
      "learning_rate": 8.084836028790438e-05,
      "loss": 0.4031,
      "step": 3762
    },
    {
      "epoch": 0.5621452046608904,
      "grad_norm": 0.2064230889081955,
      "learning_rate": 8.080223074837542e-05,
      "loss": 0.7125,
      "step": 3763
    },
    {
      "epoch": 0.5622945921720944,
      "grad_norm": 0.13225442171096802,
      "learning_rate": 8.075610544993777e-05,
      "loss": 0.7456,
      "step": 3764
    },
    {
      "epoch": 0.5624439796832985,
      "grad_norm": 0.13613517582416534,
      "learning_rate": 8.070998440278123e-05,
      "loss": 0.7818,
      "step": 3765
    },
    {
      "epoch": 0.5625933671945026,
      "grad_norm": 0.28301966190338135,
      "learning_rate": 8.066386761709467e-05,
      "loss": 0.7587,
      "step": 3766
    },
    {
      "epoch": 0.5627427547057066,
      "grad_norm": 0.20743635296821594,
      "learning_rate": 8.061775510306603e-05,
      "loss": 0.8077,
      "step": 3767
    },
    {
      "epoch": 0.5628921422169106,
      "grad_norm": 0.13272540271282196,
      "learning_rate": 8.057164687088228e-05,
      "loss": 0.7424,
      "step": 3768
    },
    {
      "epoch": 0.5630415297281147,
      "grad_norm": 0.18989917635917664,
      "learning_rate": 8.052554293072949e-05,
      "loss": 0.553,
      "step": 3769
    },
    {
      "epoch": 0.5631909172393188,
      "grad_norm": 0.1299736201763153,
      "learning_rate": 8.04794432927927e-05,
      "loss": 0.8119,
      "step": 3770
    },
    {
      "epoch": 0.5633403047505229,
      "grad_norm": 0.26445773243904114,
      "learning_rate": 8.043334796725608e-05,
      "loss": 0.883,
      "step": 3771
    },
    {
      "epoch": 0.563489692261727,
      "grad_norm": 0.1414213627576828,
      "learning_rate": 8.038725696430281e-05,
      "loss": 0.7275,
      "step": 3772
    },
    {
      "epoch": 0.5636390797729309,
      "grad_norm": 0.11140990257263184,
      "learning_rate": 8.034117029411514e-05,
      "loss": 0.6367,
      "step": 3773
    },
    {
      "epoch": 0.563788467284135,
      "grad_norm": 0.13109183311462402,
      "learning_rate": 8.029508796687432e-05,
      "loss": 0.6801,
      "step": 3774
    },
    {
      "epoch": 0.5639378547953391,
      "grad_norm": 0.1748274713754654,
      "learning_rate": 8.024900999276071e-05,
      "loss": 0.7153,
      "step": 3775
    },
    {
      "epoch": 0.5640872423065432,
      "grad_norm": 0.16637177765369415,
      "learning_rate": 8.020293638195361e-05,
      "loss": 0.6909,
      "step": 3776
    },
    {
      "epoch": 0.5642366298177472,
      "grad_norm": 0.1365109384059906,
      "learning_rate": 8.015686714463142e-05,
      "loss": 0.7146,
      "step": 3777
    },
    {
      "epoch": 0.5643860173289513,
      "grad_norm": 0.1521029770374298,
      "learning_rate": 8.011080229097159e-05,
      "loss": 0.6353,
      "step": 3778
    },
    {
      "epoch": 0.5645354048401554,
      "grad_norm": 0.24573270976543427,
      "learning_rate": 8.006474183115054e-05,
      "loss": 0.9641,
      "step": 3779
    },
    {
      "epoch": 0.5646847923513594,
      "grad_norm": 0.1258292943239212,
      "learning_rate": 8.00186857753438e-05,
      "loss": 0.7777,
      "step": 3780
    },
    {
      "epoch": 0.5648341798625635,
      "grad_norm": 0.1409403532743454,
      "learning_rate": 7.997263413372584e-05,
      "loss": 0.7949,
      "step": 3781
    },
    {
      "epoch": 0.5649835673737675,
      "grad_norm": 0.2629169225692749,
      "learning_rate": 7.992658691647027e-05,
      "loss": 0.9995,
      "step": 3782
    },
    {
      "epoch": 0.5651329548849716,
      "grad_norm": 0.22850371897220612,
      "learning_rate": 7.988054413374955e-05,
      "loss": 0.9332,
      "step": 3783
    },
    {
      "epoch": 0.5652823423961757,
      "grad_norm": 0.14643803238868713,
      "learning_rate": 7.98345057957353e-05,
      "loss": 0.5565,
      "step": 3784
    },
    {
      "epoch": 0.5654317299073798,
      "grad_norm": 0.14261296391487122,
      "learning_rate": 7.978847191259814e-05,
      "loss": 0.6517,
      "step": 3785
    },
    {
      "epoch": 0.5655811174185839,
      "grad_norm": 0.26771625876426697,
      "learning_rate": 7.974244249450767e-05,
      "loss": 0.9581,
      "step": 3786
    },
    {
      "epoch": 0.5657305049297878,
      "grad_norm": 0.11276651918888092,
      "learning_rate": 7.96964175516325e-05,
      "loss": 0.5765,
      "step": 3787
    },
    {
      "epoch": 0.5658798924409919,
      "grad_norm": 0.24681095778942108,
      "learning_rate": 7.965039709414032e-05,
      "loss": 0.7279,
      "step": 3788
    },
    {
      "epoch": 0.566029279952196,
      "grad_norm": 0.11931382119655609,
      "learning_rate": 7.960438113219772e-05,
      "loss": 0.7045,
      "step": 3789
    },
    {
      "epoch": 0.5661786674634001,
      "grad_norm": 0.1853770762681961,
      "learning_rate": 7.955836967597038e-05,
      "loss": 1.1989,
      "step": 3790
    },
    {
      "epoch": 0.5663280549746041,
      "grad_norm": 0.14835992455482483,
      "learning_rate": 7.951236273562294e-05,
      "loss": 0.6195,
      "step": 3791
    },
    {
      "epoch": 0.5664774424858082,
      "grad_norm": 0.15975846350193024,
      "learning_rate": 7.946636032131912e-05,
      "loss": 0.7317,
      "step": 3792
    },
    {
      "epoch": 0.5666268299970123,
      "grad_norm": 0.16547299921512604,
      "learning_rate": 7.94203624432215e-05,
      "loss": 0.6659,
      "step": 3793
    },
    {
      "epoch": 0.5667762175082163,
      "grad_norm": 0.2057892382144928,
      "learning_rate": 7.937436911149184e-05,
      "loss": 0.6265,
      "step": 3794
    },
    {
      "epoch": 0.5669256050194204,
      "grad_norm": 0.15118105709552765,
      "learning_rate": 7.932838033629068e-05,
      "loss": 0.5566,
      "step": 3795
    },
    {
      "epoch": 0.5670749925306244,
      "grad_norm": 0.11863881349563599,
      "learning_rate": 7.928239612777775e-05,
      "loss": 0.6984,
      "step": 3796
    },
    {
      "epoch": 0.5672243800418285,
      "grad_norm": 0.27953463792800903,
      "learning_rate": 7.923641649611162e-05,
      "loss": 1.155,
      "step": 3797
    },
    {
      "epoch": 0.5673737675530326,
      "grad_norm": 0.1522323489189148,
      "learning_rate": 7.919044145145e-05,
      "loss": 0.7591,
      "step": 3798
    },
    {
      "epoch": 0.5675231550642367,
      "grad_norm": 0.31986960768699646,
      "learning_rate": 7.914447100394947e-05,
      "loss": 0.7597,
      "step": 3799
    },
    {
      "epoch": 0.5676725425754406,
      "grad_norm": 0.19413968920707703,
      "learning_rate": 7.909850516376563e-05,
      "loss": 0.8594,
      "step": 3800
    },
    {
      "epoch": 0.5678219300866447,
      "grad_norm": 0.4680337905883789,
      "learning_rate": 7.905254394105309e-05,
      "loss": 0.8148,
      "step": 3801
    },
    {
      "epoch": 0.5679713175978488,
      "grad_norm": 0.12379107624292374,
      "learning_rate": 7.900658734596536e-05,
      "loss": 0.5341,
      "step": 3802
    },
    {
      "epoch": 0.5681207051090529,
      "grad_norm": 0.13258050382137299,
      "learning_rate": 7.8960635388655e-05,
      "loss": 0.8107,
      "step": 3803
    },
    {
      "epoch": 0.568270092620257,
      "grad_norm": 0.20536398887634277,
      "learning_rate": 7.891468807927351e-05,
      "loss": 0.8096,
      "step": 3804
    },
    {
      "epoch": 0.568419480131461,
      "grad_norm": 0.16344012320041656,
      "learning_rate": 7.886874542797142e-05,
      "loss": 0.6543,
      "step": 3805
    },
    {
      "epoch": 0.5685688676426651,
      "grad_norm": 0.17161422967910767,
      "learning_rate": 7.882280744489815e-05,
      "loss": 0.6915,
      "step": 3806
    },
    {
      "epoch": 0.5687182551538691,
      "grad_norm": 0.14980784058570862,
      "learning_rate": 7.877687414020218e-05,
      "loss": 0.7718,
      "step": 3807
    },
    {
      "epoch": 0.5688676426650732,
      "grad_norm": 0.20656977593898773,
      "learning_rate": 7.873094552403083e-05,
      "loss": 1.1037,
      "step": 3808
    },
    {
      "epoch": 0.5690170301762772,
      "grad_norm": 0.14137916266918182,
      "learning_rate": 7.868502160653051e-05,
      "loss": 0.6537,
      "step": 3809
    },
    {
      "epoch": 0.5691664176874813,
      "grad_norm": 0.28903499245643616,
      "learning_rate": 7.863910239784653e-05,
      "loss": 0.7931,
      "step": 3810
    },
    {
      "epoch": 0.5693158051986854,
      "grad_norm": 0.21949853003025055,
      "learning_rate": 7.859318790812318e-05,
      "loss": 1.0592,
      "step": 3811
    },
    {
      "epoch": 0.5694651927098895,
      "grad_norm": 0.1494913399219513,
      "learning_rate": 7.854727814750366e-05,
      "loss": 0.5317,
      "step": 3812
    },
    {
      "epoch": 0.5696145802210936,
      "grad_norm": 0.1892702877521515,
      "learning_rate": 7.850137312613021e-05,
      "loss": 0.8863,
      "step": 3813
    },
    {
      "epoch": 0.5697639677322975,
      "grad_norm": 0.1259474754333496,
      "learning_rate": 7.845547285414399e-05,
      "loss": 0.6729,
      "step": 3814
    },
    {
      "epoch": 0.5699133552435016,
      "grad_norm": 0.3680558502674103,
      "learning_rate": 7.840957734168501e-05,
      "loss": 0.8799,
      "step": 3815
    },
    {
      "epoch": 0.5700627427547057,
      "grad_norm": 0.15485581755638123,
      "learning_rate": 7.83636865988924e-05,
      "loss": 0.9335,
      "step": 3816
    },
    {
      "epoch": 0.5702121302659098,
      "grad_norm": 0.1759268194437027,
      "learning_rate": 7.831780063590411e-05,
      "loss": 0.7076,
      "step": 3817
    },
    {
      "epoch": 0.5703615177771139,
      "grad_norm": 0.1802474856376648,
      "learning_rate": 7.827191946285709e-05,
      "loss": 0.8526,
      "step": 3818
    },
    {
      "epoch": 0.5705109052883179,
      "grad_norm": 0.15598149597644806,
      "learning_rate": 7.82260430898872e-05,
      "loss": 0.5915,
      "step": 3819
    },
    {
      "epoch": 0.5706602927995219,
      "grad_norm": 0.13497863709926605,
      "learning_rate": 7.818017152712933e-05,
      "loss": 0.813,
      "step": 3820
    },
    {
      "epoch": 0.570809680310726,
      "grad_norm": 0.2554977536201477,
      "learning_rate": 7.813430478471714e-05,
      "loss": 0.8248,
      "step": 3821
    },
    {
      "epoch": 0.5709590678219301,
      "grad_norm": 0.18779698014259338,
      "learning_rate": 7.808844287278336e-05,
      "loss": 1.143,
      "step": 3822
    },
    {
      "epoch": 0.5711084553331341,
      "grad_norm": 0.12611514329910278,
      "learning_rate": 7.80425858014596e-05,
      "loss": 0.572,
      "step": 3823
    },
    {
      "epoch": 0.5712578428443382,
      "grad_norm": 0.18070462346076965,
      "learning_rate": 7.799673358087643e-05,
      "loss": 0.9139,
      "step": 3824
    },
    {
      "epoch": 0.5714072303555423,
      "grad_norm": 0.17997057735919952,
      "learning_rate": 7.795088622116329e-05,
      "loss": 0.9731,
      "step": 3825
    },
    {
      "epoch": 0.5715566178667464,
      "grad_norm": 0.1305846869945526,
      "learning_rate": 7.790504373244866e-05,
      "loss": 0.7932,
      "step": 3826
    },
    {
      "epoch": 0.5717060053779504,
      "grad_norm": 0.1776777058839798,
      "learning_rate": 7.785920612485987e-05,
      "loss": 0.9175,
      "step": 3827
    },
    {
      "epoch": 0.5718553928891544,
      "grad_norm": 0.13384713232517242,
      "learning_rate": 7.78133734085231e-05,
      "loss": 0.7657,
      "step": 3828
    },
    {
      "epoch": 0.5720047804003585,
      "grad_norm": 0.1338573694229126,
      "learning_rate": 7.776754559356356e-05,
      "loss": 0.7558,
      "step": 3829
    },
    {
      "epoch": 0.5721541679115626,
      "grad_norm": 0.12736578285694122,
      "learning_rate": 7.772172269010535e-05,
      "loss": 0.7916,
      "step": 3830
    },
    {
      "epoch": 0.5723035554227667,
      "grad_norm": 0.16737563908100128,
      "learning_rate": 7.767590470827146e-05,
      "loss": 0.6945,
      "step": 3831
    },
    {
      "epoch": 0.5724529429339708,
      "grad_norm": 0.4356246888637543,
      "learning_rate": 7.763009165818382e-05,
      "loss": 1.0445,
      "step": 3832
    },
    {
      "epoch": 0.5726023304451748,
      "grad_norm": 0.13476909697055817,
      "learning_rate": 7.758428354996328e-05,
      "loss": 0.8385,
      "step": 3833
    },
    {
      "epoch": 0.5727517179563788,
      "grad_norm": 0.11855707317590714,
      "learning_rate": 7.75384803937295e-05,
      "loss": 0.7117,
      "step": 3834
    },
    {
      "epoch": 0.5729011054675829,
      "grad_norm": 0.13812218606472015,
      "learning_rate": 7.749268219960121e-05,
      "loss": 0.8238,
      "step": 3835
    },
    {
      "epoch": 0.573050492978787,
      "grad_norm": 0.5641968846321106,
      "learning_rate": 7.74468889776959e-05,
      "loss": 1.565,
      "step": 3836
    },
    {
      "epoch": 0.573199880489991,
      "grad_norm": 0.22815346717834473,
      "learning_rate": 7.740110073813005e-05,
      "loss": 0.4943,
      "step": 3837
    },
    {
      "epoch": 0.5733492680011951,
      "grad_norm": 0.12528541684150696,
      "learning_rate": 7.735531749101898e-05,
      "loss": 0.7394,
      "step": 3838
    },
    {
      "epoch": 0.5734986555123992,
      "grad_norm": 0.13271485269069672,
      "learning_rate": 7.730953924647695e-05,
      "loss": 0.5444,
      "step": 3839
    },
    {
      "epoch": 0.5736480430236032,
      "grad_norm": 0.19810566306114197,
      "learning_rate": 7.726376601461716e-05,
      "loss": 0.9435,
      "step": 3840
    },
    {
      "epoch": 0.5737974305348073,
      "grad_norm": 0.13588392734527588,
      "learning_rate": 7.72179978055515e-05,
      "loss": 0.3494,
      "step": 3841
    },
    {
      "epoch": 0.5739468180460113,
      "grad_norm": 0.12368221580982208,
      "learning_rate": 7.7172234629391e-05,
      "loss": 0.6355,
      "step": 3842
    },
    {
      "epoch": 0.5740962055572154,
      "grad_norm": 0.1564825475215912,
      "learning_rate": 7.712647649624545e-05,
      "loss": 0.7027,
      "step": 3843
    },
    {
      "epoch": 0.5742455930684195,
      "grad_norm": 0.17468243837356567,
      "learning_rate": 7.708072341622352e-05,
      "loss": 0.9702,
      "step": 3844
    },
    {
      "epoch": 0.5743949805796236,
      "grad_norm": 0.1255885660648346,
      "learning_rate": 7.703497539943284e-05,
      "loss": 0.6916,
      "step": 3845
    },
    {
      "epoch": 0.5745443680908277,
      "grad_norm": 0.5438473224639893,
      "learning_rate": 7.698923245597986e-05,
      "loss": 1.1275,
      "step": 3846
    },
    {
      "epoch": 0.5746937556020316,
      "grad_norm": 0.13978679478168488,
      "learning_rate": 7.694349459596985e-05,
      "loss": 0.7971,
      "step": 3847
    },
    {
      "epoch": 0.5748431431132357,
      "grad_norm": 0.1568118780851364,
      "learning_rate": 7.68977618295071e-05,
      "loss": 0.781,
      "step": 3848
    },
    {
      "epoch": 0.5749925306244398,
      "grad_norm": 0.14749036729335785,
      "learning_rate": 7.685203416669469e-05,
      "loss": 0.7216,
      "step": 3849
    },
    {
      "epoch": 0.5751419181356439,
      "grad_norm": 0.23744988441467285,
      "learning_rate": 7.680631161763457e-05,
      "loss": 0.8567,
      "step": 3850
    },
    {
      "epoch": 0.5752913056468479,
      "grad_norm": 0.16781270503997803,
      "learning_rate": 7.676059419242756e-05,
      "loss": 0.7564,
      "step": 3851
    },
    {
      "epoch": 0.575440693158052,
      "grad_norm": 0.13872584700584412,
      "learning_rate": 7.671488190117341e-05,
      "loss": 0.8307,
      "step": 3852
    },
    {
      "epoch": 0.5755900806692561,
      "grad_norm": 0.19680532813072205,
      "learning_rate": 7.666917475397069e-05,
      "loss": 1.0952,
      "step": 3853
    },
    {
      "epoch": 0.5757394681804601,
      "grad_norm": 0.19261616468429565,
      "learning_rate": 7.662347276091677e-05,
      "loss": 0.7952,
      "step": 3854
    },
    {
      "epoch": 0.5758888556916641,
      "grad_norm": 0.24564647674560547,
      "learning_rate": 7.657777593210799e-05,
      "loss": 0.4974,
      "step": 3855
    },
    {
      "epoch": 0.5760382432028682,
      "grad_norm": 0.1957387924194336,
      "learning_rate": 7.653208427763949e-05,
      "loss": 0.835,
      "step": 3856
    },
    {
      "epoch": 0.5761876307140723,
      "grad_norm": 0.1952180117368698,
      "learning_rate": 7.64863978076053e-05,
      "loss": 0.5553,
      "step": 3857
    },
    {
      "epoch": 0.5763370182252764,
      "grad_norm": 0.2434726059436798,
      "learning_rate": 7.644071653209826e-05,
      "loss": 0.9338,
      "step": 3858
    },
    {
      "epoch": 0.5764864057364805,
      "grad_norm": 0.1729229986667633,
      "learning_rate": 7.639504046121012e-05,
      "loss": 0.6994,
      "step": 3859
    },
    {
      "epoch": 0.5766357932476845,
      "grad_norm": 0.13407431542873383,
      "learning_rate": 7.63493696050314e-05,
      "loss": 0.3684,
      "step": 3860
    },
    {
      "epoch": 0.5767851807588885,
      "grad_norm": 0.1371435970067978,
      "learning_rate": 7.630370397365155e-05,
      "loss": 0.7584,
      "step": 3861
    },
    {
      "epoch": 0.5769345682700926,
      "grad_norm": 0.12315504997968674,
      "learning_rate": 7.625804357715882e-05,
      "loss": 0.8642,
      "step": 3862
    },
    {
      "epoch": 0.5770839557812967,
      "grad_norm": 0.21973590552806854,
      "learning_rate": 7.62123884256403e-05,
      "loss": 0.6466,
      "step": 3863
    },
    {
      "epoch": 0.5772333432925008,
      "grad_norm": 0.22746072709560394,
      "learning_rate": 7.616673852918198e-05,
      "loss": 0.8426,
      "step": 3864
    },
    {
      "epoch": 0.5773827308037048,
      "grad_norm": 0.16150644421577454,
      "learning_rate": 7.612109389786865e-05,
      "loss": 0.7498,
      "step": 3865
    },
    {
      "epoch": 0.5775321183149089,
      "grad_norm": 0.1421440690755844,
      "learning_rate": 7.607545454178386e-05,
      "loss": 0.957,
      "step": 3866
    },
    {
      "epoch": 0.5776815058261129,
      "grad_norm": 0.15343300998210907,
      "learning_rate": 7.60298204710101e-05,
      "loss": 0.9488,
      "step": 3867
    },
    {
      "epoch": 0.577830893337317,
      "grad_norm": 0.21630048751831055,
      "learning_rate": 7.598419169562867e-05,
      "loss": 0.8253,
      "step": 3868
    },
    {
      "epoch": 0.577980280848521,
      "grad_norm": 0.15473878383636475,
      "learning_rate": 7.593856822571968e-05,
      "loss": 0.5641,
      "step": 3869
    },
    {
      "epoch": 0.5781296683597251,
      "grad_norm": 0.1355929672718048,
      "learning_rate": 7.589295007136206e-05,
      "loss": 0.5465,
      "step": 3870
    },
    {
      "epoch": 0.5782790558709292,
      "grad_norm": 0.3231397271156311,
      "learning_rate": 7.584733724263365e-05,
      "loss": 1.01,
      "step": 3871
    },
    {
      "epoch": 0.5784284433821333,
      "grad_norm": 0.14716920256614685,
      "learning_rate": 7.580172974961101e-05,
      "loss": 0.5791,
      "step": 3872
    },
    {
      "epoch": 0.5785778308933374,
      "grad_norm": 0.1762653887271881,
      "learning_rate": 7.575612760236953e-05,
      "loss": 0.8764,
      "step": 3873
    },
    {
      "epoch": 0.5787272184045413,
      "grad_norm": 0.1616542488336563,
      "learning_rate": 7.571053081098346e-05,
      "loss": 0.8111,
      "step": 3874
    },
    {
      "epoch": 0.5788766059157454,
      "grad_norm": 0.21095097064971924,
      "learning_rate": 7.566493938552585e-05,
      "loss": 0.7253,
      "step": 3875
    },
    {
      "epoch": 0.5790259934269495,
      "grad_norm": 0.15852631628513336,
      "learning_rate": 7.561935333606858e-05,
      "loss": 0.7678,
      "step": 3876
    },
    {
      "epoch": 0.5791753809381536,
      "grad_norm": 0.5843408107757568,
      "learning_rate": 7.557377267268234e-05,
      "loss": 1.2136,
      "step": 3877
    },
    {
      "epoch": 0.5793247684493577,
      "grad_norm": 0.1144263967871666,
      "learning_rate": 7.552819740543661e-05,
      "loss": 0.5518,
      "step": 3878
    },
    {
      "epoch": 0.5794741559605617,
      "grad_norm": 0.14258526265621185,
      "learning_rate": 7.548262754439968e-05,
      "loss": 0.6326,
      "step": 3879
    },
    {
      "epoch": 0.5796235434717658,
      "grad_norm": 0.16237901151180267,
      "learning_rate": 7.543706309963868e-05,
      "loss": 0.5512,
      "step": 3880
    },
    {
      "epoch": 0.5797729309829698,
      "grad_norm": 0.1285848170518875,
      "learning_rate": 7.539150408121947e-05,
      "loss": 0.7349,
      "step": 3881
    },
    {
      "epoch": 0.5799223184941739,
      "grad_norm": 0.12993387877941132,
      "learning_rate": 7.534595049920679e-05,
      "loss": 0.7224,
      "step": 3882
    },
    {
      "epoch": 0.5800717060053779,
      "grad_norm": 0.24460987746715546,
      "learning_rate": 7.530040236366413e-05,
      "loss": 0.7773,
      "step": 3883
    },
    {
      "epoch": 0.580221093516582,
      "grad_norm": 0.1344483494758606,
      "learning_rate": 7.525485968465384e-05,
      "loss": 0.6204,
      "step": 3884
    },
    {
      "epoch": 0.5803704810277861,
      "grad_norm": 0.22525547444820404,
      "learning_rate": 7.520932247223701e-05,
      "loss": 0.5978,
      "step": 3885
    },
    {
      "epoch": 0.5805198685389902,
      "grad_norm": 0.1786930114030838,
      "learning_rate": 7.516379073647346e-05,
      "loss": 0.631,
      "step": 3886
    },
    {
      "epoch": 0.5806692560501941,
      "grad_norm": 0.17504426836967468,
      "learning_rate": 7.511826448742194e-05,
      "loss": 0.735,
      "step": 3887
    },
    {
      "epoch": 0.5808186435613982,
      "grad_norm": 0.3059908151626587,
      "learning_rate": 7.50727437351399e-05,
      "loss": 0.8541,
      "step": 3888
    },
    {
      "epoch": 0.5809680310726023,
      "grad_norm": 0.14975513517856598,
      "learning_rate": 7.502722848968359e-05,
      "loss": 0.9295,
      "step": 3889
    },
    {
      "epoch": 0.5811174185838064,
      "grad_norm": 0.1718021184206009,
      "learning_rate": 7.498171876110805e-05,
      "loss": 0.7961,
      "step": 3890
    },
    {
      "epoch": 0.5812668060950105,
      "grad_norm": 0.12831704318523407,
      "learning_rate": 7.493621455946716e-05,
      "loss": 0.706,
      "step": 3891
    },
    {
      "epoch": 0.5814161936062145,
      "grad_norm": 0.12870027124881744,
      "learning_rate": 7.489071589481342e-05,
      "loss": 0.7288,
      "step": 3892
    },
    {
      "epoch": 0.5815655811174186,
      "grad_norm": 0.11215419322252274,
      "learning_rate": 7.484522277719824e-05,
      "loss": 0.5511,
      "step": 3893
    },
    {
      "epoch": 0.5817149686286226,
      "grad_norm": 0.12614436447620392,
      "learning_rate": 7.479973521667179e-05,
      "loss": 0.6345,
      "step": 3894
    },
    {
      "epoch": 0.5818643561398267,
      "grad_norm": 0.12616190314292908,
      "learning_rate": 7.475425322328298e-05,
      "loss": 0.5542,
      "step": 3895
    },
    {
      "epoch": 0.5820137436510308,
      "grad_norm": 0.18607626855373383,
      "learning_rate": 7.470877680707951e-05,
      "loss": 0.6146,
      "step": 3896
    },
    {
      "epoch": 0.5821631311622348,
      "grad_norm": 0.30349409580230713,
      "learning_rate": 7.466330597810783e-05,
      "loss": 0.6975,
      "step": 3897
    },
    {
      "epoch": 0.5823125186734389,
      "grad_norm": 0.18538470566272736,
      "learning_rate": 7.461784074641318e-05,
      "loss": 0.6454,
      "step": 3898
    },
    {
      "epoch": 0.582461906184643,
      "grad_norm": 0.2065892517566681,
      "learning_rate": 7.457238112203953e-05,
      "loss": 0.8177,
      "step": 3899
    },
    {
      "epoch": 0.5826112936958471,
      "grad_norm": 0.1513538956642151,
      "learning_rate": 7.452692711502964e-05,
      "loss": 0.7144,
      "step": 3900
    },
    {
      "epoch": 0.582760681207051,
      "grad_norm": 0.12807802855968475,
      "learning_rate": 7.448147873542503e-05,
      "loss": 0.6136,
      "step": 3901
    },
    {
      "epoch": 0.5829100687182551,
      "grad_norm": 0.11978505551815033,
      "learning_rate": 7.443603599326596e-05,
      "loss": 0.5578,
      "step": 3902
    },
    {
      "epoch": 0.5830594562294592,
      "grad_norm": 0.418393075466156,
      "learning_rate": 7.439059889859146e-05,
      "loss": 1.1315,
      "step": 3903
    },
    {
      "epoch": 0.5832088437406633,
      "grad_norm": 0.13452911376953125,
      "learning_rate": 7.43451674614393e-05,
      "loss": 0.4885,
      "step": 3904
    },
    {
      "epoch": 0.5833582312518674,
      "grad_norm": 0.30148932337760925,
      "learning_rate": 7.429974169184598e-05,
      "loss": 0.4897,
      "step": 3905
    },
    {
      "epoch": 0.5835076187630714,
      "grad_norm": 0.11458752304315567,
      "learning_rate": 7.42543215998468e-05,
      "loss": 0.581,
      "step": 3906
    },
    {
      "epoch": 0.5836570062742755,
      "grad_norm": 0.18898528814315796,
      "learning_rate": 7.420890719547576e-05,
      "loss": 0.7764,
      "step": 3907
    },
    {
      "epoch": 0.5838063937854795,
      "grad_norm": 0.19349533319473267,
      "learning_rate": 7.416349848876562e-05,
      "loss": 0.7552,
      "step": 3908
    },
    {
      "epoch": 0.5839557812966836,
      "grad_norm": 0.14663155376911163,
      "learning_rate": 7.411809548974792e-05,
      "loss": 0.5841,
      "step": 3909
    },
    {
      "epoch": 0.5841051688078877,
      "grad_norm": 0.13865816593170166,
      "learning_rate": 7.407269820845286e-05,
      "loss": 0.8356,
      "step": 3910
    },
    {
      "epoch": 0.5842545563190917,
      "grad_norm": 0.12688858807086945,
      "learning_rate": 7.402730665490947e-05,
      "loss": 0.5219,
      "step": 3911
    },
    {
      "epoch": 0.5844039438302958,
      "grad_norm": 0.3065115809440613,
      "learning_rate": 7.398192083914541e-05,
      "loss": 0.7056,
      "step": 3912
    },
    {
      "epoch": 0.5845533313414999,
      "grad_norm": 0.15706004202365875,
      "learning_rate": 7.393654077118711e-05,
      "loss": 0.8203,
      "step": 3913
    },
    {
      "epoch": 0.5847027188527039,
      "grad_norm": 0.1403438299894333,
      "learning_rate": 7.389116646105977e-05,
      "loss": 0.4756,
      "step": 3914
    },
    {
      "epoch": 0.5848521063639079,
      "grad_norm": 0.1498555988073349,
      "learning_rate": 7.384579791878731e-05,
      "loss": 1.0536,
      "step": 3915
    },
    {
      "epoch": 0.585001493875112,
      "grad_norm": 0.16070841252803802,
      "learning_rate": 7.380043515439237e-05,
      "loss": 0.6498,
      "step": 3916
    },
    {
      "epoch": 0.5851508813863161,
      "grad_norm": 0.13967616856098175,
      "learning_rate": 7.375507817789633e-05,
      "loss": 0.5907,
      "step": 3917
    },
    {
      "epoch": 0.5853002688975202,
      "grad_norm": 0.2041584849357605,
      "learning_rate": 7.370972699931915e-05,
      "loss": 0.6411,
      "step": 3918
    },
    {
      "epoch": 0.5854496564087243,
      "grad_norm": 0.13946913182735443,
      "learning_rate": 7.366438162867973e-05,
      "loss": 0.9397,
      "step": 3919
    },
    {
      "epoch": 0.5855990439199283,
      "grad_norm": 0.16613413393497467,
      "learning_rate": 7.361904207599551e-05,
      "loss": 0.6985,
      "step": 3920
    },
    {
      "epoch": 0.5857484314311323,
      "grad_norm": 0.1317937970161438,
      "learning_rate": 7.357370835128275e-05,
      "loss": 0.7462,
      "step": 3921
    },
    {
      "epoch": 0.5858978189423364,
      "grad_norm": 0.11932416260242462,
      "learning_rate": 7.352838046455639e-05,
      "loss": 0.4421,
      "step": 3922
    },
    {
      "epoch": 0.5860472064535405,
      "grad_norm": 0.14499500393867493,
      "learning_rate": 7.34830584258301e-05,
      "loss": 0.5846,
      "step": 3923
    },
    {
      "epoch": 0.5861965939647446,
      "grad_norm": 0.26846227049827576,
      "learning_rate": 7.34377422451162e-05,
      "loss": 0.8292,
      "step": 3924
    },
    {
      "epoch": 0.5863459814759486,
      "grad_norm": 0.22512462735176086,
      "learning_rate": 7.339243193242576e-05,
      "loss": 0.5067,
      "step": 3925
    },
    {
      "epoch": 0.5864953689871527,
      "grad_norm": 0.1390652060508728,
      "learning_rate": 7.334712749776853e-05,
      "loss": 0.6381,
      "step": 3926
    },
    {
      "epoch": 0.5866447564983568,
      "grad_norm": 0.4324862062931061,
      "learning_rate": 7.330182895115301e-05,
      "loss": 0.9083,
      "step": 3927
    },
    {
      "epoch": 0.5867941440095608,
      "grad_norm": 0.14873288571834564,
      "learning_rate": 7.325653630258633e-05,
      "loss": 0.77,
      "step": 3928
    },
    {
      "epoch": 0.5869435315207648,
      "grad_norm": 0.11334080249071121,
      "learning_rate": 7.32112495620744e-05,
      "loss": 0.7447,
      "step": 3929
    },
    {
      "epoch": 0.5870929190319689,
      "grad_norm": 0.393247127532959,
      "learning_rate": 7.316596873962177e-05,
      "loss": 0.7102,
      "step": 3930
    },
    {
      "epoch": 0.587242306543173,
      "grad_norm": 0.17770008742809296,
      "learning_rate": 7.312069384523164e-05,
      "loss": 0.5249,
      "step": 3931
    },
    {
      "epoch": 0.5873916940543771,
      "grad_norm": 0.12082204967737198,
      "learning_rate": 7.3075424888906e-05,
      "loss": 0.7426,
      "step": 3932
    },
    {
      "epoch": 0.5875410815655812,
      "grad_norm": 0.32135534286499023,
      "learning_rate": 7.303016188064547e-05,
      "loss": 1.1489,
      "step": 3933
    },
    {
      "epoch": 0.5876904690767851,
      "grad_norm": 0.12218035012483597,
      "learning_rate": 7.298490483044935e-05,
      "loss": 0.6026,
      "step": 3934
    },
    {
      "epoch": 0.5878398565879892,
      "grad_norm": 0.12067753076553345,
      "learning_rate": 7.293965374831566e-05,
      "loss": 0.7148,
      "step": 3935
    },
    {
      "epoch": 0.5879892440991933,
      "grad_norm": 0.13009998202323914,
      "learning_rate": 7.28944086442411e-05,
      "loss": 0.9748,
      "step": 3936
    },
    {
      "epoch": 0.5881386316103974,
      "grad_norm": 0.17588762938976288,
      "learning_rate": 7.284916952822096e-05,
      "loss": 0.8978,
      "step": 3937
    },
    {
      "epoch": 0.5882880191216014,
      "grad_norm": 0.3584859073162079,
      "learning_rate": 7.280393641024932e-05,
      "loss": 1.3935,
      "step": 3938
    },
    {
      "epoch": 0.5884374066328055,
      "grad_norm": 0.18953457474708557,
      "learning_rate": 7.275870930031889e-05,
      "loss": 0.8321,
      "step": 3939
    },
    {
      "epoch": 0.5885867941440096,
      "grad_norm": 0.30029547214508057,
      "learning_rate": 7.271348820842106e-05,
      "loss": 0.7472,
      "step": 3940
    },
    {
      "epoch": 0.5887361816552136,
      "grad_norm": 0.1402878314256668,
      "learning_rate": 7.266827314454587e-05,
      "loss": 0.4582,
      "step": 3941
    },
    {
      "epoch": 0.5888855691664177,
      "grad_norm": 0.17240741848945618,
      "learning_rate": 7.262306411868207e-05,
      "loss": 0.6566,
      "step": 3942
    },
    {
      "epoch": 0.5890349566776217,
      "grad_norm": 0.1249086931347847,
      "learning_rate": 7.257786114081705e-05,
      "loss": 0.516,
      "step": 3943
    },
    {
      "epoch": 0.5891843441888258,
      "grad_norm": 0.20941947400569916,
      "learning_rate": 7.253266422093683e-05,
      "loss": 0.9071,
      "step": 3944
    },
    {
      "epoch": 0.5893337317000299,
      "grad_norm": 0.4041518568992615,
      "learning_rate": 7.248747336902613e-05,
      "loss": 0.8304,
      "step": 3945
    },
    {
      "epoch": 0.589483119211234,
      "grad_norm": 0.23520752787590027,
      "learning_rate": 7.244228859506836e-05,
      "loss": 0.6214,
      "step": 3946
    },
    {
      "epoch": 0.589632506722438,
      "grad_norm": 0.19522571563720703,
      "learning_rate": 7.23971099090455e-05,
      "loss": 0.4527,
      "step": 3947
    },
    {
      "epoch": 0.589781894233642,
      "grad_norm": 0.1716553121805191,
      "learning_rate": 7.23519373209383e-05,
      "loss": 1.0067,
      "step": 3948
    },
    {
      "epoch": 0.5899312817448461,
      "grad_norm": 0.14052586257457733,
      "learning_rate": 7.230677084072608e-05,
      "loss": 0.9433,
      "step": 3949
    },
    {
      "epoch": 0.5900806692560502,
      "grad_norm": 0.16659864783287048,
      "learning_rate": 7.226161047838679e-05,
      "loss": 0.4078,
      "step": 3950
    },
    {
      "epoch": 0.5902300567672543,
      "grad_norm": 0.1860942393541336,
      "learning_rate": 7.22164562438971e-05,
      "loss": 0.9152,
      "step": 3951
    },
    {
      "epoch": 0.5903794442784583,
      "grad_norm": 0.16548529267311096,
      "learning_rate": 7.21713081472323e-05,
      "loss": 0.9183,
      "step": 3952
    },
    {
      "epoch": 0.5905288317896624,
      "grad_norm": 0.17878258228302002,
      "learning_rate": 7.212616619836631e-05,
      "loss": 0.7566,
      "step": 3953
    },
    {
      "epoch": 0.5906782193008665,
      "grad_norm": 0.1585194170475006,
      "learning_rate": 7.208103040727172e-05,
      "loss": 0.7415,
      "step": 3954
    },
    {
      "epoch": 0.5908276068120705,
      "grad_norm": 0.2812114655971527,
      "learning_rate": 7.20359007839197e-05,
      "loss": 0.7656,
      "step": 3955
    },
    {
      "epoch": 0.5909769943232746,
      "grad_norm": 0.12271907925605774,
      "learning_rate": 7.199077733828019e-05,
      "loss": 0.5591,
      "step": 3956
    },
    {
      "epoch": 0.5911263818344786,
      "grad_norm": 0.11963808536529541,
      "learning_rate": 7.194566008032155e-05,
      "loss": 0.7295,
      "step": 3957
    },
    {
      "epoch": 0.5912757693456827,
      "grad_norm": 0.2306731790304184,
      "learning_rate": 7.190054902001097e-05,
      "loss": 0.5213,
      "step": 3958
    },
    {
      "epoch": 0.5914251568568868,
      "grad_norm": 0.1385939121246338,
      "learning_rate": 7.185544416731416e-05,
      "loss": 0.6152,
      "step": 3959
    },
    {
      "epoch": 0.5915745443680909,
      "grad_norm": 0.1822139173746109,
      "learning_rate": 7.181034553219554e-05,
      "loss": 0.7582,
      "step": 3960
    },
    {
      "epoch": 0.5917239318792948,
      "grad_norm": 0.10867682844400406,
      "learning_rate": 7.176525312461808e-05,
      "loss": 0.5932,
      "step": 3961
    },
    {
      "epoch": 0.5918733193904989,
      "grad_norm": 0.21352946758270264,
      "learning_rate": 7.172016695454349e-05,
      "loss": 0.8689,
      "step": 3962
    },
    {
      "epoch": 0.592022706901703,
      "grad_norm": 0.12637072801589966,
      "learning_rate": 7.167508703193188e-05,
      "loss": 0.4706,
      "step": 3963
    },
    {
      "epoch": 0.5921720944129071,
      "grad_norm": 0.1369561403989792,
      "learning_rate": 7.16300133667422e-05,
      "loss": 0.7104,
      "step": 3964
    },
    {
      "epoch": 0.5923214819241112,
      "grad_norm": 0.08935172855854034,
      "learning_rate": 7.158494596893193e-05,
      "loss": 0.3752,
      "step": 3965
    },
    {
      "epoch": 0.5924708694353152,
      "grad_norm": 0.6286527514457703,
      "learning_rate": 7.153988484845715e-05,
      "loss": 1.7147,
      "step": 3966
    },
    {
      "epoch": 0.5926202569465193,
      "grad_norm": 0.12249976396560669,
      "learning_rate": 7.149483001527261e-05,
      "loss": 0.5909,
      "step": 3967
    },
    {
      "epoch": 0.5927696444577233,
      "grad_norm": 0.1625247299671173,
      "learning_rate": 7.144978147933162e-05,
      "loss": 1.0839,
      "step": 3968
    },
    {
      "epoch": 0.5929190319689274,
      "grad_norm": 0.1329154223203659,
      "learning_rate": 7.14047392505861e-05,
      "loss": 0.9195,
      "step": 3969
    },
    {
      "epoch": 0.5930684194801314,
      "grad_norm": 0.19672340154647827,
      "learning_rate": 7.135970333898661e-05,
      "loss": 0.7483,
      "step": 3970
    },
    {
      "epoch": 0.5932178069913355,
      "grad_norm": 0.15305165946483612,
      "learning_rate": 7.13146737544823e-05,
      "loss": 0.5936,
      "step": 3971
    },
    {
      "epoch": 0.5933671945025396,
      "grad_norm": 0.21835391223430634,
      "learning_rate": 7.12696505070209e-05,
      "loss": 0.6522,
      "step": 3972
    },
    {
      "epoch": 0.5935165820137437,
      "grad_norm": 0.20291920006275177,
      "learning_rate": 7.122463360654876e-05,
      "loss": 0.6598,
      "step": 3973
    },
    {
      "epoch": 0.5936659695249478,
      "grad_norm": 0.1658279299736023,
      "learning_rate": 7.117962306301084e-05,
      "loss": 1.0237,
      "step": 3974
    },
    {
      "epoch": 0.5938153570361517,
      "grad_norm": 0.19565875828266144,
      "learning_rate": 7.113461888635071e-05,
      "loss": 0.9106,
      "step": 3975
    },
    {
      "epoch": 0.5939647445473558,
      "grad_norm": 0.17244844138622284,
      "learning_rate": 7.108962108651044e-05,
      "loss": 0.6446,
      "step": 3976
    },
    {
      "epoch": 0.5941141320585599,
      "grad_norm": 0.1427401453256607,
      "learning_rate": 7.10446296734308e-05,
      "loss": 0.6813,
      "step": 3977
    },
    {
      "epoch": 0.594263519569764,
      "grad_norm": 0.17747339606285095,
      "learning_rate": 7.099964465705106e-05,
      "loss": 0.7115,
      "step": 3978
    },
    {
      "epoch": 0.5944129070809681,
      "grad_norm": 0.1962990015745163,
      "learning_rate": 7.09546660473092e-05,
      "loss": 0.7572,
      "step": 3979
    },
    {
      "epoch": 0.5945622945921721,
      "grad_norm": 0.16694936156272888,
      "learning_rate": 7.090969385414163e-05,
      "loss": 0.9288,
      "step": 3980
    },
    {
      "epoch": 0.5947116821033761,
      "grad_norm": 0.12869974970817566,
      "learning_rate": 7.086472808748348e-05,
      "loss": 0.7422,
      "step": 3981
    },
    {
      "epoch": 0.5948610696145802,
      "grad_norm": 0.10596796870231628,
      "learning_rate": 7.081976875726842e-05,
      "loss": 0.6125,
      "step": 3982
    },
    {
      "epoch": 0.5950104571257843,
      "grad_norm": 0.14334218204021454,
      "learning_rate": 7.077481587342856e-05,
      "loss": 0.5951,
      "step": 3983
    },
    {
      "epoch": 0.5951598446369883,
      "grad_norm": 0.1588943898677826,
      "learning_rate": 7.072986944589479e-05,
      "loss": 0.9095,
      "step": 3984
    },
    {
      "epoch": 0.5953092321481924,
      "grad_norm": 0.3941342532634735,
      "learning_rate": 7.068492948459648e-05,
      "loss": 1.333,
      "step": 3985
    },
    {
      "epoch": 0.5954586196593965,
      "grad_norm": 0.17836987972259521,
      "learning_rate": 7.063999599946159e-05,
      "loss": 0.617,
      "step": 3986
    },
    {
      "epoch": 0.5956080071706006,
      "grad_norm": 0.17555870115756989,
      "learning_rate": 7.059506900041657e-05,
      "loss": 0.7269,
      "step": 3987
    },
    {
      "epoch": 0.5957573946818046,
      "grad_norm": 0.1384737342596054,
      "learning_rate": 7.055014849738664e-05,
      "loss": 0.8111,
      "step": 3988
    },
    {
      "epoch": 0.5959067821930086,
      "grad_norm": 0.11168132722377777,
      "learning_rate": 7.050523450029531e-05,
      "loss": 0.5818,
      "step": 3989
    },
    {
      "epoch": 0.5960561697042127,
      "grad_norm": 0.12163358926773071,
      "learning_rate": 7.046032701906486e-05,
      "loss": 0.6866,
      "step": 3990
    },
    {
      "epoch": 0.5962055572154168,
      "grad_norm": 0.15894478559494019,
      "learning_rate": 7.041542606361607e-05,
      "loss": 0.6012,
      "step": 3991
    },
    {
      "epoch": 0.5963549447266209,
      "grad_norm": 0.1395583301782608,
      "learning_rate": 7.037053164386824e-05,
      "loss": 0.8139,
      "step": 3992
    },
    {
      "epoch": 0.596504332237825,
      "grad_norm": 0.1398267298936844,
      "learning_rate": 7.032564376973927e-05,
      "loss": 0.9327,
      "step": 3993
    },
    {
      "epoch": 0.596653719749029,
      "grad_norm": 0.20389768481254578,
      "learning_rate": 7.028076245114566e-05,
      "loss": 0.8323,
      "step": 3994
    },
    {
      "epoch": 0.596803107260233,
      "grad_norm": 0.14743159711360931,
      "learning_rate": 7.023588769800231e-05,
      "loss": 0.6823,
      "step": 3995
    },
    {
      "epoch": 0.5969524947714371,
      "grad_norm": 0.10580015182495117,
      "learning_rate": 7.01910195202228e-05,
      "loss": 0.5874,
      "step": 3996
    },
    {
      "epoch": 0.5971018822826412,
      "grad_norm": 0.15174062550067902,
      "learning_rate": 7.014615792771923e-05,
      "loss": 0.8839,
      "step": 3997
    },
    {
      "epoch": 0.5972512697938452,
      "grad_norm": 0.1828785389661789,
      "learning_rate": 7.010130293040225e-05,
      "loss": 0.7773,
      "step": 3998
    },
    {
      "epoch": 0.5974006573050493,
      "grad_norm": 0.1852671056985855,
      "learning_rate": 7.005645453818101e-05,
      "loss": 0.5767,
      "step": 3999
    },
    {
      "epoch": 0.5975500448162534,
      "grad_norm": 0.13542863726615906,
      "learning_rate": 7.001161276096326e-05,
      "loss": 0.6538,
      "step": 4000
    },
    {
      "epoch": 0.5976994323274574,
      "grad_norm": 0.12856455147266388,
      "learning_rate": 6.996677760865527e-05,
      "loss": 0.6141,
      "step": 4001
    },
    {
      "epoch": 0.5978488198386614,
      "grad_norm": 0.1546035259962082,
      "learning_rate": 6.992194909116178e-05,
      "loss": 0.6924,
      "step": 4002
    },
    {
      "epoch": 0.5979982073498655,
      "grad_norm": 0.1371324062347412,
      "learning_rate": 6.98771272183861e-05,
      "loss": 0.6092,
      "step": 4003
    },
    {
      "epoch": 0.5981475948610696,
      "grad_norm": 0.14232684671878815,
      "learning_rate": 6.98323120002302e-05,
      "loss": 0.6003,
      "step": 4004
    },
    {
      "epoch": 0.5982969823722737,
      "grad_norm": 0.12170089781284332,
      "learning_rate": 6.978750344659436e-05,
      "loss": 0.6995,
      "step": 4005
    },
    {
      "epoch": 0.5984463698834778,
      "grad_norm": 0.12605628371238708,
      "learning_rate": 6.97427015673776e-05,
      "loss": 0.7143,
      "step": 4006
    },
    {
      "epoch": 0.5985957573946818,
      "grad_norm": 0.1482643485069275,
      "learning_rate": 6.969790637247732e-05,
      "loss": 0.742,
      "step": 4007
    },
    {
      "epoch": 0.5987451449058858,
      "grad_norm": 0.16465796530246735,
      "learning_rate": 6.965311787178946e-05,
      "loss": 0.8375,
      "step": 4008
    },
    {
      "epoch": 0.5988945324170899,
      "grad_norm": 0.22890914976596832,
      "learning_rate": 6.960833607520853e-05,
      "loss": 1.3131,
      "step": 4009
    },
    {
      "epoch": 0.599043919928294,
      "grad_norm": 0.2087405025959015,
      "learning_rate": 6.956356099262753e-05,
      "loss": 0.8925,
      "step": 4010
    },
    {
      "epoch": 0.5991933074394981,
      "grad_norm": 0.11993660777807236,
      "learning_rate": 6.9518792633938e-05,
      "loss": 0.7592,
      "step": 4011
    },
    {
      "epoch": 0.5993426949507021,
      "grad_norm": 0.16774313151836395,
      "learning_rate": 6.947403100902999e-05,
      "loss": 0.9456,
      "step": 4012
    },
    {
      "epoch": 0.5994920824619062,
      "grad_norm": 0.17252188920974731,
      "learning_rate": 6.942927612779202e-05,
      "loss": 0.7122,
      "step": 4013
    },
    {
      "epoch": 0.5996414699731103,
      "grad_norm": 0.11605922877788544,
      "learning_rate": 6.938452800011119e-05,
      "loss": 0.4305,
      "step": 4014
    },
    {
      "epoch": 0.5997908574843143,
      "grad_norm": 0.19367732107639313,
      "learning_rate": 6.933978663587301e-05,
      "loss": 0.8919,
      "step": 4015
    },
    {
      "epoch": 0.5999402449955183,
      "grad_norm": 0.12205325067043304,
      "learning_rate": 6.929505204496162e-05,
      "loss": 0.757,
      "step": 4016
    },
    {
      "epoch": 0.6000896325067224,
      "grad_norm": 0.43195393681526184,
      "learning_rate": 6.925032423725958e-05,
      "loss": 1.0258,
      "step": 4017
    },
    {
      "epoch": 0.6002390200179265,
      "grad_norm": 0.15651105344295502,
      "learning_rate": 6.920560322264795e-05,
      "loss": 0.4986,
      "step": 4018
    },
    {
      "epoch": 0.6003884075291306,
      "grad_norm": 0.12385055422782898,
      "learning_rate": 6.916088901100634e-05,
      "loss": 0.8634,
      "step": 4019
    },
    {
      "epoch": 0.6005377950403347,
      "grad_norm": 0.13062192499637604,
      "learning_rate": 6.911618161221282e-05,
      "loss": 0.8793,
      "step": 4020
    },
    {
      "epoch": 0.6006871825515387,
      "grad_norm": 0.15317974984645844,
      "learning_rate": 6.907148103614393e-05,
      "loss": 0.5953,
      "step": 4021
    },
    {
      "epoch": 0.6008365700627427,
      "grad_norm": 0.12144477665424347,
      "learning_rate": 6.902678729267478e-05,
      "loss": 0.7294,
      "step": 4022
    },
    {
      "epoch": 0.6009859575739468,
      "grad_norm": 0.16116252541542053,
      "learning_rate": 6.898210039167891e-05,
      "loss": 0.5593,
      "step": 4023
    },
    {
      "epoch": 0.6011353450851509,
      "grad_norm": 0.18061742186546326,
      "learning_rate": 6.893742034302835e-05,
      "loss": 0.728,
      "step": 4024
    },
    {
      "epoch": 0.601284732596355,
      "grad_norm": 0.2507433593273163,
      "learning_rate": 6.889274715659364e-05,
      "loss": 0.8281,
      "step": 4025
    },
    {
      "epoch": 0.601434120107559,
      "grad_norm": 0.12418647110462189,
      "learning_rate": 6.88480808422438e-05,
      "loss": 0.5085,
      "step": 4026
    },
    {
      "epoch": 0.6015835076187631,
      "grad_norm": 0.1820501685142517,
      "learning_rate": 6.880342140984637e-05,
      "loss": 0.9774,
      "step": 4027
    },
    {
      "epoch": 0.6017328951299671,
      "grad_norm": 0.18762658536434174,
      "learning_rate": 6.875876886926723e-05,
      "loss": 0.5719,
      "step": 4028
    },
    {
      "epoch": 0.6018822826411712,
      "grad_norm": 0.5203716158866882,
      "learning_rate": 6.871412323037087e-05,
      "loss": 1.2679,
      "step": 4029
    },
    {
      "epoch": 0.6020316701523752,
      "grad_norm": 0.2447790950536728,
      "learning_rate": 6.866948450302025e-05,
      "loss": 1.0555,
      "step": 4030
    },
    {
      "epoch": 0.6021810576635793,
      "grad_norm": 0.10851575434207916,
      "learning_rate": 6.862485269707675e-05,
      "loss": 0.5775,
      "step": 4031
    },
    {
      "epoch": 0.6023304451747834,
      "grad_norm": 0.1273108720779419,
      "learning_rate": 6.858022782240024e-05,
      "loss": 0.6277,
      "step": 4032
    },
    {
      "epoch": 0.6024798326859875,
      "grad_norm": 0.6666232943534851,
      "learning_rate": 6.85356098888491e-05,
      "loss": 1.3718,
      "step": 4033
    },
    {
      "epoch": 0.6026292201971916,
      "grad_norm": 0.11813711374998093,
      "learning_rate": 6.849099890628008e-05,
      "loss": 0.6874,
      "step": 4034
    },
    {
      "epoch": 0.6027786077083955,
      "grad_norm": 0.15924116969108582,
      "learning_rate": 6.844639488454847e-05,
      "loss": 0.4754,
      "step": 4035
    },
    {
      "epoch": 0.6029279952195996,
      "grad_norm": 0.11908827722072601,
      "learning_rate": 6.840179783350802e-05,
      "loss": 0.617,
      "step": 4036
    },
    {
      "epoch": 0.6030773827308037,
      "grad_norm": 0.1404748111963272,
      "learning_rate": 6.835720776301092e-05,
      "loss": 0.588,
      "step": 4037
    },
    {
      "epoch": 0.6032267702420078,
      "grad_norm": 0.14913105964660645,
      "learning_rate": 6.831262468290782e-05,
      "loss": 0.6494,
      "step": 4038
    },
    {
      "epoch": 0.6033761577532119,
      "grad_norm": 0.2038773000240326,
      "learning_rate": 6.826804860304783e-05,
      "loss": 0.9746,
      "step": 4039
    },
    {
      "epoch": 0.6035255452644159,
      "grad_norm": 0.23243093490600586,
      "learning_rate": 6.822347953327852e-05,
      "loss": 0.938,
      "step": 4040
    },
    {
      "epoch": 0.60367493277562,
      "grad_norm": 0.1794816106557846,
      "learning_rate": 6.817891748344589e-05,
      "loss": 0.6799,
      "step": 4041
    },
    {
      "epoch": 0.603824320286824,
      "grad_norm": 0.411724328994751,
      "learning_rate": 6.813436246339439e-05,
      "loss": 0.7609,
      "step": 4042
    },
    {
      "epoch": 0.6039737077980281,
      "grad_norm": 0.14042165875434875,
      "learning_rate": 6.808981448296697e-05,
      "loss": 0.8538,
      "step": 4043
    },
    {
      "epoch": 0.6041230953092321,
      "grad_norm": 0.17963655292987823,
      "learning_rate": 6.804527355200496e-05,
      "loss": 0.7629,
      "step": 4044
    },
    {
      "epoch": 0.6042724828204362,
      "grad_norm": 0.1791219562292099,
      "learning_rate": 6.800073968034814e-05,
      "loss": 0.7883,
      "step": 4045
    },
    {
      "epoch": 0.6044218703316403,
      "grad_norm": 0.28506964445114136,
      "learning_rate": 6.795621287783482e-05,
      "loss": 0.5571,
      "step": 4046
    },
    {
      "epoch": 0.6045712578428444,
      "grad_norm": 0.14071182906627655,
      "learning_rate": 6.79116931543016e-05,
      "loss": 0.5766,
      "step": 4047
    },
    {
      "epoch": 0.6047206453540483,
      "grad_norm": 0.15433645248413086,
      "learning_rate": 6.786718051958357e-05,
      "loss": 0.8602,
      "step": 4048
    },
    {
      "epoch": 0.6048700328652524,
      "grad_norm": 0.18325099349021912,
      "learning_rate": 6.782267498351433e-05,
      "loss": 0.7521,
      "step": 4049
    },
    {
      "epoch": 0.6050194203764565,
      "grad_norm": 0.29621168971061707,
      "learning_rate": 6.777817655592587e-05,
      "loss": 1.1007,
      "step": 4050
    },
    {
      "epoch": 0.6051688078876606,
      "grad_norm": 0.16080495715141296,
      "learning_rate": 6.773368524664858e-05,
      "loss": 0.8578,
      "step": 4051
    },
    {
      "epoch": 0.6053181953988647,
      "grad_norm": 0.12383773922920227,
      "learning_rate": 6.768920106551128e-05,
      "loss": 0.3336,
      "step": 4052
    },
    {
      "epoch": 0.6054675829100687,
      "grad_norm": 0.12978647649288177,
      "learning_rate": 6.764472402234127e-05,
      "loss": 0.5824,
      "step": 4053
    },
    {
      "epoch": 0.6056169704212728,
      "grad_norm": 0.23082269728183746,
      "learning_rate": 6.760025412696419e-05,
      "loss": 1.1223,
      "step": 4054
    },
    {
      "epoch": 0.6057663579324768,
      "grad_norm": 0.2578333020210266,
      "learning_rate": 6.755579138920414e-05,
      "loss": 0.8953,
      "step": 4055
    },
    {
      "epoch": 0.6059157454436809,
      "grad_norm": 0.16579687595367432,
      "learning_rate": 6.751133581888367e-05,
      "loss": 0.7032,
      "step": 4056
    },
    {
      "epoch": 0.606065132954885,
      "grad_norm": 0.16161248087882996,
      "learning_rate": 6.746688742582372e-05,
      "loss": 0.8465,
      "step": 4057
    },
    {
      "epoch": 0.606214520466089,
      "grad_norm": 0.15905603766441345,
      "learning_rate": 6.742244621984362e-05,
      "loss": 0.5033,
      "step": 4058
    },
    {
      "epoch": 0.6063639079772931,
      "grad_norm": 0.1540277898311615,
      "learning_rate": 6.737801221076117e-05,
      "loss": 1.07,
      "step": 4059
    },
    {
      "epoch": 0.6065132954884972,
      "grad_norm": 0.1749313622713089,
      "learning_rate": 6.733358540839253e-05,
      "loss": 0.7973,
      "step": 4060
    },
    {
      "epoch": 0.6066626829997013,
      "grad_norm": 0.1666993349790573,
      "learning_rate": 6.728916582255226e-05,
      "loss": 0.9205,
      "step": 4061
    },
    {
      "epoch": 0.6068120705109052,
      "grad_norm": 0.16259324550628662,
      "learning_rate": 6.724475346305338e-05,
      "loss": 0.7615,
      "step": 4062
    },
    {
      "epoch": 0.6069614580221093,
      "grad_norm": 0.39470216631889343,
      "learning_rate": 6.720034833970728e-05,
      "loss": 0.8519,
      "step": 4063
    },
    {
      "epoch": 0.6071108455333134,
      "grad_norm": 0.18858696520328522,
      "learning_rate": 6.715595046232374e-05,
      "loss": 0.9141,
      "step": 4064
    },
    {
      "epoch": 0.6072602330445175,
      "grad_norm": 0.237192764878273,
      "learning_rate": 6.7111559840711e-05,
      "loss": 0.712,
      "step": 4065
    },
    {
      "epoch": 0.6074096205557216,
      "grad_norm": 0.14514686167240143,
      "learning_rate": 6.70671764846756e-05,
      "loss": 0.6391,
      "step": 4066
    },
    {
      "epoch": 0.6075590080669256,
      "grad_norm": 0.1541067659854889,
      "learning_rate": 6.702280040402252e-05,
      "loss": 0.6782,
      "step": 4067
    },
    {
      "epoch": 0.6077083955781297,
      "grad_norm": 0.2077447772026062,
      "learning_rate": 6.697843160855518e-05,
      "loss": 1.0379,
      "step": 4068
    },
    {
      "epoch": 0.6078577830893337,
      "grad_norm": 0.20825546979904175,
      "learning_rate": 6.693407010807533e-05,
      "loss": 0.9249,
      "step": 4069
    },
    {
      "epoch": 0.6080071706005378,
      "grad_norm": 0.17848896980285645,
      "learning_rate": 6.688971591238313e-05,
      "loss": 0.8954,
      "step": 4070
    },
    {
      "epoch": 0.6081565581117419,
      "grad_norm": 0.29883426427841187,
      "learning_rate": 6.684536903127712e-05,
      "loss": 0.8891,
      "step": 4071
    },
    {
      "epoch": 0.6083059456229459,
      "grad_norm": 0.22169750928878784,
      "learning_rate": 6.68010294745543e-05,
      "loss": 0.8083,
      "step": 4072
    },
    {
      "epoch": 0.60845533313415,
      "grad_norm": 0.12304669618606567,
      "learning_rate": 6.675669725200984e-05,
      "loss": 0.5928,
      "step": 4073
    },
    {
      "epoch": 0.6086047206453541,
      "grad_norm": 0.14341673254966736,
      "learning_rate": 6.671237237343751e-05,
      "loss": 0.8291,
      "step": 4074
    },
    {
      "epoch": 0.6087541081565581,
      "grad_norm": 0.23723852634429932,
      "learning_rate": 6.666805484862937e-05,
      "loss": 0.7033,
      "step": 4075
    },
    {
      "epoch": 0.6089034956677621,
      "grad_norm": 0.17535237967967987,
      "learning_rate": 6.66237446873759e-05,
      "loss": 0.8932,
      "step": 4076
    },
    {
      "epoch": 0.6090528831789662,
      "grad_norm": 0.34510093927383423,
      "learning_rate": 6.65794418994658e-05,
      "loss": 1.2633,
      "step": 4077
    },
    {
      "epoch": 0.6092022706901703,
      "grad_norm": 0.128973126411438,
      "learning_rate": 6.653514649468644e-05,
      "loss": 0.8202,
      "step": 4078
    },
    {
      "epoch": 0.6093516582013744,
      "grad_norm": 0.14690040051937103,
      "learning_rate": 6.649085848282324e-05,
      "loss": 0.8704,
      "step": 4079
    },
    {
      "epoch": 0.6095010457125785,
      "grad_norm": 0.15245862305164337,
      "learning_rate": 6.644657787366013e-05,
      "loss": 0.6943,
      "step": 4080
    },
    {
      "epoch": 0.6096504332237825,
      "grad_norm": 0.12698648869991302,
      "learning_rate": 6.640230467697945e-05,
      "loss": 0.7293,
      "step": 4081
    },
    {
      "epoch": 0.6097998207349865,
      "grad_norm": 0.16013909876346588,
      "learning_rate": 6.635803890256181e-05,
      "loss": 0.7269,
      "step": 4082
    },
    {
      "epoch": 0.6099492082461906,
      "grad_norm": 0.1540268212556839,
      "learning_rate": 6.631378056018624e-05,
      "loss": 0.4553,
      "step": 4083
    },
    {
      "epoch": 0.6100985957573947,
      "grad_norm": 0.11974772065877914,
      "learning_rate": 6.626952965963012e-05,
      "loss": 0.6139,
      "step": 4084
    },
    {
      "epoch": 0.6102479832685987,
      "grad_norm": 0.1744173914194107,
      "learning_rate": 6.622528621066918e-05,
      "loss": 0.6543,
      "step": 4085
    },
    {
      "epoch": 0.6103973707798028,
      "grad_norm": 0.16362139582633972,
      "learning_rate": 6.618105022307746e-05,
      "loss": 0.7958,
      "step": 4086
    },
    {
      "epoch": 0.6105467582910069,
      "grad_norm": 0.17188912630081177,
      "learning_rate": 6.613682170662739e-05,
      "loss": 0.8732,
      "step": 4087
    },
    {
      "epoch": 0.610696145802211,
      "grad_norm": 0.1675652414560318,
      "learning_rate": 6.609260067108979e-05,
      "loss": 0.5542,
      "step": 4088
    },
    {
      "epoch": 0.610845533313415,
      "grad_norm": 0.16193367540836334,
      "learning_rate": 6.604838712623375e-05,
      "loss": 0.6665,
      "step": 4089
    },
    {
      "epoch": 0.610994920824619,
      "grad_norm": 0.147139310836792,
      "learning_rate": 6.600418108182678e-05,
      "loss": 0.6148,
      "step": 4090
    },
    {
      "epoch": 0.6111443083358231,
      "grad_norm": 0.14002157747745514,
      "learning_rate": 6.595998254763471e-05,
      "loss": 0.6167,
      "step": 4091
    },
    {
      "epoch": 0.6112936958470272,
      "grad_norm": 0.14420591294765472,
      "learning_rate": 6.59157915334216e-05,
      "loss": 0.518,
      "step": 4092
    },
    {
      "epoch": 0.6114430833582313,
      "grad_norm": 0.14218035340309143,
      "learning_rate": 6.587160804895002e-05,
      "loss": 0.798,
      "step": 4093
    },
    {
      "epoch": 0.6115924708694354,
      "grad_norm": 0.27137085795402527,
      "learning_rate": 6.582743210398079e-05,
      "loss": 0.8526,
      "step": 4094
    },
    {
      "epoch": 0.6117418583806393,
      "grad_norm": 0.21052435040473938,
      "learning_rate": 6.578326370827306e-05,
      "loss": 0.4418,
      "step": 4095
    },
    {
      "epoch": 0.6118912458918434,
      "grad_norm": 0.12852846086025238,
      "learning_rate": 6.573910287158437e-05,
      "loss": 0.9071,
      "step": 4096
    },
    {
      "epoch": 0.6120406334030475,
      "grad_norm": 0.1987370252609253,
      "learning_rate": 6.56949496036705e-05,
      "loss": 0.6996,
      "step": 4097
    },
    {
      "epoch": 0.6121900209142516,
      "grad_norm": 0.20965783298015594,
      "learning_rate": 6.565080391428568e-05,
      "loss": 0.8524,
      "step": 4098
    },
    {
      "epoch": 0.6123394084254556,
      "grad_norm": 0.20166808366775513,
      "learning_rate": 6.560666581318229e-05,
      "loss": 1.1299,
      "step": 4099
    },
    {
      "epoch": 0.6124887959366597,
      "grad_norm": 0.20372559130191803,
      "learning_rate": 6.556253531011119e-05,
      "loss": 0.8237,
      "step": 4100
    },
    {
      "epoch": 0.6126381834478638,
      "grad_norm": 0.16287700831890106,
      "learning_rate": 6.551841241482148e-05,
      "loss": 1.0301,
      "step": 4101
    },
    {
      "epoch": 0.6127875709590678,
      "grad_norm": 0.18008066713809967,
      "learning_rate": 6.547429713706066e-05,
      "loss": 1.0209,
      "step": 4102
    },
    {
      "epoch": 0.6129369584702719,
      "grad_norm": 0.5854524374008179,
      "learning_rate": 6.543018948657445e-05,
      "loss": 1.7191,
      "step": 4103
    },
    {
      "epoch": 0.6130863459814759,
      "grad_norm": 0.2777935266494751,
      "learning_rate": 6.538608947310694e-05,
      "loss": 1.2866,
      "step": 4104
    },
    {
      "epoch": 0.61323573349268,
      "grad_norm": 0.13467517495155334,
      "learning_rate": 6.534199710640052e-05,
      "loss": 0.5006,
      "step": 4105
    },
    {
      "epoch": 0.6133851210038841,
      "grad_norm": 0.22652876377105713,
      "learning_rate": 6.529791239619589e-05,
      "loss": 0.9822,
      "step": 4106
    },
    {
      "epoch": 0.6135345085150882,
      "grad_norm": 0.19572187960147858,
      "learning_rate": 6.525383535223207e-05,
      "loss": 0.9968,
      "step": 4107
    },
    {
      "epoch": 0.6136838960262923,
      "grad_norm": 0.15496830642223358,
      "learning_rate": 6.520976598424637e-05,
      "loss": 0.6834,
      "step": 4108
    },
    {
      "epoch": 0.6138332835374962,
      "grad_norm": 0.2195952832698822,
      "learning_rate": 6.516570430197441e-05,
      "loss": 0.6965,
      "step": 4109
    },
    {
      "epoch": 0.6139826710487003,
      "grad_norm": 0.6817240118980408,
      "learning_rate": 6.512165031515012e-05,
      "loss": 1.4026,
      "step": 4110
    },
    {
      "epoch": 0.6141320585599044,
      "grad_norm": 0.19333189725875854,
      "learning_rate": 6.507760403350574e-05,
      "loss": 1.1389,
      "step": 4111
    },
    {
      "epoch": 0.6142814460711085,
      "grad_norm": 0.19699609279632568,
      "learning_rate": 6.503356546677173e-05,
      "loss": 0.5924,
      "step": 4112
    },
    {
      "epoch": 0.6144308335823125,
      "grad_norm": 0.14416566491127014,
      "learning_rate": 6.498953462467698e-05,
      "loss": 0.711,
      "step": 4113
    },
    {
      "epoch": 0.6145802210935166,
      "grad_norm": 0.1390933394432068,
      "learning_rate": 6.494551151694854e-05,
      "loss": 0.7822,
      "step": 4114
    },
    {
      "epoch": 0.6147296086047206,
      "grad_norm": 0.2005484253168106,
      "learning_rate": 6.490149615331186e-05,
      "loss": 0.621,
      "step": 4115
    },
    {
      "epoch": 0.6148789961159247,
      "grad_norm": 0.17496226727962494,
      "learning_rate": 6.485748854349061e-05,
      "loss": 0.772,
      "step": 4116
    },
    {
      "epoch": 0.6150283836271287,
      "grad_norm": 0.14194275438785553,
      "learning_rate": 6.48134886972068e-05,
      "loss": 0.897,
      "step": 4117
    },
    {
      "epoch": 0.6151777711383328,
      "grad_norm": 0.13411028683185577,
      "learning_rate": 6.476949662418062e-05,
      "loss": 0.687,
      "step": 4118
    },
    {
      "epoch": 0.6153271586495369,
      "grad_norm": 0.14102494716644287,
      "learning_rate": 6.472551233413066e-05,
      "loss": 0.9165,
      "step": 4119
    },
    {
      "epoch": 0.615476546160741,
      "grad_norm": 0.14294709265232086,
      "learning_rate": 6.468153583677377e-05,
      "loss": 0.805,
      "step": 4120
    },
    {
      "epoch": 0.6156259336719451,
      "grad_norm": 0.13600099086761475,
      "learning_rate": 6.463756714182498e-05,
      "loss": 0.8253,
      "step": 4121
    },
    {
      "epoch": 0.615775321183149,
      "grad_norm": 0.3051573932170868,
      "learning_rate": 6.459360625899775e-05,
      "loss": 0.9874,
      "step": 4122
    },
    {
      "epoch": 0.6159247086943531,
      "grad_norm": 0.28640317916870117,
      "learning_rate": 6.454965319800375e-05,
      "loss": 0.8259,
      "step": 4123
    },
    {
      "epoch": 0.6160740962055572,
      "grad_norm": 0.16837233304977417,
      "learning_rate": 6.450570796855282e-05,
      "loss": 0.5979,
      "step": 4124
    },
    {
      "epoch": 0.6162234837167613,
      "grad_norm": 0.15624776482582092,
      "learning_rate": 6.446177058035318e-05,
      "loss": 0.6844,
      "step": 4125
    },
    {
      "epoch": 0.6163728712279654,
      "grad_norm": 0.19457146525382996,
      "learning_rate": 6.441784104311135e-05,
      "loss": 0.846,
      "step": 4126
    },
    {
      "epoch": 0.6165222587391694,
      "grad_norm": 0.17119362950325012,
      "learning_rate": 6.437391936653202e-05,
      "loss": 0.8117,
      "step": 4127
    },
    {
      "epoch": 0.6166716462503735,
      "grad_norm": 0.15421448647975922,
      "learning_rate": 6.433000556031816e-05,
      "loss": 0.7478,
      "step": 4128
    },
    {
      "epoch": 0.6168210337615775,
      "grad_norm": 0.15223398804664612,
      "learning_rate": 6.428609963417107e-05,
      "loss": 0.8508,
      "step": 4129
    },
    {
      "epoch": 0.6169704212727816,
      "grad_norm": 0.12258417159318924,
      "learning_rate": 6.424220159779029e-05,
      "loss": 0.6715,
      "step": 4130
    },
    {
      "epoch": 0.6171198087839856,
      "grad_norm": 0.14299075305461884,
      "learning_rate": 6.41983114608735e-05,
      "loss": 0.835,
      "step": 4131
    },
    {
      "epoch": 0.6172691962951897,
      "grad_norm": 0.13055945932865143,
      "learning_rate": 6.415442923311679e-05,
      "loss": 0.7379,
      "step": 4132
    },
    {
      "epoch": 0.6174185838063938,
      "grad_norm": 0.14645138382911682,
      "learning_rate": 6.411055492421444e-05,
      "loss": 0.7369,
      "step": 4133
    },
    {
      "epoch": 0.6175679713175979,
      "grad_norm": 0.13017840683460236,
      "learning_rate": 6.406668854385895e-05,
      "loss": 0.7943,
      "step": 4134
    },
    {
      "epoch": 0.617717358828802,
      "grad_norm": 0.19769428670406342,
      "learning_rate": 6.402283010174112e-05,
      "loss": 0.7685,
      "step": 4135
    },
    {
      "epoch": 0.6178667463400059,
      "grad_norm": 0.15109023451805115,
      "learning_rate": 6.397897960755002e-05,
      "loss": 0.5233,
      "step": 4136
    },
    {
      "epoch": 0.61801613385121,
      "grad_norm": 0.15133798122406006,
      "learning_rate": 6.39351370709728e-05,
      "loss": 0.5306,
      "step": 4137
    },
    {
      "epoch": 0.6181655213624141,
      "grad_norm": 0.14193843305110931,
      "learning_rate": 6.389130250169505e-05,
      "loss": 0.7883,
      "step": 4138
    },
    {
      "epoch": 0.6183149088736182,
      "grad_norm": 0.13686639070510864,
      "learning_rate": 6.384747590940053e-05,
      "loss": 0.8284,
      "step": 4139
    },
    {
      "epoch": 0.6184642963848223,
      "grad_norm": 0.1430334895849228,
      "learning_rate": 6.38036573037712e-05,
      "loss": 0.5106,
      "step": 4140
    },
    {
      "epoch": 0.6186136838960263,
      "grad_norm": 0.35056570172309875,
      "learning_rate": 6.375984669448727e-05,
      "loss": 0.9588,
      "step": 4141
    },
    {
      "epoch": 0.6187630714072303,
      "grad_norm": 0.1276337057352066,
      "learning_rate": 6.371604409122725e-05,
      "loss": 0.4558,
      "step": 4142
    },
    {
      "epoch": 0.6189124589184344,
      "grad_norm": 0.16884633898735046,
      "learning_rate": 6.367224950366782e-05,
      "loss": 0.9239,
      "step": 4143
    },
    {
      "epoch": 0.6190618464296385,
      "grad_norm": 0.10548844188451767,
      "learning_rate": 6.362846294148383e-05,
      "loss": 0.3594,
      "step": 4144
    },
    {
      "epoch": 0.6192112339408425,
      "grad_norm": 0.2143741399049759,
      "learning_rate": 6.35846844143485e-05,
      "loss": 0.9021,
      "step": 4145
    },
    {
      "epoch": 0.6193606214520466,
      "grad_norm": 0.1424761712551117,
      "learning_rate": 6.354091393193315e-05,
      "loss": 0.656,
      "step": 4146
    },
    {
      "epoch": 0.6195100089632507,
      "grad_norm": 0.21021342277526855,
      "learning_rate": 6.349715150390741e-05,
      "loss": 0.6758,
      "step": 4147
    },
    {
      "epoch": 0.6196593964744548,
      "grad_norm": 0.11950641870498657,
      "learning_rate": 6.345339713993905e-05,
      "loss": 0.6583,
      "step": 4148
    },
    {
      "epoch": 0.6198087839856588,
      "grad_norm": 0.14674589037895203,
      "learning_rate": 6.340965084969418e-05,
      "loss": 0.805,
      "step": 4149
    },
    {
      "epoch": 0.6199581714968628,
      "grad_norm": 0.15847551822662354,
      "learning_rate": 6.336591264283697e-05,
      "loss": 0.8246,
      "step": 4150
    },
    {
      "epoch": 0.6201075590080669,
      "grad_norm": 0.08982459455728531,
      "learning_rate": 6.332218252902991e-05,
      "loss": 0.3832,
      "step": 4151
    },
    {
      "epoch": 0.620256946519271,
      "grad_norm": 0.12686467170715332,
      "learning_rate": 6.327846051793367e-05,
      "loss": 0.7103,
      "step": 4152
    },
    {
      "epoch": 0.6204063340304751,
      "grad_norm": 0.326980859041214,
      "learning_rate": 6.323474661920714e-05,
      "loss": 0.8609,
      "step": 4153
    },
    {
      "epoch": 0.6205557215416792,
      "grad_norm": 0.2614046037197113,
      "learning_rate": 6.319104084250742e-05,
      "loss": 0.6782,
      "step": 4154
    },
    {
      "epoch": 0.6207051090528832,
      "grad_norm": 0.14015229046344757,
      "learning_rate": 6.31473431974898e-05,
      "loss": 0.5961,
      "step": 4155
    },
    {
      "epoch": 0.6208544965640872,
      "grad_norm": 0.5132714509963989,
      "learning_rate": 6.310365369380779e-05,
      "loss": 0.9472,
      "step": 4156
    },
    {
      "epoch": 0.6210038840752913,
      "grad_norm": 0.2223353087902069,
      "learning_rate": 6.30599723411131e-05,
      "loss": 0.7347,
      "step": 4157
    },
    {
      "epoch": 0.6211532715864954,
      "grad_norm": 0.1852208971977234,
      "learning_rate": 6.301629914905559e-05,
      "loss": 0.6962,
      "step": 4158
    },
    {
      "epoch": 0.6213026590976994,
      "grad_norm": 0.16240467131137848,
      "learning_rate": 6.29726341272834e-05,
      "loss": 0.7868,
      "step": 4159
    },
    {
      "epoch": 0.6214520466089035,
      "grad_norm": 0.13596655428409576,
      "learning_rate": 6.292897728544282e-05,
      "loss": 0.7596,
      "step": 4160
    },
    {
      "epoch": 0.6216014341201076,
      "grad_norm": 0.12929636240005493,
      "learning_rate": 6.288532863317832e-05,
      "loss": 0.7095,
      "step": 4161
    },
    {
      "epoch": 0.6217508216313116,
      "grad_norm": 0.2984831631183624,
      "learning_rate": 6.284168818013263e-05,
      "loss": 0.8506,
      "step": 4162
    },
    {
      "epoch": 0.6219002091425156,
      "grad_norm": 0.13210748136043549,
      "learning_rate": 6.279805593594655e-05,
      "loss": 0.6764,
      "step": 4163
    },
    {
      "epoch": 0.6220495966537197,
      "grad_norm": 0.3239263594150543,
      "learning_rate": 6.275443191025916e-05,
      "loss": 0.7897,
      "step": 4164
    },
    {
      "epoch": 0.6221989841649238,
      "grad_norm": 0.17348460853099823,
      "learning_rate": 6.27108161127077e-05,
      "loss": 0.8929,
      "step": 4165
    },
    {
      "epoch": 0.6223483716761279,
      "grad_norm": 0.2410515695810318,
      "learning_rate": 6.266720855292756e-05,
      "loss": 0.8991,
      "step": 4166
    },
    {
      "epoch": 0.622497759187332,
      "grad_norm": 0.14342686533927917,
      "learning_rate": 6.26236092405524e-05,
      "loss": 0.9586,
      "step": 4167
    },
    {
      "epoch": 0.622647146698536,
      "grad_norm": 0.11895397305488586,
      "learning_rate": 6.258001818521397e-05,
      "loss": 0.5633,
      "step": 4168
    },
    {
      "epoch": 0.62279653420974,
      "grad_norm": 0.12431944906711578,
      "learning_rate": 6.253643539654226e-05,
      "loss": 0.8115,
      "step": 4169
    },
    {
      "epoch": 0.6229459217209441,
      "grad_norm": 0.3534824550151825,
      "learning_rate": 6.249286088416534e-05,
      "loss": 0.5441,
      "step": 4170
    },
    {
      "epoch": 0.6230953092321482,
      "grad_norm": 0.12586544454097748,
      "learning_rate": 6.244929465770953e-05,
      "loss": 0.5613,
      "step": 4171
    },
    {
      "epoch": 0.6232446967433523,
      "grad_norm": 0.18011412024497986,
      "learning_rate": 6.240573672679929e-05,
      "loss": 0.5647,
      "step": 4172
    },
    {
      "epoch": 0.6233940842545563,
      "grad_norm": 0.1330413818359375,
      "learning_rate": 6.236218710105727e-05,
      "loss": 0.9752,
      "step": 4173
    },
    {
      "epoch": 0.6235434717657604,
      "grad_norm": 0.24614588916301727,
      "learning_rate": 6.231864579010426e-05,
      "loss": 0.77,
      "step": 4174
    },
    {
      "epoch": 0.6236928592769645,
      "grad_norm": 0.16739724576473236,
      "learning_rate": 6.227511280355928e-05,
      "loss": 0.621,
      "step": 4175
    },
    {
      "epoch": 0.6238422467881685,
      "grad_norm": 0.24803446233272552,
      "learning_rate": 6.22315881510394e-05,
      "loss": 0.9678,
      "step": 4176
    },
    {
      "epoch": 0.6239916342993725,
      "grad_norm": 0.11379440873861313,
      "learning_rate": 6.218807184215989e-05,
      "loss": 0.6344,
      "step": 4177
    },
    {
      "epoch": 0.6241410218105766,
      "grad_norm": 0.1917128562927246,
      "learning_rate": 6.214456388653423e-05,
      "loss": 0.7536,
      "step": 4178
    },
    {
      "epoch": 0.6242904093217807,
      "grad_norm": 0.1548663228750229,
      "learning_rate": 6.210106429377402e-05,
      "loss": 0.4562,
      "step": 4179
    },
    {
      "epoch": 0.6244397968329848,
      "grad_norm": 0.11434723436832428,
      "learning_rate": 6.205757307348898e-05,
      "loss": 0.5659,
      "step": 4180
    },
    {
      "epoch": 0.6245891843441889,
      "grad_norm": 0.15858866274356842,
      "learning_rate": 6.201409023528702e-05,
      "loss": 0.5923,
      "step": 4181
    },
    {
      "epoch": 0.6247385718553929,
      "grad_norm": 0.18767806887626648,
      "learning_rate": 6.197061578877424e-05,
      "loss": 1.0656,
      "step": 4182
    },
    {
      "epoch": 0.6248879593665969,
      "grad_norm": 0.23377855122089386,
      "learning_rate": 6.192714974355473e-05,
      "loss": 0.8212,
      "step": 4183
    },
    {
      "epoch": 0.625037346877801,
      "grad_norm": 0.22544582188129425,
      "learning_rate": 6.18836921092309e-05,
      "loss": 1.0159,
      "step": 4184
    },
    {
      "epoch": 0.6251867343890051,
      "grad_norm": 0.1542222499847412,
      "learning_rate": 6.184024289540322e-05,
      "loss": 0.8497,
      "step": 4185
    },
    {
      "epoch": 0.6253361219002092,
      "grad_norm": 0.12284936010837555,
      "learning_rate": 6.179680211167031e-05,
      "loss": 0.7708,
      "step": 4186
    },
    {
      "epoch": 0.6254855094114132,
      "grad_norm": 0.14446687698364258,
      "learning_rate": 6.175336976762891e-05,
      "loss": 0.4772,
      "step": 4187
    },
    {
      "epoch": 0.6256348969226173,
      "grad_norm": 0.14486733078956604,
      "learning_rate": 6.170994587287398e-05,
      "loss": 0.6652,
      "step": 4188
    },
    {
      "epoch": 0.6257842844338213,
      "grad_norm": 0.21637411415576935,
      "learning_rate": 6.166653043699843e-05,
      "loss": 0.9393,
      "step": 4189
    },
    {
      "epoch": 0.6259336719450254,
      "grad_norm": 0.1360471099615097,
      "learning_rate": 6.162312346959348e-05,
      "loss": 0.7053,
      "step": 4190
    },
    {
      "epoch": 0.6260830594562294,
      "grad_norm": 0.11286638677120209,
      "learning_rate": 6.157972498024842e-05,
      "loss": 0.5405,
      "step": 4191
    },
    {
      "epoch": 0.6262324469674335,
      "grad_norm": 0.14911124110221863,
      "learning_rate": 6.153633497855065e-05,
      "loss": 0.6676,
      "step": 4192
    },
    {
      "epoch": 0.6263818344786376,
      "grad_norm": 0.21269449591636658,
      "learning_rate": 6.149295347408572e-05,
      "loss": 0.8238,
      "step": 4193
    },
    {
      "epoch": 0.6265312219898417,
      "grad_norm": 0.13775356113910675,
      "learning_rate": 6.144958047643732e-05,
      "loss": 0.8042,
      "step": 4194
    },
    {
      "epoch": 0.6266806095010458,
      "grad_norm": 0.1658342480659485,
      "learning_rate": 6.140621599518716e-05,
      "loss": 0.8137,
      "step": 4195
    },
    {
      "epoch": 0.6268299970122497,
      "grad_norm": 0.12744642794132233,
      "learning_rate": 6.136286003991518e-05,
      "loss": 0.7202,
      "step": 4196
    },
    {
      "epoch": 0.6269793845234538,
      "grad_norm": 0.19294977188110352,
      "learning_rate": 6.131951262019941e-05,
      "loss": 0.9991,
      "step": 4197
    },
    {
      "epoch": 0.6271287720346579,
      "grad_norm": 0.22246606647968292,
      "learning_rate": 6.127617374561597e-05,
      "loss": 1.1513,
      "step": 4198
    },
    {
      "epoch": 0.627278159545862,
      "grad_norm": 0.24722294509410858,
      "learning_rate": 6.12328434257391e-05,
      "loss": 1.2377,
      "step": 4199
    },
    {
      "epoch": 0.627427547057066,
      "grad_norm": 0.15174803137779236,
      "learning_rate": 6.118952167014115e-05,
      "loss": 0.7193,
      "step": 4200
    },
    {
      "epoch": 0.6275769345682701,
      "grad_norm": 0.41749218106269836,
      "learning_rate": 6.114620848839263e-05,
      "loss": 0.9505,
      "step": 4201
    },
    {
      "epoch": 0.6277263220794742,
      "grad_norm": 0.12974758446216583,
      "learning_rate": 6.110290389006204e-05,
      "loss": 0.7595,
      "step": 4202
    },
    {
      "epoch": 0.6278757095906782,
      "grad_norm": 0.12282485514879227,
      "learning_rate": 6.105960788471605e-05,
      "loss": 0.5761,
      "step": 4203
    },
    {
      "epoch": 0.6280250971018823,
      "grad_norm": 0.17054803669452667,
      "learning_rate": 6.10163204819195e-05,
      "loss": 0.8347,
      "step": 4204
    },
    {
      "epoch": 0.6281744846130863,
      "grad_norm": 0.14993737637996674,
      "learning_rate": 6.097304169123521e-05,
      "loss": 0.5341,
      "step": 4205
    },
    {
      "epoch": 0.6283238721242904,
      "grad_norm": 0.27448564767837524,
      "learning_rate": 6.0929771522224165e-05,
      "loss": 1.0248,
      "step": 4206
    },
    {
      "epoch": 0.6284732596354945,
      "grad_norm": 0.12859925627708435,
      "learning_rate": 6.0886509984445474e-05,
      "loss": 0.8102,
      "step": 4207
    },
    {
      "epoch": 0.6286226471466986,
      "grad_norm": 0.11825615167617798,
      "learning_rate": 6.0843257087456196e-05,
      "loss": 0.3353,
      "step": 4208
    },
    {
      "epoch": 0.6287720346579025,
      "grad_norm": 0.17913062870502472,
      "learning_rate": 6.080001284081166e-05,
      "loss": 0.6474,
      "step": 4209
    },
    {
      "epoch": 0.6289214221691066,
      "grad_norm": 0.21600456535816193,
      "learning_rate": 6.075677725406516e-05,
      "loss": 1.0472,
      "step": 4210
    },
    {
      "epoch": 0.6290708096803107,
      "grad_norm": 0.28075292706489563,
      "learning_rate": 6.0713550336768154e-05,
      "loss": 0.7068,
      "step": 4211
    },
    {
      "epoch": 0.6292201971915148,
      "grad_norm": 0.14417524635791779,
      "learning_rate": 6.067033209847015e-05,
      "loss": 0.6503,
      "step": 4212
    },
    {
      "epoch": 0.6293695847027189,
      "grad_norm": 0.13622713088989258,
      "learning_rate": 6.062712254871874e-05,
      "loss": 0.8026,
      "step": 4213
    },
    {
      "epoch": 0.629518972213923,
      "grad_norm": 0.19704632461071014,
      "learning_rate": 6.058392169705962e-05,
      "loss": 0.6711,
      "step": 4214
    },
    {
      "epoch": 0.629668359725127,
      "grad_norm": 0.15503256022930145,
      "learning_rate": 6.054072955303649e-05,
      "loss": 0.5447,
      "step": 4215
    },
    {
      "epoch": 0.629817747236331,
      "grad_norm": 0.1630752980709076,
      "learning_rate": 6.0497546126191206e-05,
      "loss": 0.6852,
      "step": 4216
    },
    {
      "epoch": 0.6299671347475351,
      "grad_norm": 0.14187735319137573,
      "learning_rate": 6.0454371426063646e-05,
      "loss": 0.6853,
      "step": 4217
    },
    {
      "epoch": 0.6301165222587392,
      "grad_norm": 0.1882721185684204,
      "learning_rate": 6.041120546219183e-05,
      "loss": 0.7355,
      "step": 4218
    },
    {
      "epoch": 0.6302659097699432,
      "grad_norm": 0.1194940060377121,
      "learning_rate": 6.036804824411177e-05,
      "loss": 0.693,
      "step": 4219
    },
    {
      "epoch": 0.6304152972811473,
      "grad_norm": 0.15827606618404388,
      "learning_rate": 6.0324899781357624e-05,
      "loss": 0.859,
      "step": 4220
    },
    {
      "epoch": 0.6305646847923514,
      "grad_norm": 0.23284271359443665,
      "learning_rate": 6.028176008346151e-05,
      "loss": 0.8566,
      "step": 4221
    },
    {
      "epoch": 0.6307140723035555,
      "grad_norm": 0.1301899254322052,
      "learning_rate": 6.023862915995371e-05,
      "loss": 0.9564,
      "step": 4222
    },
    {
      "epoch": 0.6308634598147594,
      "grad_norm": 0.15400893986225128,
      "learning_rate": 6.01955070203625e-05,
      "loss": 0.568,
      "step": 4223
    },
    {
      "epoch": 0.6310128473259635,
      "grad_norm": 0.2992945909500122,
      "learning_rate": 6.0152393674214256e-05,
      "loss": 1.0013,
      "step": 4224
    },
    {
      "epoch": 0.6311622348371676,
      "grad_norm": 0.3447125554084778,
      "learning_rate": 6.010928913103342e-05,
      "loss": 0.8055,
      "step": 4225
    },
    {
      "epoch": 0.6313116223483717,
      "grad_norm": 0.13957270979881287,
      "learning_rate": 6.006619340034243e-05,
      "loss": 0.5773,
      "step": 4226
    },
    {
      "epoch": 0.6314610098595758,
      "grad_norm": 0.15160028636455536,
      "learning_rate": 6.0023106491661875e-05,
      "loss": 0.5433,
      "step": 4227
    },
    {
      "epoch": 0.6316103973707798,
      "grad_norm": 0.24746893346309662,
      "learning_rate": 5.998002841451027e-05,
      "loss": 0.8205,
      "step": 4228
    },
    {
      "epoch": 0.6317597848819839,
      "grad_norm": 0.15253928303718567,
      "learning_rate": 5.9936959178404275e-05,
      "loss": 0.6094,
      "step": 4229
    },
    {
      "epoch": 0.6319091723931879,
      "grad_norm": 0.12057177722454071,
      "learning_rate": 5.9893898792858564e-05,
      "loss": 0.619,
      "step": 4230
    },
    {
      "epoch": 0.632058559904392,
      "grad_norm": 0.15078209340572357,
      "learning_rate": 5.9850847267385846e-05,
      "loss": 0.9344,
      "step": 4231
    },
    {
      "epoch": 0.632207947415596,
      "grad_norm": 0.11977210640907288,
      "learning_rate": 5.980780461149691e-05,
      "loss": 0.6508,
      "step": 4232
    },
    {
      "epoch": 0.6323573349268001,
      "grad_norm": 0.3247925043106079,
      "learning_rate": 5.976477083470058e-05,
      "loss": 0.9373,
      "step": 4233
    },
    {
      "epoch": 0.6325067224380042,
      "grad_norm": 0.12837880849838257,
      "learning_rate": 5.972174594650363e-05,
      "loss": 0.5755,
      "step": 4234
    },
    {
      "epoch": 0.6326561099492083,
      "grad_norm": 0.13902558386325836,
      "learning_rate": 5.967872995641097e-05,
      "loss": 0.8365,
      "step": 4235
    },
    {
      "epoch": 0.6328054974604123,
      "grad_norm": 0.21605095267295837,
      "learning_rate": 5.9635722873925514e-05,
      "loss": 0.7108,
      "step": 4236
    },
    {
      "epoch": 0.6329548849716163,
      "grad_norm": 0.17556419968605042,
      "learning_rate": 5.959272470854822e-05,
      "loss": 0.7933,
      "step": 4237
    },
    {
      "epoch": 0.6331042724828204,
      "grad_norm": 0.15964846312999725,
      "learning_rate": 5.954973546977806e-05,
      "loss": 0.6179,
      "step": 4238
    },
    {
      "epoch": 0.6332536599940245,
      "grad_norm": 0.12367497384548187,
      "learning_rate": 5.9506755167112005e-05,
      "loss": 0.8741,
      "step": 4239
    },
    {
      "epoch": 0.6334030475052286,
      "grad_norm": 0.18512670695781708,
      "learning_rate": 5.946378381004518e-05,
      "loss": 0.6874,
      "step": 4240
    },
    {
      "epoch": 0.6335524350164327,
      "grad_norm": 0.19619376957416534,
      "learning_rate": 5.9420821408070526e-05,
      "loss": 0.9387,
      "step": 4241
    },
    {
      "epoch": 0.6337018225276367,
      "grad_norm": 0.17050646245479584,
      "learning_rate": 5.937786797067917e-05,
      "loss": 0.7746,
      "step": 4242
    },
    {
      "epoch": 0.6338512100388407,
      "grad_norm": 0.13103164732456207,
      "learning_rate": 5.9334923507360196e-05,
      "loss": 0.6229,
      "step": 4243
    },
    {
      "epoch": 0.6340005975500448,
      "grad_norm": 0.16050748527050018,
      "learning_rate": 5.929198802760072e-05,
      "loss": 0.6919,
      "step": 4244
    },
    {
      "epoch": 0.6341499850612489,
      "grad_norm": 0.14817118644714355,
      "learning_rate": 5.9249061540885873e-05,
      "loss": 0.709,
      "step": 4245
    },
    {
      "epoch": 0.634299372572453,
      "grad_norm": 0.24491626024246216,
      "learning_rate": 5.920614405669882e-05,
      "loss": 0.4981,
      "step": 4246
    },
    {
      "epoch": 0.634448760083657,
      "grad_norm": 0.1774425059556961,
      "learning_rate": 5.916323558452066e-05,
      "loss": 0.5933,
      "step": 4247
    },
    {
      "epoch": 0.6345981475948611,
      "grad_norm": 0.21614766120910645,
      "learning_rate": 5.9120336133830564e-05,
      "loss": 1.0614,
      "step": 4248
    },
    {
      "epoch": 0.6347475351060652,
      "grad_norm": 0.1952279806137085,
      "learning_rate": 5.907744571410574e-05,
      "loss": 0.9411,
      "step": 4249
    },
    {
      "epoch": 0.6348969226172692,
      "grad_norm": 0.1394856572151184,
      "learning_rate": 5.903456433482133e-05,
      "loss": 0.7803,
      "step": 4250
    },
    {
      "epoch": 0.6350463101284732,
      "grad_norm": 0.1635223627090454,
      "learning_rate": 5.899169200545052e-05,
      "loss": 0.8491,
      "step": 4251
    },
    {
      "epoch": 0.6351956976396773,
      "grad_norm": 0.2811232805252075,
      "learning_rate": 5.894882873546448e-05,
      "loss": 1.4229,
      "step": 4252
    },
    {
      "epoch": 0.6353450851508814,
      "grad_norm": 0.19918611645698547,
      "learning_rate": 5.890597453433242e-05,
      "loss": 1.0013,
      "step": 4253
    },
    {
      "epoch": 0.6354944726620855,
      "grad_norm": 0.1347842812538147,
      "learning_rate": 5.886312941152146e-05,
      "loss": 0.7892,
      "step": 4254
    },
    {
      "epoch": 0.6356438601732896,
      "grad_norm": 0.1497023105621338,
      "learning_rate": 5.882029337649675e-05,
      "loss": 0.7868,
      "step": 4255
    },
    {
      "epoch": 0.6357932476844935,
      "grad_norm": 0.17877355217933655,
      "learning_rate": 5.877746643872152e-05,
      "loss": 0.5078,
      "step": 4256
    },
    {
      "epoch": 0.6359426351956976,
      "grad_norm": 0.12498283386230469,
      "learning_rate": 5.873464860765687e-05,
      "loss": 0.4329,
      "step": 4257
    },
    {
      "epoch": 0.6360920227069017,
      "grad_norm": 0.26655346155166626,
      "learning_rate": 5.8691839892761965e-05,
      "loss": 0.6838,
      "step": 4258
    },
    {
      "epoch": 0.6362414102181058,
      "grad_norm": 0.1580439954996109,
      "learning_rate": 5.864904030349396e-05,
      "loss": 0.8722,
      "step": 4259
    },
    {
      "epoch": 0.6363907977293098,
      "grad_norm": 0.22915318608283997,
      "learning_rate": 5.860624984930787e-05,
      "loss": 0.438,
      "step": 4260
    },
    {
      "epoch": 0.6365401852405139,
      "grad_norm": 0.17232942581176758,
      "learning_rate": 5.8563468539656815e-05,
      "loss": 0.536,
      "step": 4261
    },
    {
      "epoch": 0.636689572751718,
      "grad_norm": 0.1340131163597107,
      "learning_rate": 5.852069638399191e-05,
      "loss": 0.6002,
      "step": 4262
    },
    {
      "epoch": 0.636838960262922,
      "grad_norm": 0.1637933850288391,
      "learning_rate": 5.847793339176215e-05,
      "loss": 0.9123,
      "step": 4263
    },
    {
      "epoch": 0.636988347774126,
      "grad_norm": 0.1157214418053627,
      "learning_rate": 5.843517957241459e-05,
      "loss": 0.8508,
      "step": 4264
    },
    {
      "epoch": 0.6371377352853301,
      "grad_norm": 0.1234390139579773,
      "learning_rate": 5.839243493539424e-05,
      "loss": 0.6414,
      "step": 4265
    },
    {
      "epoch": 0.6372871227965342,
      "grad_norm": 0.15302959084510803,
      "learning_rate": 5.8349699490144e-05,
      "loss": 0.8569,
      "step": 4266
    },
    {
      "epoch": 0.6374365103077383,
      "grad_norm": 0.2314283698797226,
      "learning_rate": 5.8306973246104876e-05,
      "loss": 0.7364,
      "step": 4267
    },
    {
      "epoch": 0.6375858978189424,
      "grad_norm": 0.12339307367801666,
      "learning_rate": 5.8264256212715726e-05,
      "loss": 0.7548,
      "step": 4268
    },
    {
      "epoch": 0.6377352853301465,
      "grad_norm": 0.25690895318984985,
      "learning_rate": 5.822154839941345e-05,
      "loss": 1.0046,
      "step": 4269
    },
    {
      "epoch": 0.6378846728413504,
      "grad_norm": 0.15167945623397827,
      "learning_rate": 5.817884981563286e-05,
      "loss": 0.7034,
      "step": 4270
    },
    {
      "epoch": 0.6380340603525545,
      "grad_norm": 0.39939120411872864,
      "learning_rate": 5.813616047080677e-05,
      "loss": 0.8236,
      "step": 4271
    },
    {
      "epoch": 0.6381834478637586,
      "grad_norm": 0.17225416004657745,
      "learning_rate": 5.809348037436595e-05,
      "loss": 0.5085,
      "step": 4272
    },
    {
      "epoch": 0.6383328353749627,
      "grad_norm": 0.13655978441238403,
      "learning_rate": 5.805080953573905e-05,
      "loss": 0.5492,
      "step": 4273
    },
    {
      "epoch": 0.6384822228861667,
      "grad_norm": 0.15536777675151825,
      "learning_rate": 5.8008147964352756e-05,
      "loss": 0.4512,
      "step": 4274
    },
    {
      "epoch": 0.6386316103973708,
      "grad_norm": 0.24335040152072906,
      "learning_rate": 5.79654956696317e-05,
      "loss": 1.0092,
      "step": 4275
    },
    {
      "epoch": 0.6387809979085748,
      "grad_norm": 0.1810038834810257,
      "learning_rate": 5.792285266099844e-05,
      "loss": 0.7738,
      "step": 4276
    },
    {
      "epoch": 0.6389303854197789,
      "grad_norm": 0.1720893681049347,
      "learning_rate": 5.788021894787349e-05,
      "loss": 0.6069,
      "step": 4277
    },
    {
      "epoch": 0.639079772930983,
      "grad_norm": 0.1522633284330368,
      "learning_rate": 5.783759453967532e-05,
      "loss": 0.5263,
      "step": 4278
    },
    {
      "epoch": 0.639229160442187,
      "grad_norm": 0.1657140702009201,
      "learning_rate": 5.779497944582033e-05,
      "loss": 0.4911,
      "step": 4279
    },
    {
      "epoch": 0.6393785479533911,
      "grad_norm": 0.2722562849521637,
      "learning_rate": 5.7752373675722884e-05,
      "loss": 1.0562,
      "step": 4280
    },
    {
      "epoch": 0.6395279354645952,
      "grad_norm": 0.13559791445732117,
      "learning_rate": 5.770977723879525e-05,
      "loss": 0.8421,
      "step": 4281
    },
    {
      "epoch": 0.6396773229757993,
      "grad_norm": 0.20714382827281952,
      "learning_rate": 5.766719014444768e-05,
      "loss": 0.7187,
      "step": 4282
    },
    {
      "epoch": 0.6398267104870032,
      "grad_norm": 0.13428863883018494,
      "learning_rate": 5.762461240208832e-05,
      "loss": 0.5754,
      "step": 4283
    },
    {
      "epoch": 0.6399760979982073,
      "grad_norm": 0.12575839459896088,
      "learning_rate": 5.758204402112326e-05,
      "loss": 0.6942,
      "step": 4284
    },
    {
      "epoch": 0.6401254855094114,
      "grad_norm": 0.18061308562755585,
      "learning_rate": 5.753948501095659e-05,
      "loss": 0.6805,
      "step": 4285
    },
    {
      "epoch": 0.6402748730206155,
      "grad_norm": 0.22430044412612915,
      "learning_rate": 5.749693538099018e-05,
      "loss": 0.8835,
      "step": 4286
    },
    {
      "epoch": 0.6404242605318196,
      "grad_norm": 0.16515129804611206,
      "learning_rate": 5.745439514062396e-05,
      "loss": 0.7904,
      "step": 4287
    },
    {
      "epoch": 0.6405736480430236,
      "grad_norm": 0.1512410044670105,
      "learning_rate": 5.741186429925574e-05,
      "loss": 0.6813,
      "step": 4288
    },
    {
      "epoch": 0.6407230355542277,
      "grad_norm": 0.1422632932662964,
      "learning_rate": 5.736934286628125e-05,
      "loss": 0.5973,
      "step": 4289
    },
    {
      "epoch": 0.6408724230654317,
      "grad_norm": 0.14442236721515656,
      "learning_rate": 5.732683085109416e-05,
      "loss": 0.5539,
      "step": 4290
    },
    {
      "epoch": 0.6410218105766358,
      "grad_norm": 0.1460275799036026,
      "learning_rate": 5.7284328263086095e-05,
      "loss": 0.758,
      "step": 4291
    },
    {
      "epoch": 0.6411711980878398,
      "grad_norm": 0.13237841427326202,
      "learning_rate": 5.7241835111646444e-05,
      "loss": 0.492,
      "step": 4292
    },
    {
      "epoch": 0.6413205855990439,
      "grad_norm": 0.19501985609531403,
      "learning_rate": 5.7199351406162685e-05,
      "loss": 0.3966,
      "step": 4293
    },
    {
      "epoch": 0.641469973110248,
      "grad_norm": 0.15059512853622437,
      "learning_rate": 5.7156877156020106e-05,
      "loss": 0.5157,
      "step": 4294
    },
    {
      "epoch": 0.6416193606214521,
      "grad_norm": 0.15084552764892578,
      "learning_rate": 5.7114412370601976e-05,
      "loss": 0.8353,
      "step": 4295
    },
    {
      "epoch": 0.6417687481326562,
      "grad_norm": 0.18661732971668243,
      "learning_rate": 5.707195705928943e-05,
      "loss": 0.7844,
      "step": 4296
    },
    {
      "epoch": 0.6419181356438601,
      "grad_norm": 0.32447463274002075,
      "learning_rate": 5.702951123146151e-05,
      "loss": 0.7082,
      "step": 4297
    },
    {
      "epoch": 0.6420675231550642,
      "grad_norm": 0.2401759773492813,
      "learning_rate": 5.6987074896495176e-05,
      "loss": 0.8312,
      "step": 4298
    },
    {
      "epoch": 0.6422169106662683,
      "grad_norm": 0.13105745613574982,
      "learning_rate": 5.6944648063765286e-05,
      "loss": 0.6383,
      "step": 4299
    },
    {
      "epoch": 0.6423662981774724,
      "grad_norm": 0.12388674169778824,
      "learning_rate": 5.69022307426446e-05,
      "loss": 0.3819,
      "step": 4300
    },
    {
      "epoch": 0.6425156856886765,
      "grad_norm": 0.14097605645656586,
      "learning_rate": 5.685982294250378e-05,
      "loss": 0.5156,
      "step": 4301
    },
    {
      "epoch": 0.6426650731998805,
      "grad_norm": 0.14637187123298645,
      "learning_rate": 5.681742467271137e-05,
      "loss": 0.758,
      "step": 4302
    },
    {
      "epoch": 0.6428144607110845,
      "grad_norm": 0.300611287355423,
      "learning_rate": 5.6775035942633836e-05,
      "loss": 0.798,
      "step": 4303
    },
    {
      "epoch": 0.6429638482222886,
      "grad_norm": 0.16687364876270294,
      "learning_rate": 5.673265676163555e-05,
      "loss": 0.4944,
      "step": 4304
    },
    {
      "epoch": 0.6431132357334927,
      "grad_norm": 0.2866893708705902,
      "learning_rate": 5.669028713907865e-05,
      "loss": 0.6453,
      "step": 4305
    },
    {
      "epoch": 0.6432626232446967,
      "grad_norm": 0.19865016639232635,
      "learning_rate": 5.664792708432333e-05,
      "loss": 1.0234,
      "step": 4306
    },
    {
      "epoch": 0.6434120107559008,
      "grad_norm": 0.154793381690979,
      "learning_rate": 5.660557660672756e-05,
      "loss": 0.5983,
      "step": 4307
    },
    {
      "epoch": 0.6435613982671049,
      "grad_norm": 0.3678831160068512,
      "learning_rate": 5.6563235715647264e-05,
      "loss": 0.8873,
      "step": 4308
    },
    {
      "epoch": 0.643710785778309,
      "grad_norm": 0.1949298083782196,
      "learning_rate": 5.65209044204362e-05,
      "loss": 0.5051,
      "step": 4309
    },
    {
      "epoch": 0.643860173289513,
      "grad_norm": 0.15958526730537415,
      "learning_rate": 5.647858273044602e-05,
      "loss": 0.6281,
      "step": 4310
    },
    {
      "epoch": 0.644009560800717,
      "grad_norm": 0.1689438372850418,
      "learning_rate": 5.6436270655026304e-05,
      "loss": 0.6853,
      "step": 4311
    },
    {
      "epoch": 0.6441589483119211,
      "grad_norm": 0.3383398950099945,
      "learning_rate": 5.639396820352436e-05,
      "loss": 0.8766,
      "step": 4312
    },
    {
      "epoch": 0.6443083358231252,
      "grad_norm": 0.2607225775718689,
      "learning_rate": 5.635167538528554e-05,
      "loss": 1.2208,
      "step": 4313
    },
    {
      "epoch": 0.6444577233343293,
      "grad_norm": 0.19487948715686798,
      "learning_rate": 5.6309392209652924e-05,
      "loss": 0.7021,
      "step": 4314
    },
    {
      "epoch": 0.6446071108455333,
      "grad_norm": 0.13561616837978363,
      "learning_rate": 5.6267118685967645e-05,
      "loss": 0.8826,
      "step": 4315
    },
    {
      "epoch": 0.6447564983567374,
      "grad_norm": 0.1572197675704956,
      "learning_rate": 5.622485482356854e-05,
      "loss": 0.6118,
      "step": 4316
    },
    {
      "epoch": 0.6449058858679414,
      "grad_norm": 0.2165667563676834,
      "learning_rate": 5.61826006317924e-05,
      "loss": 0.6418,
      "step": 4317
    },
    {
      "epoch": 0.6450552733791455,
      "grad_norm": 0.2371535301208496,
      "learning_rate": 5.614035611997378e-05,
      "loss": 1.0856,
      "step": 4318
    },
    {
      "epoch": 0.6452046608903496,
      "grad_norm": 0.19905902445316315,
      "learning_rate": 5.609812129744518e-05,
      "loss": 0.8566,
      "step": 4319
    },
    {
      "epoch": 0.6453540484015536,
      "grad_norm": 0.21877242624759674,
      "learning_rate": 5.605589617353697e-05,
      "loss": 0.7791,
      "step": 4320
    },
    {
      "epoch": 0.6455034359127577,
      "grad_norm": 0.17051340639591217,
      "learning_rate": 5.601368075757734e-05,
      "loss": 0.6315,
      "step": 4321
    },
    {
      "epoch": 0.6456528234239618,
      "grad_norm": 0.160380020737648,
      "learning_rate": 5.597147505889233e-05,
      "loss": 0.8897,
      "step": 4322
    },
    {
      "epoch": 0.6458022109351658,
      "grad_norm": 0.13271751999855042,
      "learning_rate": 5.5929279086805884e-05,
      "loss": 0.6365,
      "step": 4323
    },
    {
      "epoch": 0.6459515984463698,
      "grad_norm": 0.13175153732299805,
      "learning_rate": 5.588709285063971e-05,
      "loss": 0.5973,
      "step": 4324
    },
    {
      "epoch": 0.6461009859575739,
      "grad_norm": 0.1766090989112854,
      "learning_rate": 5.584491635971343e-05,
      "loss": 0.8247,
      "step": 4325
    },
    {
      "epoch": 0.646250373468778,
      "grad_norm": 0.17424088716506958,
      "learning_rate": 5.580274962334451e-05,
      "loss": 0.595,
      "step": 4326
    },
    {
      "epoch": 0.6463997609799821,
      "grad_norm": 0.14612527191638947,
      "learning_rate": 5.576059265084823e-05,
      "loss": 0.5801,
      "step": 4327
    },
    {
      "epoch": 0.6465491484911862,
      "grad_norm": 0.21146363019943237,
      "learning_rate": 5.571844545153777e-05,
      "loss": 0.5388,
      "step": 4328
    },
    {
      "epoch": 0.6466985360023902,
      "grad_norm": 0.35164639353752136,
      "learning_rate": 5.567630803472408e-05,
      "loss": 0.7727,
      "step": 4329
    },
    {
      "epoch": 0.6468479235135942,
      "grad_norm": 0.27677881717681885,
      "learning_rate": 5.5634180409716e-05,
      "loss": 0.6973,
      "step": 4330
    },
    {
      "epoch": 0.6469973110247983,
      "grad_norm": 0.13974158465862274,
      "learning_rate": 5.559206258582019e-05,
      "loss": 0.7346,
      "step": 4331
    },
    {
      "epoch": 0.6471466985360024,
      "grad_norm": 0.12208545953035355,
      "learning_rate": 5.5549954572341134e-05,
      "loss": 0.5683,
      "step": 4332
    },
    {
      "epoch": 0.6472960860472065,
      "grad_norm": 0.28291261196136475,
      "learning_rate": 5.550785637858117e-05,
      "loss": 1.0898,
      "step": 4333
    },
    {
      "epoch": 0.6474454735584105,
      "grad_norm": 0.12974488735198975,
      "learning_rate": 5.546576801384045e-05,
      "loss": 0.7323,
      "step": 4334
    },
    {
      "epoch": 0.6475948610696146,
      "grad_norm": 0.11376846581697464,
      "learning_rate": 5.5423689487416964e-05,
      "loss": 0.6046,
      "step": 4335
    },
    {
      "epoch": 0.6477442485808187,
      "grad_norm": 0.13923761248588562,
      "learning_rate": 5.538162080860655e-05,
      "loss": 0.5531,
      "step": 4336
    },
    {
      "epoch": 0.6478936360920227,
      "grad_norm": 0.19857728481292725,
      "learning_rate": 5.533956198670278e-05,
      "loss": 0.6154,
      "step": 4337
    },
    {
      "epoch": 0.6480430236032267,
      "grad_norm": 0.13640432059764862,
      "learning_rate": 5.529751303099717e-05,
      "loss": 0.776,
      "step": 4338
    },
    {
      "epoch": 0.6481924111144308,
      "grad_norm": 0.23706190288066864,
      "learning_rate": 5.525547395077897e-05,
      "loss": 0.682,
      "step": 4339
    },
    {
      "epoch": 0.6483417986256349,
      "grad_norm": 0.20699051022529602,
      "learning_rate": 5.52134447553353e-05,
      "loss": 0.483,
      "step": 4340
    },
    {
      "epoch": 0.648491186136839,
      "grad_norm": 0.17999127507209778,
      "learning_rate": 5.5171425453951045e-05,
      "loss": 0.7038,
      "step": 4341
    },
    {
      "epoch": 0.6486405736480431,
      "grad_norm": 0.20031103491783142,
      "learning_rate": 5.512941605590898e-05,
      "loss": 0.67,
      "step": 4342
    },
    {
      "epoch": 0.6487899611592471,
      "grad_norm": 0.25791049003601074,
      "learning_rate": 5.5087416570489634e-05,
      "loss": 0.6574,
      "step": 4343
    },
    {
      "epoch": 0.6489393486704511,
      "grad_norm": 0.1533007174730301,
      "learning_rate": 5.5045427006971325e-05,
      "loss": 1.017,
      "step": 4344
    },
    {
      "epoch": 0.6490887361816552,
      "grad_norm": 0.1280319094657898,
      "learning_rate": 5.5003447374630236e-05,
      "loss": 0.7176,
      "step": 4345
    },
    {
      "epoch": 0.6492381236928593,
      "grad_norm": 0.14068573713302612,
      "learning_rate": 5.4961477682740325e-05,
      "loss": 0.5841,
      "step": 4346
    },
    {
      "epoch": 0.6493875112040634,
      "grad_norm": 0.4040509760379791,
      "learning_rate": 5.491951794057337e-05,
      "loss": 0.9508,
      "step": 4347
    },
    {
      "epoch": 0.6495368987152674,
      "grad_norm": 0.25920066237449646,
      "learning_rate": 5.487756815739894e-05,
      "loss": 0.6375,
      "step": 4348
    },
    {
      "epoch": 0.6496862862264715,
      "grad_norm": 0.160513773560524,
      "learning_rate": 5.48356283424844e-05,
      "loss": 0.5801,
      "step": 4349
    },
    {
      "epoch": 0.6498356737376755,
      "grad_norm": 0.16294503211975098,
      "learning_rate": 5.4793698505094926e-05,
      "loss": 0.6725,
      "step": 4350
    },
    {
      "epoch": 0.6499850612488796,
      "grad_norm": 0.12700767815113068,
      "learning_rate": 5.475177865449349e-05,
      "loss": 0.7172,
      "step": 4351
    },
    {
      "epoch": 0.6501344487600836,
      "grad_norm": 3.2553062438964844,
      "learning_rate": 5.4709868799940845e-05,
      "loss": 2.5579,
      "step": 4352
    },
    {
      "epoch": 0.6502838362712877,
      "grad_norm": 0.1973828673362732,
      "learning_rate": 5.4667968950695534e-05,
      "loss": 0.7354,
      "step": 4353
    },
    {
      "epoch": 0.6504332237824918,
      "grad_norm": 0.1930742710828781,
      "learning_rate": 5.4626079116013906e-05,
      "loss": 0.7306,
      "step": 4354
    },
    {
      "epoch": 0.6505826112936959,
      "grad_norm": 0.13104519248008728,
      "learning_rate": 5.4584199305150086e-05,
      "loss": 0.4675,
      "step": 4355
    },
    {
      "epoch": 0.6507319988049,
      "grad_norm": 0.14616692066192627,
      "learning_rate": 5.4542329527356025e-05,
      "loss": 0.6896,
      "step": 4356
    },
    {
      "epoch": 0.6508813863161039,
      "grad_norm": 0.15907815098762512,
      "learning_rate": 5.450046979188136e-05,
      "loss": 0.69,
      "step": 4357
    },
    {
      "epoch": 0.651030773827308,
      "grad_norm": 0.2896324694156647,
      "learning_rate": 5.445862010797358e-05,
      "loss": 1.2466,
      "step": 4358
    },
    {
      "epoch": 0.6511801613385121,
      "grad_norm": 0.13103102147579193,
      "learning_rate": 5.441678048487795e-05,
      "loss": 0.569,
      "step": 4359
    },
    {
      "epoch": 0.6513295488497162,
      "grad_norm": 0.13212913274765015,
      "learning_rate": 5.437495093183753e-05,
      "loss": 0.7324,
      "step": 4360
    },
    {
      "epoch": 0.6514789363609202,
      "grad_norm": 0.14805541932582855,
      "learning_rate": 5.43331314580931e-05,
      "loss": 0.5414,
      "step": 4361
    },
    {
      "epoch": 0.6516283238721243,
      "grad_norm": 0.16160765290260315,
      "learning_rate": 5.42913220728833e-05,
      "loss": 0.8668,
      "step": 4362
    },
    {
      "epoch": 0.6517777113833284,
      "grad_norm": 0.13007156550884247,
      "learning_rate": 5.42495227854444e-05,
      "loss": 0.5814,
      "step": 4363
    },
    {
      "epoch": 0.6519270988945324,
      "grad_norm": 0.17280595004558563,
      "learning_rate": 5.420773360501057e-05,
      "loss": 0.5648,
      "step": 4364
    },
    {
      "epoch": 0.6520764864057365,
      "grad_norm": 0.15236331522464752,
      "learning_rate": 5.41659545408137e-05,
      "loss": 0.5797,
      "step": 4365
    },
    {
      "epoch": 0.6522258739169405,
      "grad_norm": 0.2440706342458725,
      "learning_rate": 5.412418560208343e-05,
      "loss": 0.7628,
      "step": 4366
    },
    {
      "epoch": 0.6523752614281446,
      "grad_norm": 0.14022861421108246,
      "learning_rate": 5.408242679804722e-05,
      "loss": 0.7207,
      "step": 4367
    },
    {
      "epoch": 0.6525246489393487,
      "grad_norm": 0.15127135813236237,
      "learning_rate": 5.4040678137930214e-05,
      "loss": 0.8349,
      "step": 4368
    },
    {
      "epoch": 0.6526740364505528,
      "grad_norm": 0.16863341629505157,
      "learning_rate": 5.3998939630955367e-05,
      "loss": 0.8681,
      "step": 4369
    },
    {
      "epoch": 0.6528234239617567,
      "grad_norm": 0.14937616884708405,
      "learning_rate": 5.395721128634338e-05,
      "loss": 0.7884,
      "step": 4370
    },
    {
      "epoch": 0.6529728114729608,
      "grad_norm": 0.11547055840492249,
      "learning_rate": 5.3915493113312696e-05,
      "loss": 0.545,
      "step": 4371
    },
    {
      "epoch": 0.6531221989841649,
      "grad_norm": 0.17821939289569855,
      "learning_rate": 5.387378512107952e-05,
      "loss": 0.5403,
      "step": 4372
    },
    {
      "epoch": 0.653271586495369,
      "grad_norm": 0.11405037343502045,
      "learning_rate": 5.3832087318857815e-05,
      "loss": 0.6533,
      "step": 4373
    },
    {
      "epoch": 0.6534209740065731,
      "grad_norm": 0.16139666736125946,
      "learning_rate": 5.379039971585929e-05,
      "loss": 0.727,
      "step": 4374
    },
    {
      "epoch": 0.6535703615177771,
      "grad_norm": 0.14676588773727417,
      "learning_rate": 5.374872232129342e-05,
      "loss": 0.8044,
      "step": 4375
    },
    {
      "epoch": 0.6537197490289812,
      "grad_norm": 0.20285168290138245,
      "learning_rate": 5.3707055144367336e-05,
      "loss": 0.9481,
      "step": 4376
    },
    {
      "epoch": 0.6538691365401852,
      "grad_norm": 0.2261962592601776,
      "learning_rate": 5.366539819428601e-05,
      "loss": 0.8913,
      "step": 4377
    },
    {
      "epoch": 0.6540185240513893,
      "grad_norm": 0.21459020674228668,
      "learning_rate": 5.362375148025213e-05,
      "loss": 0.8753,
      "step": 4378
    },
    {
      "epoch": 0.6541679115625934,
      "grad_norm": 0.25149455666542053,
      "learning_rate": 5.3582115011466106e-05,
      "loss": 0.9388,
      "step": 4379
    },
    {
      "epoch": 0.6543172990737974,
      "grad_norm": 0.17654316127300262,
      "learning_rate": 5.35404887971261e-05,
      "loss": 0.7448,
      "step": 4380
    },
    {
      "epoch": 0.6544666865850015,
      "grad_norm": 0.14424292743206024,
      "learning_rate": 5.3498872846428e-05,
      "loss": 0.4914,
      "step": 4381
    },
    {
      "epoch": 0.6546160740962056,
      "grad_norm": 0.15362778306007385,
      "learning_rate": 5.345726716856545e-05,
      "loss": 0.6981,
      "step": 4382
    },
    {
      "epoch": 0.6547654616074097,
      "grad_norm": 0.21319247782230377,
      "learning_rate": 5.341567177272977e-05,
      "loss": 1.1037,
      "step": 4383
    },
    {
      "epoch": 0.6549148491186136,
      "grad_norm": 0.14228132367134094,
      "learning_rate": 5.3374086668110034e-05,
      "loss": 0.7545,
      "step": 4384
    },
    {
      "epoch": 0.6550642366298177,
      "grad_norm": 0.1360132098197937,
      "learning_rate": 5.333251186389308e-05,
      "loss": 0.7362,
      "step": 4385
    },
    {
      "epoch": 0.6552136241410218,
      "grad_norm": 0.22222623229026794,
      "learning_rate": 5.329094736926342e-05,
      "loss": 0.9088,
      "step": 4386
    },
    {
      "epoch": 0.6553630116522259,
      "grad_norm": 0.24109241366386414,
      "learning_rate": 5.324939319340328e-05,
      "loss": 0.7934,
      "step": 4387
    },
    {
      "epoch": 0.65551239916343,
      "grad_norm": 0.15872132778167725,
      "learning_rate": 5.320784934549277e-05,
      "loss": 0.5394,
      "step": 4388
    },
    {
      "epoch": 0.655661786674634,
      "grad_norm": 0.22934986650943756,
      "learning_rate": 5.316631583470943e-05,
      "loss": 1.2354,
      "step": 4389
    },
    {
      "epoch": 0.6558111741858381,
      "grad_norm": 0.12874899804592133,
      "learning_rate": 5.312479267022874e-05,
      "loss": 0.8693,
      "step": 4390
    },
    {
      "epoch": 0.6559605616970421,
      "grad_norm": 0.15565863251686096,
      "learning_rate": 5.308327986122381e-05,
      "loss": 0.612,
      "step": 4391
    },
    {
      "epoch": 0.6561099492082462,
      "grad_norm": 0.16037599742412567,
      "learning_rate": 5.304177741686549e-05,
      "loss": 0.701,
      "step": 4392
    },
    {
      "epoch": 0.6562593367194502,
      "grad_norm": 0.13753099739551544,
      "learning_rate": 5.300028534632231e-05,
      "loss": 0.5664,
      "step": 4393
    },
    {
      "epoch": 0.6564087242306543,
      "grad_norm": 0.4384874403476715,
      "learning_rate": 5.295880365876058e-05,
      "loss": 1.0331,
      "step": 4394
    },
    {
      "epoch": 0.6565581117418584,
      "grad_norm": 0.12920920550823212,
      "learning_rate": 5.2917332363344164e-05,
      "loss": 0.8081,
      "step": 4395
    },
    {
      "epoch": 0.6567074992530625,
      "grad_norm": 0.18578462302684784,
      "learning_rate": 5.2875871469234786e-05,
      "loss": 0.7401,
      "step": 4396
    },
    {
      "epoch": 0.6568568867642665,
      "grad_norm": 0.13123682141304016,
      "learning_rate": 5.2834420985591815e-05,
      "loss": 0.8012,
      "step": 4397
    },
    {
      "epoch": 0.6570062742754705,
      "grad_norm": 0.16064737737178802,
      "learning_rate": 5.27929809215723e-05,
      "loss": 0.5681,
      "step": 4398
    },
    {
      "epoch": 0.6571556617866746,
      "grad_norm": 0.27724146842956543,
      "learning_rate": 5.275155128633102e-05,
      "loss": 0.746,
      "step": 4399
    },
    {
      "epoch": 0.6573050492978787,
      "grad_norm": 0.23610037565231323,
      "learning_rate": 5.271013208902045e-05,
      "loss": 0.8909,
      "step": 4400
    },
    {
      "epoch": 0.6574544368090828,
      "grad_norm": 0.14107497036457062,
      "learning_rate": 5.266872333879077e-05,
      "loss": 0.801,
      "step": 4401
    },
    {
      "epoch": 0.6576038243202869,
      "grad_norm": 0.15323635935783386,
      "learning_rate": 5.262732504478975e-05,
      "loss": 0.5591,
      "step": 4402
    },
    {
      "epoch": 0.6577532118314909,
      "grad_norm": 0.1836881786584854,
      "learning_rate": 5.258593721616294e-05,
      "loss": 0.9078,
      "step": 4403
    },
    {
      "epoch": 0.6579025993426949,
      "grad_norm": 0.10958611220121384,
      "learning_rate": 5.254455986205362e-05,
      "loss": 0.5929,
      "step": 4404
    },
    {
      "epoch": 0.658051986853899,
      "grad_norm": 0.133941650390625,
      "learning_rate": 5.2503192991602715e-05,
      "loss": 0.9424,
      "step": 4405
    },
    {
      "epoch": 0.6582013743651031,
      "grad_norm": 0.14269699156284332,
      "learning_rate": 5.246183661394876e-05,
      "loss": 0.5525,
      "step": 4406
    },
    {
      "epoch": 0.6583507618763071,
      "grad_norm": 0.1848260760307312,
      "learning_rate": 5.24204907382281e-05,
      "loss": 0.7575,
      "step": 4407
    },
    {
      "epoch": 0.6585001493875112,
      "grad_norm": 0.14129601418972015,
      "learning_rate": 5.237915537357463e-05,
      "loss": 0.6241,
      "step": 4408
    },
    {
      "epoch": 0.6586495368987153,
      "grad_norm": 0.1946703940629959,
      "learning_rate": 5.2337830529120016e-05,
      "loss": 0.6364,
      "step": 4409
    },
    {
      "epoch": 0.6587989244099194,
      "grad_norm": 0.2839517295360565,
      "learning_rate": 5.2296516213993564e-05,
      "loss": 0.7595,
      "step": 4410
    },
    {
      "epoch": 0.6589483119211234,
      "grad_norm": 0.1412980705499649,
      "learning_rate": 5.225521243732226e-05,
      "loss": 0.4664,
      "step": 4411
    },
    {
      "epoch": 0.6590976994323274,
      "grad_norm": 0.15037702023983002,
      "learning_rate": 5.2213919208230775e-05,
      "loss": 0.6259,
      "step": 4412
    },
    {
      "epoch": 0.6592470869435315,
      "grad_norm": 0.2382907122373581,
      "learning_rate": 5.217263653584141e-05,
      "loss": 1.0086,
      "step": 4413
    },
    {
      "epoch": 0.6593964744547356,
      "grad_norm": 0.1397772878408432,
      "learning_rate": 5.2131364429274246e-05,
      "loss": 0.5348,
      "step": 4414
    },
    {
      "epoch": 0.6595458619659397,
      "grad_norm": 0.16269522905349731,
      "learning_rate": 5.209010289764682e-05,
      "loss": 0.5219,
      "step": 4415
    },
    {
      "epoch": 0.6596952494771438,
      "grad_norm": 0.12522614002227783,
      "learning_rate": 5.204885195007453e-05,
      "loss": 0.5255,
      "step": 4416
    },
    {
      "epoch": 0.6598446369883477,
      "grad_norm": 0.16688966751098633,
      "learning_rate": 5.200761159567035e-05,
      "loss": 0.5055,
      "step": 4417
    },
    {
      "epoch": 0.6599940244995518,
      "grad_norm": 0.13111922144889832,
      "learning_rate": 5.196638184354492e-05,
      "loss": 0.4997,
      "step": 4418
    },
    {
      "epoch": 0.6601434120107559,
      "grad_norm": 0.22333525121212006,
      "learning_rate": 5.1925162702806565e-05,
      "loss": 0.5295,
      "step": 4419
    },
    {
      "epoch": 0.66029279952196,
      "grad_norm": 0.16879773139953613,
      "learning_rate": 5.188395418256122e-05,
      "loss": 0.6754,
      "step": 4420
    },
    {
      "epoch": 0.660442187033164,
      "grad_norm": 0.09198059141635895,
      "learning_rate": 5.184275629191253e-05,
      "loss": 0.3471,
      "step": 4421
    },
    {
      "epoch": 0.6605915745443681,
      "grad_norm": 0.1221717819571495,
      "learning_rate": 5.180156903996174e-05,
      "loss": 0.6655,
      "step": 4422
    },
    {
      "epoch": 0.6607409620555722,
      "grad_norm": 0.19353681802749634,
      "learning_rate": 5.176039243580777e-05,
      "loss": 0.6705,
      "step": 4423
    },
    {
      "epoch": 0.6608903495667762,
      "grad_norm": 0.22404973208904266,
      "learning_rate": 5.171922648854719e-05,
      "loss": 0.9713,
      "step": 4424
    },
    {
      "epoch": 0.6610397370779802,
      "grad_norm": 0.30907249450683594,
      "learning_rate": 5.167807120727422e-05,
      "loss": 1.2869,
      "step": 4425
    },
    {
      "epoch": 0.6611891245891843,
      "grad_norm": 0.19484591484069824,
      "learning_rate": 5.163692660108068e-05,
      "loss": 0.7894,
      "step": 4426
    },
    {
      "epoch": 0.6613385121003884,
      "grad_norm": 0.12278755754232407,
      "learning_rate": 5.159579267905613e-05,
      "loss": 0.8016,
      "step": 4427
    },
    {
      "epoch": 0.6614878996115925,
      "grad_norm": 0.2530829608440399,
      "learning_rate": 5.155466945028762e-05,
      "loss": 0.965,
      "step": 4428
    },
    {
      "epoch": 0.6616372871227966,
      "grad_norm": 0.1466372162103653,
      "learning_rate": 5.1513556923859975e-05,
      "loss": 0.5226,
      "step": 4429
    },
    {
      "epoch": 0.6617866746340006,
      "grad_norm": 0.1391781121492386,
      "learning_rate": 5.147245510885557e-05,
      "loss": 0.8242,
      "step": 4430
    },
    {
      "epoch": 0.6619360621452046,
      "grad_norm": 0.1526176780462265,
      "learning_rate": 5.143136401435449e-05,
      "loss": 0.7541,
      "step": 4431
    },
    {
      "epoch": 0.6620854496564087,
      "grad_norm": 0.1452173888683319,
      "learning_rate": 5.1390283649434365e-05,
      "loss": 0.5751,
      "step": 4432
    },
    {
      "epoch": 0.6622348371676128,
      "grad_norm": 0.13828393816947937,
      "learning_rate": 5.1349214023170564e-05,
      "loss": 0.7391,
      "step": 4433
    },
    {
      "epoch": 0.6623842246788169,
      "grad_norm": 0.15598390996456146,
      "learning_rate": 5.130815514463595e-05,
      "loss": 0.7735,
      "step": 4434
    },
    {
      "epoch": 0.6625336121900209,
      "grad_norm": 0.19063301384449005,
      "learning_rate": 5.1267107022901096e-05,
      "loss": 0.9643,
      "step": 4435
    },
    {
      "epoch": 0.662682999701225,
      "grad_norm": 0.1422838717699051,
      "learning_rate": 5.122606966703418e-05,
      "loss": 0.8309,
      "step": 4436
    },
    {
      "epoch": 0.662832387212429,
      "grad_norm": 0.20462340116500854,
      "learning_rate": 5.118504308610104e-05,
      "loss": 0.972,
      "step": 4437
    },
    {
      "epoch": 0.6629817747236331,
      "grad_norm": 0.296310693025589,
      "learning_rate": 5.1144027289165045e-05,
      "loss": 0.8646,
      "step": 4438
    },
    {
      "epoch": 0.6631311622348371,
      "grad_norm": 0.27805647253990173,
      "learning_rate": 5.110302228528727e-05,
      "loss": 0.8684,
      "step": 4439
    },
    {
      "epoch": 0.6632805497460412,
      "grad_norm": 0.13401229679584503,
      "learning_rate": 5.1062028083526356e-05,
      "loss": 0.9577,
      "step": 4440
    },
    {
      "epoch": 0.6634299372572453,
      "grad_norm": 0.11802790313959122,
      "learning_rate": 5.102104469293858e-05,
      "loss": 0.4246,
      "step": 4441
    },
    {
      "epoch": 0.6635793247684494,
      "grad_norm": 0.2153034210205078,
      "learning_rate": 5.098007212257782e-05,
      "loss": 0.9872,
      "step": 4442
    },
    {
      "epoch": 0.6637287122796535,
      "grad_norm": 0.1380615234375,
      "learning_rate": 5.093911038149556e-05,
      "loss": 0.8298,
      "step": 4443
    },
    {
      "epoch": 0.6638780997908574,
      "grad_norm": 0.14291872084140778,
      "learning_rate": 5.08981594787409e-05,
      "loss": 0.9165,
      "step": 4444
    },
    {
      "epoch": 0.6640274873020615,
      "grad_norm": 0.13186267018318176,
      "learning_rate": 5.085721942336055e-05,
      "loss": 0.6629,
      "step": 4445
    },
    {
      "epoch": 0.6641768748132656,
      "grad_norm": 0.147129088640213,
      "learning_rate": 5.0816290224398844e-05,
      "loss": 0.8709,
      "step": 4446
    },
    {
      "epoch": 0.6643262623244697,
      "grad_norm": 0.13977491855621338,
      "learning_rate": 5.0775371890897615e-05,
      "loss": 0.592,
      "step": 4447
    },
    {
      "epoch": 0.6644756498356738,
      "grad_norm": 0.12048250436782837,
      "learning_rate": 5.073446443189642e-05,
      "loss": 0.7133,
      "step": 4448
    },
    {
      "epoch": 0.6646250373468778,
      "grad_norm": 0.1864444762468338,
      "learning_rate": 5.069356785643236e-05,
      "loss": 1.0664,
      "step": 4449
    },
    {
      "epoch": 0.6647744248580819,
      "grad_norm": 0.12989215552806854,
      "learning_rate": 5.0652682173540125e-05,
      "loss": 0.7196,
      "step": 4450
    },
    {
      "epoch": 0.6649238123692859,
      "grad_norm": 0.15337535738945007,
      "learning_rate": 5.061180739225202e-05,
      "loss": 0.6556,
      "step": 4451
    },
    {
      "epoch": 0.66507319988049,
      "grad_norm": 0.135578915476799,
      "learning_rate": 5.0570943521597934e-05,
      "loss": 0.7241,
      "step": 4452
    },
    {
      "epoch": 0.665222587391694,
      "grad_norm": 0.20100988447666168,
      "learning_rate": 5.053009057060538e-05,
      "loss": 0.5912,
      "step": 4453
    },
    {
      "epoch": 0.6653719749028981,
      "grad_norm": 0.15946964919567108,
      "learning_rate": 5.048924854829934e-05,
      "loss": 0.5722,
      "step": 4454
    },
    {
      "epoch": 0.6655213624141022,
      "grad_norm": 0.12619760632514954,
      "learning_rate": 5.0448417463702505e-05,
      "loss": 0.6504,
      "step": 4455
    },
    {
      "epoch": 0.6656707499253063,
      "grad_norm": 0.2949730157852173,
      "learning_rate": 5.040759732583511e-05,
      "loss": 0.6658,
      "step": 4456
    },
    {
      "epoch": 0.6658201374365104,
      "grad_norm": 0.13583524525165558,
      "learning_rate": 5.036678814371497e-05,
      "loss": 0.7316,
      "step": 4457
    },
    {
      "epoch": 0.6659695249477143,
      "grad_norm": 0.15064877271652222,
      "learning_rate": 5.032598992635748e-05,
      "loss": 0.8557,
      "step": 4458
    },
    {
      "epoch": 0.6661189124589184,
      "grad_norm": 0.12154415994882584,
      "learning_rate": 5.028520268277561e-05,
      "loss": 0.4996,
      "step": 4459
    },
    {
      "epoch": 0.6662682999701225,
      "grad_norm": 0.21336646378040314,
      "learning_rate": 5.024442642197991e-05,
      "loss": 1.1475,
      "step": 4460
    },
    {
      "epoch": 0.6664176874813266,
      "grad_norm": 0.1531991958618164,
      "learning_rate": 5.0203661152978496e-05,
      "loss": 0.521,
      "step": 4461
    },
    {
      "epoch": 0.6665670749925307,
      "grad_norm": 0.10438970476388931,
      "learning_rate": 5.016290688477707e-05,
      "loss": 0.4676,
      "step": 4462
    },
    {
      "epoch": 0.6667164625037347,
      "grad_norm": 0.20441101491451263,
      "learning_rate": 5.012216362637889e-05,
      "loss": 0.94,
      "step": 4463
    },
    {
      "epoch": 0.6668658500149387,
      "grad_norm": 0.21757328510284424,
      "learning_rate": 5.008143138678479e-05,
      "loss": 0.6193,
      "step": 4464
    },
    {
      "epoch": 0.6670152375261428,
      "grad_norm": 0.16931599378585815,
      "learning_rate": 5.004071017499321e-05,
      "loss": 0.9392,
      "step": 4465
    },
    {
      "epoch": 0.6671646250373469,
      "grad_norm": 0.2077966183423996,
      "learning_rate": 5.000000000000002e-05,
      "loss": 0.4509,
      "step": 4466
    },
    {
      "epoch": 0.6673140125485509,
      "grad_norm": 0.1529855579137802,
      "learning_rate": 4.9959300870798796e-05,
      "loss": 0.5744,
      "step": 4467
    },
    {
      "epoch": 0.667463400059755,
      "grad_norm": 0.13916553556919098,
      "learning_rate": 4.991861279638061e-05,
      "loss": 0.4172,
      "step": 4468
    },
    {
      "epoch": 0.6676127875709591,
      "grad_norm": 0.25168976187705994,
      "learning_rate": 4.987793578573411e-05,
      "loss": 0.9297,
      "step": 4469
    },
    {
      "epoch": 0.6677621750821632,
      "grad_norm": 0.1205325573682785,
      "learning_rate": 4.983726984784548e-05,
      "loss": 0.4897,
      "step": 4470
    },
    {
      "epoch": 0.6679115625933671,
      "grad_norm": 0.14898672699928284,
      "learning_rate": 4.979661499169848e-05,
      "loss": 0.796,
      "step": 4471
    },
    {
      "epoch": 0.6680609501045712,
      "grad_norm": 0.16903848946094513,
      "learning_rate": 4.975597122627445e-05,
      "loss": 0.36,
      "step": 4472
    },
    {
      "epoch": 0.6682103376157753,
      "grad_norm": 0.14433594048023224,
      "learning_rate": 4.9715338560552146e-05,
      "loss": 0.7662,
      "step": 4473
    },
    {
      "epoch": 0.6683597251269794,
      "grad_norm": 0.22180941700935364,
      "learning_rate": 4.9674717003508045e-05,
      "loss": 0.7335,
      "step": 4474
    },
    {
      "epoch": 0.6685091126381835,
      "grad_norm": 0.13435396552085876,
      "learning_rate": 4.963410656411606e-05,
      "loss": 0.899,
      "step": 4475
    },
    {
      "epoch": 0.6686585001493875,
      "grad_norm": 0.2958034574985504,
      "learning_rate": 4.959350725134765e-05,
      "loss": 0.6643,
      "step": 4476
    },
    {
      "epoch": 0.6688078876605916,
      "grad_norm": 0.3351992666721344,
      "learning_rate": 4.955291907417192e-05,
      "loss": 0.9904,
      "step": 4477
    },
    {
      "epoch": 0.6689572751717956,
      "grad_norm": 0.17713822424411774,
      "learning_rate": 4.951234204155545e-05,
      "loss": 0.5249,
      "step": 4478
    },
    {
      "epoch": 0.6691066626829997,
      "grad_norm": 0.10996868461370468,
      "learning_rate": 4.947177616246227e-05,
      "loss": 0.5438,
      "step": 4479
    },
    {
      "epoch": 0.6692560501942038,
      "grad_norm": 0.18839603662490845,
      "learning_rate": 4.943122144585406e-05,
      "loss": 1.13,
      "step": 4480
    },
    {
      "epoch": 0.6694054377054078,
      "grad_norm": 0.23967738449573517,
      "learning_rate": 4.9390677900689996e-05,
      "loss": 0.5882,
      "step": 4481
    },
    {
      "epoch": 0.6695548252166119,
      "grad_norm": 0.31552135944366455,
      "learning_rate": 4.9350145535926796e-05,
      "loss": 0.8318,
      "step": 4482
    },
    {
      "epoch": 0.669704212727816,
      "grad_norm": 0.19218727946281433,
      "learning_rate": 4.93096243605187e-05,
      "loss": 0.814,
      "step": 4483
    },
    {
      "epoch": 0.66985360023902,
      "grad_norm": 0.5546085238456726,
      "learning_rate": 4.9269114383417484e-05,
      "loss": 1.3436,
      "step": 4484
    },
    {
      "epoch": 0.670002987750224,
      "grad_norm": 0.13659346103668213,
      "learning_rate": 4.922861561357246e-05,
      "loss": 0.5671,
      "step": 4485
    },
    {
      "epoch": 0.6701523752614281,
      "grad_norm": 0.1545904576778412,
      "learning_rate": 4.9188128059930394e-05,
      "loss": 0.4794,
      "step": 4486
    },
    {
      "epoch": 0.6703017627726322,
      "grad_norm": 0.1288261115550995,
      "learning_rate": 4.914765173143566e-05,
      "loss": 0.5095,
      "step": 4487
    },
    {
      "epoch": 0.6704511502838363,
      "grad_norm": 0.3280554413795471,
      "learning_rate": 4.910718663703012e-05,
      "loss": 0.8769,
      "step": 4488
    },
    {
      "epoch": 0.6706005377950404,
      "grad_norm": 0.19561098515987396,
      "learning_rate": 4.906673278565315e-05,
      "loss": 0.7571,
      "step": 4489
    },
    {
      "epoch": 0.6707499253062444,
      "grad_norm": 0.22987867891788483,
      "learning_rate": 4.902629018624164e-05,
      "loss": 0.8129,
      "step": 4490
    },
    {
      "epoch": 0.6708993128174484,
      "grad_norm": 0.1844412386417389,
      "learning_rate": 4.898585884773006e-05,
      "loss": 0.9104,
      "step": 4491
    },
    {
      "epoch": 0.6710487003286525,
      "grad_norm": 0.15339171886444092,
      "learning_rate": 4.89454387790502e-05,
      "loss": 0.4642,
      "step": 4492
    },
    {
      "epoch": 0.6711980878398566,
      "grad_norm": 0.23105745017528534,
      "learning_rate": 4.890502998913162e-05,
      "loss": 0.8847,
      "step": 4493
    },
    {
      "epoch": 0.6713474753510607,
      "grad_norm": 0.3503510653972626,
      "learning_rate": 4.886463248690122e-05,
      "loss": 0.5983,
      "step": 4494
    },
    {
      "epoch": 0.6714968628622647,
      "grad_norm": 0.1930682361125946,
      "learning_rate": 4.882424628128344e-05,
      "loss": 0.7637,
      "step": 4495
    },
    {
      "epoch": 0.6716462503734688,
      "grad_norm": 0.15850479900836945,
      "learning_rate": 4.878387138120023e-05,
      "loss": 0.5656,
      "step": 4496
    },
    {
      "epoch": 0.6717956378846729,
      "grad_norm": 0.1405087262392044,
      "learning_rate": 4.8743507795571065e-05,
      "loss": 0.412,
      "step": 4497
    },
    {
      "epoch": 0.6719450253958769,
      "grad_norm": 0.1367122083902359,
      "learning_rate": 4.8703155533312925e-05,
      "loss": 0.7986,
      "step": 4498
    },
    {
      "epoch": 0.6720944129070809,
      "grad_norm": 0.13022398948669434,
      "learning_rate": 4.866281460334019e-05,
      "loss": 0.5323,
      "step": 4499
    },
    {
      "epoch": 0.672243800418285,
      "grad_norm": 0.33614295721054077,
      "learning_rate": 4.862248501456484e-05,
      "loss": 1.0431,
      "step": 4500
    },
    {
      "epoch": 0.6723931879294891,
      "grad_norm": 0.2007324993610382,
      "learning_rate": 4.858216677589634e-05,
      "loss": 0.6156,
      "step": 4501
    },
    {
      "epoch": 0.6725425754406932,
      "grad_norm": 0.12399091571569443,
      "learning_rate": 4.8541859896241616e-05,
      "loss": 0.6383,
      "step": 4502
    },
    {
      "epoch": 0.6726919629518973,
      "grad_norm": 0.1809137463569641,
      "learning_rate": 4.850156438450508e-05,
      "loss": 0.7692,
      "step": 4503
    },
    {
      "epoch": 0.6728413504631013,
      "grad_norm": 0.12796355783939362,
      "learning_rate": 4.846128024958872e-05,
      "loss": 0.6038,
      "step": 4504
    },
    {
      "epoch": 0.6729907379743053,
      "grad_norm": 0.1552809625864029,
      "learning_rate": 4.842100750039184e-05,
      "loss": 0.5506,
      "step": 4505
    },
    {
      "epoch": 0.6731401254855094,
      "grad_norm": 0.2010272592306137,
      "learning_rate": 4.838074614581137e-05,
      "loss": 0.8199,
      "step": 4506
    },
    {
      "epoch": 0.6732895129967135,
      "grad_norm": 0.26998692750930786,
      "learning_rate": 4.834049619474168e-05,
      "loss": 1.1083,
      "step": 4507
    },
    {
      "epoch": 0.6734389005079175,
      "grad_norm": 0.5911502242088318,
      "learning_rate": 4.830025765607463e-05,
      "loss": 1.3742,
      "step": 4508
    },
    {
      "epoch": 0.6735882880191216,
      "grad_norm": 0.15808479487895966,
      "learning_rate": 4.8260030538699555e-05,
      "loss": 0.4939,
      "step": 4509
    },
    {
      "epoch": 0.6737376755303257,
      "grad_norm": 0.14046266674995422,
      "learning_rate": 4.821981485150323e-05,
      "loss": 0.6917,
      "step": 4510
    },
    {
      "epoch": 0.6738870630415297,
      "grad_norm": 0.14021186530590057,
      "learning_rate": 4.817961060336997e-05,
      "loss": 0.4552,
      "step": 4511
    },
    {
      "epoch": 0.6740364505527338,
      "grad_norm": 0.15818122029304504,
      "learning_rate": 4.813941780318153e-05,
      "loss": 0.8561,
      "step": 4512
    },
    {
      "epoch": 0.6741858380639378,
      "grad_norm": 0.1356274038553238,
      "learning_rate": 4.809923645981712e-05,
      "loss": 0.348,
      "step": 4513
    },
    {
      "epoch": 0.6743352255751419,
      "grad_norm": 0.17271605134010315,
      "learning_rate": 4.805906658215344e-05,
      "loss": 0.8597,
      "step": 4514
    },
    {
      "epoch": 0.674484613086346,
      "grad_norm": 0.1019844338297844,
      "learning_rate": 4.801890817906466e-05,
      "loss": 0.4226,
      "step": 4515
    },
    {
      "epoch": 0.6746340005975501,
      "grad_norm": 0.13891182839870453,
      "learning_rate": 4.797876125942239e-05,
      "loss": 0.6635,
      "step": 4516
    },
    {
      "epoch": 0.6747833881087542,
      "grad_norm": 0.26426297426223755,
      "learning_rate": 4.7938625832095776e-05,
      "loss": 0.8882,
      "step": 4517
    },
    {
      "epoch": 0.6749327756199581,
      "grad_norm": 0.17758791148662567,
      "learning_rate": 4.789850190595129e-05,
      "loss": 0.6809,
      "step": 4518
    },
    {
      "epoch": 0.6750821631311622,
      "grad_norm": 0.3215659558773041,
      "learning_rate": 4.7858389489852976e-05,
      "loss": 0.9718,
      "step": 4519
    },
    {
      "epoch": 0.6752315506423663,
      "grad_norm": 0.20558084547519684,
      "learning_rate": 4.7818288592662295e-05,
      "loss": 1.0349,
      "step": 4520
    },
    {
      "epoch": 0.6753809381535704,
      "grad_norm": 0.1549457162618637,
      "learning_rate": 4.7778199223238196e-05,
      "loss": 0.6701,
      "step": 4521
    },
    {
      "epoch": 0.6755303256647744,
      "grad_norm": 0.17171570658683777,
      "learning_rate": 4.773812139043703e-05,
      "loss": 0.8164,
      "step": 4522
    },
    {
      "epoch": 0.6756797131759785,
      "grad_norm": 0.12353827059268951,
      "learning_rate": 4.769805510311267e-05,
      "loss": 0.6907,
      "step": 4523
    },
    {
      "epoch": 0.6758291006871826,
      "grad_norm": 0.1420106142759323,
      "learning_rate": 4.7658000370116315e-05,
      "loss": 0.6423,
      "step": 4524
    },
    {
      "epoch": 0.6759784881983866,
      "grad_norm": 0.34141021966934204,
      "learning_rate": 4.761795720029675e-05,
      "loss": 0.7609,
      "step": 4525
    },
    {
      "epoch": 0.6761278757095907,
      "grad_norm": 0.3230818808078766,
      "learning_rate": 4.757792560250012e-05,
      "loss": 0.9609,
      "step": 4526
    },
    {
      "epoch": 0.6762772632207947,
      "grad_norm": 0.330085813999176,
      "learning_rate": 4.753790558557004e-05,
      "loss": 1.0968,
      "step": 4527
    },
    {
      "epoch": 0.6764266507319988,
      "grad_norm": 0.1571723371744156,
      "learning_rate": 4.749789715834758e-05,
      "loss": 0.6914,
      "step": 4528
    },
    {
      "epoch": 0.6765760382432029,
      "grad_norm": 0.14600616693496704,
      "learning_rate": 4.7457900329671225e-05,
      "loss": 0.5531,
      "step": 4529
    },
    {
      "epoch": 0.676725425754407,
      "grad_norm": 0.19893884658813477,
      "learning_rate": 4.741791510837691e-05,
      "loss": 0.6681,
      "step": 4530
    },
    {
      "epoch": 0.6768748132656109,
      "grad_norm": 0.13083714246749878,
      "learning_rate": 4.7377941503298003e-05,
      "loss": 0.5976,
      "step": 4531
    },
    {
      "epoch": 0.677024200776815,
      "grad_norm": 0.17252083122730255,
      "learning_rate": 4.733797952326532e-05,
      "loss": 0.8144,
      "step": 4532
    },
    {
      "epoch": 0.6771735882880191,
      "grad_norm": 0.1382385939359665,
      "learning_rate": 4.7298029177107074e-05,
      "loss": 0.6146,
      "step": 4533
    },
    {
      "epoch": 0.6773229757992232,
      "grad_norm": 0.26033079624176025,
      "learning_rate": 4.725809047364894e-05,
      "loss": 1.0008,
      "step": 4534
    },
    {
      "epoch": 0.6774723633104273,
      "grad_norm": 0.11880939453840256,
      "learning_rate": 4.7218163421714015e-05,
      "loss": 0.7401,
      "step": 4535
    },
    {
      "epoch": 0.6776217508216313,
      "grad_norm": 0.153743177652359,
      "learning_rate": 4.717824803012284e-05,
      "loss": 0.9233,
      "step": 4536
    },
    {
      "epoch": 0.6777711383328354,
      "grad_norm": 0.1336580365896225,
      "learning_rate": 4.713834430769329e-05,
      "loss": 0.4855,
      "step": 4537
    },
    {
      "epoch": 0.6779205258440394,
      "grad_norm": 0.11718132346868515,
      "learning_rate": 4.7098452263240776e-05,
      "loss": 0.5023,
      "step": 4538
    },
    {
      "epoch": 0.6780699133552435,
      "grad_norm": 0.1642831563949585,
      "learning_rate": 4.705857190557807e-05,
      "loss": 0.7514,
      "step": 4539
    },
    {
      "epoch": 0.6782193008664475,
      "grad_norm": 0.15965546667575836,
      "learning_rate": 4.7018703243515385e-05,
      "loss": 0.7447,
      "step": 4540
    },
    {
      "epoch": 0.6783686883776516,
      "grad_norm": 0.21381601691246033,
      "learning_rate": 4.697884628586032e-05,
      "loss": 0.9145,
      "step": 4541
    },
    {
      "epoch": 0.6785180758888557,
      "grad_norm": 0.18727773427963257,
      "learning_rate": 4.693900104141794e-05,
      "loss": 0.4883,
      "step": 4542
    },
    {
      "epoch": 0.6786674634000598,
      "grad_norm": 0.18034525215625763,
      "learning_rate": 4.6899167518990696e-05,
      "loss": 0.7014,
      "step": 4543
    },
    {
      "epoch": 0.6788168509112639,
      "grad_norm": 0.5324631333351135,
      "learning_rate": 4.6859345727378387e-05,
      "loss": 1.3106,
      "step": 4544
    },
    {
      "epoch": 0.6789662384224678,
      "grad_norm": 0.1174982562661171,
      "learning_rate": 4.681953567537831e-05,
      "loss": 0.6262,
      "step": 4545
    },
    {
      "epoch": 0.6791156259336719,
      "grad_norm": 0.15351428091526031,
      "learning_rate": 4.6779737371785146e-05,
      "loss": 0.4436,
      "step": 4546
    },
    {
      "epoch": 0.679265013444876,
      "grad_norm": 0.14033591747283936,
      "learning_rate": 4.673995082539096e-05,
      "loss": 0.6703,
      "step": 4547
    },
    {
      "epoch": 0.6794144009560801,
      "grad_norm": 0.13986103236675262,
      "learning_rate": 4.670017604498522e-05,
      "loss": 0.5515,
      "step": 4548
    },
    {
      "epoch": 0.6795637884672842,
      "grad_norm": 0.14375442266464233,
      "learning_rate": 4.6660413039354825e-05,
      "loss": 0.8003,
      "step": 4549
    },
    {
      "epoch": 0.6797131759784882,
      "grad_norm": 0.13436606526374817,
      "learning_rate": 4.662066181728405e-05,
      "loss": 0.8013,
      "step": 4550
    },
    {
      "epoch": 0.6798625634896922,
      "grad_norm": 0.1385025680065155,
      "learning_rate": 4.658092238755456e-05,
      "loss": 0.8058,
      "step": 4551
    },
    {
      "epoch": 0.6800119510008963,
      "grad_norm": 0.15490847826004028,
      "learning_rate": 4.654119475894543e-05,
      "loss": 0.4885,
      "step": 4552
    },
    {
      "epoch": 0.6801613385121004,
      "grad_norm": 0.13009488582611084,
      "learning_rate": 4.650147894023311e-05,
      "loss": 0.7491,
      "step": 4553
    },
    {
      "epoch": 0.6803107260233044,
      "grad_norm": 0.2830427587032318,
      "learning_rate": 4.6461774940191486e-05,
      "loss": 1.4083,
      "step": 4554
    },
    {
      "epoch": 0.6804601135345085,
      "grad_norm": 0.24221758544445038,
      "learning_rate": 4.642208276759177e-05,
      "loss": 0.8809,
      "step": 4555
    },
    {
      "epoch": 0.6806095010457126,
      "grad_norm": 0.12590539455413818,
      "learning_rate": 4.638240243120263e-05,
      "loss": 0.7281,
      "step": 4556
    },
    {
      "epoch": 0.6807588885569167,
      "grad_norm": 0.11840452998876572,
      "learning_rate": 4.634273393979003e-05,
      "loss": 0.6034,
      "step": 4557
    },
    {
      "epoch": 0.6809082760681207,
      "grad_norm": 0.13331001996994019,
      "learning_rate": 4.630307730211737e-05,
      "loss": 0.9003,
      "step": 4558
    },
    {
      "epoch": 0.6810576635793247,
      "grad_norm": 0.14055581390857697,
      "learning_rate": 4.626343252694546e-05,
      "loss": 0.771,
      "step": 4559
    },
    {
      "epoch": 0.6812070510905288,
      "grad_norm": 0.152981698513031,
      "learning_rate": 4.622379962303246e-05,
      "loss": 0.492,
      "step": 4560
    },
    {
      "epoch": 0.6813564386017329,
      "grad_norm": 0.12766066193580627,
      "learning_rate": 4.618417859913388e-05,
      "loss": 0.6439,
      "step": 4561
    },
    {
      "epoch": 0.681505826112937,
      "grad_norm": 0.2109127640724182,
      "learning_rate": 4.6144569464002684e-05,
      "loss": 0.6997,
      "step": 4562
    },
    {
      "epoch": 0.681655213624141,
      "grad_norm": 0.29572319984436035,
      "learning_rate": 4.610497222638909e-05,
      "loss": 1.0485,
      "step": 4563
    },
    {
      "epoch": 0.6818046011353451,
      "grad_norm": 0.1985575556755066,
      "learning_rate": 4.606538689504077e-05,
      "loss": 0.8154,
      "step": 4564
    },
    {
      "epoch": 0.6819539886465491,
      "grad_norm": 0.23162192106246948,
      "learning_rate": 4.602581347870274e-05,
      "loss": 0.8142,
      "step": 4565
    },
    {
      "epoch": 0.6821033761577532,
      "grad_norm": 0.3184506297111511,
      "learning_rate": 4.5986251986117454e-05,
      "loss": 0.6865,
      "step": 4566
    },
    {
      "epoch": 0.6822527636689573,
      "grad_norm": 0.32151034474372864,
      "learning_rate": 4.594670242602464e-05,
      "loss": 0.8294,
      "step": 4567
    },
    {
      "epoch": 0.6824021511801613,
      "grad_norm": 0.153967022895813,
      "learning_rate": 4.590716480716141e-05,
      "loss": 0.6821,
      "step": 4568
    },
    {
      "epoch": 0.6825515386913654,
      "grad_norm": 0.185753732919693,
      "learning_rate": 4.586763913826231e-05,
      "loss": 0.8719,
      "step": 4569
    },
    {
      "epoch": 0.6827009262025695,
      "grad_norm": 0.20840121805667877,
      "learning_rate": 4.582812542805909e-05,
      "loss": 0.8819,
      "step": 4570
    },
    {
      "epoch": 0.6828503137137736,
      "grad_norm": 0.2339048981666565,
      "learning_rate": 4.578862368528101e-05,
      "loss": 0.5016,
      "step": 4571
    },
    {
      "epoch": 0.6829997012249776,
      "grad_norm": 0.19241803884506226,
      "learning_rate": 4.5749133918654607e-05,
      "loss": 0.6897,
      "step": 4572
    },
    {
      "epoch": 0.6831490887361816,
      "grad_norm": 0.14188329875469208,
      "learning_rate": 4.570965613690382e-05,
      "loss": 0.6347,
      "step": 4573
    },
    {
      "epoch": 0.6832984762473857,
      "grad_norm": 0.19061778485774994,
      "learning_rate": 4.567019034874992e-05,
      "loss": 0.6403,
      "step": 4574
    },
    {
      "epoch": 0.6834478637585898,
      "grad_norm": 0.1619061529636383,
      "learning_rate": 4.5630736562911536e-05,
      "loss": 0.7989,
      "step": 4575
    },
    {
      "epoch": 0.6835972512697939,
      "grad_norm": 0.1447991579771042,
      "learning_rate": 4.559129478810457e-05,
      "loss": 0.5794,
      "step": 4576
    },
    {
      "epoch": 0.683746638780998,
      "grad_norm": 0.3063763678073883,
      "learning_rate": 4.5551865033042373e-05,
      "loss": 1.1537,
      "step": 4577
    },
    {
      "epoch": 0.6838960262922019,
      "grad_norm": 0.15508949756622314,
      "learning_rate": 4.551244730643561e-05,
      "loss": 0.8632,
      "step": 4578
    },
    {
      "epoch": 0.684045413803406,
      "grad_norm": 0.14480948448181152,
      "learning_rate": 4.547304161699227e-05,
      "loss": 0.8403,
      "step": 4579
    },
    {
      "epoch": 0.6841948013146101,
      "grad_norm": 0.23806864023208618,
      "learning_rate": 4.5433647973417703e-05,
      "loss": 0.9249,
      "step": 4580
    },
    {
      "epoch": 0.6843441888258142,
      "grad_norm": 0.4203002154827118,
      "learning_rate": 4.5394266384414594e-05,
      "loss": 1.2495,
      "step": 4581
    },
    {
      "epoch": 0.6844935763370182,
      "grad_norm": 0.14766840636730194,
      "learning_rate": 4.535489685868294e-05,
      "loss": 0.8401,
      "step": 4582
    },
    {
      "epoch": 0.6846429638482223,
      "grad_norm": 0.40573596954345703,
      "learning_rate": 4.5315539404920106e-05,
      "loss": 1.4615,
      "step": 4583
    },
    {
      "epoch": 0.6847923513594264,
      "grad_norm": 0.11742778867483139,
      "learning_rate": 4.5276194031820774e-05,
      "loss": 0.8263,
      "step": 4584
    },
    {
      "epoch": 0.6849417388706304,
      "grad_norm": 0.2617970108985901,
      "learning_rate": 4.523686074807696e-05,
      "loss": 0.6695,
      "step": 4585
    },
    {
      "epoch": 0.6850911263818344,
      "grad_norm": 0.1355372816324234,
      "learning_rate": 4.5197539562378014e-05,
      "loss": 0.8588,
      "step": 4586
    },
    {
      "epoch": 0.6852405138930385,
      "grad_norm": 0.19781404733657837,
      "learning_rate": 4.51582304834106e-05,
      "loss": 0.6208,
      "step": 4587
    },
    {
      "epoch": 0.6853899014042426,
      "grad_norm": 0.13103139400482178,
      "learning_rate": 4.511893351985876e-05,
      "loss": 0.6902,
      "step": 4588
    },
    {
      "epoch": 0.6855392889154467,
      "grad_norm": 0.19514009356498718,
      "learning_rate": 4.5079648680403744e-05,
      "loss": 0.9294,
      "step": 4589
    },
    {
      "epoch": 0.6856886764266508,
      "grad_norm": 0.25215545296669006,
      "learning_rate": 4.5040375973724216e-05,
      "loss": 0.5194,
      "step": 4590
    },
    {
      "epoch": 0.6858380639378548,
      "grad_norm": 0.15621864795684814,
      "learning_rate": 4.500111540849617e-05,
      "loss": 0.7363,
      "step": 4591
    },
    {
      "epoch": 0.6859874514490588,
      "grad_norm": 0.1355849653482437,
      "learning_rate": 4.496186699339284e-05,
      "loss": 0.8574,
      "step": 4592
    },
    {
      "epoch": 0.6861368389602629,
      "grad_norm": 0.14042329788208008,
      "learning_rate": 4.492263073708487e-05,
      "loss": 0.8806,
      "step": 4593
    },
    {
      "epoch": 0.686286226471467,
      "grad_norm": 0.21296457946300507,
      "learning_rate": 4.488340664824019e-05,
      "loss": 0.6703,
      "step": 4594
    },
    {
      "epoch": 0.6864356139826711,
      "grad_norm": 0.1357143223285675,
      "learning_rate": 4.4844194735523937e-05,
      "loss": 0.7396,
      "step": 4595
    },
    {
      "epoch": 0.6865850014938751,
      "grad_norm": 0.15597866475582123,
      "learning_rate": 4.4804995007598694e-05,
      "loss": 0.5506,
      "step": 4596
    },
    {
      "epoch": 0.6867343890050792,
      "grad_norm": 1.5002219676971436,
      "learning_rate": 4.476580747312429e-05,
      "loss": 1.6704,
      "step": 4597
    },
    {
      "epoch": 0.6868837765162832,
      "grad_norm": 0.19931535422801971,
      "learning_rate": 4.472663214075788e-05,
      "loss": 0.5737,
      "step": 4598
    },
    {
      "epoch": 0.6870331640274873,
      "grad_norm": 0.11923231184482574,
      "learning_rate": 4.4687469019153926e-05,
      "loss": 0.6822,
      "step": 4599
    },
    {
      "epoch": 0.6871825515386913,
      "grad_norm": 0.15625718235969543,
      "learning_rate": 4.464831811696417e-05,
      "loss": 0.9639,
      "step": 4600
    },
    {
      "epoch": 0.6873319390498954,
      "grad_norm": 0.22136351466178894,
      "learning_rate": 4.460917944283766e-05,
      "loss": 0.6769,
      "step": 4601
    },
    {
      "epoch": 0.6874813265610995,
      "grad_norm": 0.17018884420394897,
      "learning_rate": 4.457005300542077e-05,
      "loss": 1.0884,
      "step": 4602
    },
    {
      "epoch": 0.6876307140723036,
      "grad_norm": 0.2000696063041687,
      "learning_rate": 4.4530938813357114e-05,
      "loss": 0.7887,
      "step": 4603
    },
    {
      "epoch": 0.6877801015835077,
      "grad_norm": 0.254130095243454,
      "learning_rate": 4.4491836875287674e-05,
      "loss": 1.1327,
      "step": 4604
    },
    {
      "epoch": 0.6879294890947116,
      "grad_norm": 0.14630961418151855,
      "learning_rate": 4.4452747199850665e-05,
      "loss": 0.8076,
      "step": 4605
    },
    {
      "epoch": 0.6880788766059157,
      "grad_norm": 0.14232397079467773,
      "learning_rate": 4.441366979568162e-05,
      "loss": 0.81,
      "step": 4606
    },
    {
      "epoch": 0.6882282641171198,
      "grad_norm": 0.1393236666917801,
      "learning_rate": 4.437460467141341e-05,
      "loss": 0.4635,
      "step": 4607
    },
    {
      "epoch": 0.6883776516283239,
      "grad_norm": 0.11958955973386765,
      "learning_rate": 4.433555183567604e-05,
      "loss": 0.5686,
      "step": 4608
    },
    {
      "epoch": 0.688527039139528,
      "grad_norm": 0.20627152919769287,
      "learning_rate": 4.429651129709696e-05,
      "loss": 0.9623,
      "step": 4609
    },
    {
      "epoch": 0.688676426650732,
      "grad_norm": 0.187338188290596,
      "learning_rate": 4.425748306430082e-05,
      "loss": 0.7901,
      "step": 4610
    },
    {
      "epoch": 0.6888258141619361,
      "grad_norm": 0.12486065924167633,
      "learning_rate": 4.421846714590959e-05,
      "loss": 0.7458,
      "step": 4611
    },
    {
      "epoch": 0.6889752016731401,
      "grad_norm": 0.12297758460044861,
      "learning_rate": 4.41794635505425e-05,
      "loss": 0.7187,
      "step": 4612
    },
    {
      "epoch": 0.6891245891843442,
      "grad_norm": 0.16105423867702484,
      "learning_rate": 4.414047228681607e-05,
      "loss": 0.5793,
      "step": 4613
    },
    {
      "epoch": 0.6892739766955482,
      "grad_norm": 0.1391417235136032,
      "learning_rate": 4.410149336334411e-05,
      "loss": 0.7027,
      "step": 4614
    },
    {
      "epoch": 0.6894233642067523,
      "grad_norm": 0.2927907705307007,
      "learning_rate": 4.4062526788737624e-05,
      "loss": 0.7344,
      "step": 4615
    },
    {
      "epoch": 0.6895727517179564,
      "grad_norm": 0.12467020750045776,
      "learning_rate": 4.4023572571604965e-05,
      "loss": 0.5281,
      "step": 4616
    },
    {
      "epoch": 0.6897221392291605,
      "grad_norm": 0.16625529527664185,
      "learning_rate": 4.398463072055176e-05,
      "loss": 0.3399,
      "step": 4617
    },
    {
      "epoch": 0.6898715267403646,
      "grad_norm": 0.11127953231334686,
      "learning_rate": 4.3945701244180847e-05,
      "loss": 0.6143,
      "step": 4618
    },
    {
      "epoch": 0.6900209142515685,
      "grad_norm": 0.19436371326446533,
      "learning_rate": 4.3906784151092396e-05,
      "loss": 0.9569,
      "step": 4619
    },
    {
      "epoch": 0.6901703017627726,
      "grad_norm": 0.13548624515533447,
      "learning_rate": 4.38678794498838e-05,
      "loss": 0.5024,
      "step": 4620
    },
    {
      "epoch": 0.6903196892739767,
      "grad_norm": 0.14494569599628448,
      "learning_rate": 4.3828987149149716e-05,
      "loss": 0.6327,
      "step": 4621
    },
    {
      "epoch": 0.6904690767851808,
      "grad_norm": 0.1488719880580902,
      "learning_rate": 4.379010725748206e-05,
      "loss": 0.6515,
      "step": 4622
    },
    {
      "epoch": 0.6906184642963848,
      "grad_norm": 0.3272029161453247,
      "learning_rate": 4.3751239783470045e-05,
      "loss": 0.7691,
      "step": 4623
    },
    {
      "epoch": 0.6907678518075889,
      "grad_norm": 0.17457172274589539,
      "learning_rate": 4.37123847357001e-05,
      "loss": 0.5782,
      "step": 4624
    },
    {
      "epoch": 0.6909172393187929,
      "grad_norm": 0.14382411539554596,
      "learning_rate": 4.367354212275591e-05,
      "loss": 0.6409,
      "step": 4625
    },
    {
      "epoch": 0.691066626829997,
      "grad_norm": 0.22149758040905,
      "learning_rate": 4.363471195321842e-05,
      "loss": 0.5463,
      "step": 4626
    },
    {
      "epoch": 0.6912160143412011,
      "grad_norm": 0.1555415540933609,
      "learning_rate": 4.359589423566589e-05,
      "loss": 0.7806,
      "step": 4627
    },
    {
      "epoch": 0.6913654018524051,
      "grad_norm": 0.1793929785490036,
      "learning_rate": 4.3557088978673676e-05,
      "loss": 0.489,
      "step": 4628
    },
    {
      "epoch": 0.6915147893636092,
      "grad_norm": 0.11405190825462341,
      "learning_rate": 4.351829619081451e-05,
      "loss": 0.5278,
      "step": 4629
    },
    {
      "epoch": 0.6916641768748133,
      "grad_norm": 0.13263565301895142,
      "learning_rate": 4.347951588065835e-05,
      "loss": 0.5413,
      "step": 4630
    },
    {
      "epoch": 0.6918135643860174,
      "grad_norm": 0.17374169826507568,
      "learning_rate": 4.3440748056772373e-05,
      "loss": 0.4566,
      "step": 4631
    },
    {
      "epoch": 0.6919629518972213,
      "grad_norm": 0.14342959225177765,
      "learning_rate": 4.3401992727721e-05,
      "loss": 0.5965,
      "step": 4632
    },
    {
      "epoch": 0.6921123394084254,
      "grad_norm": 0.1537477821111679,
      "learning_rate": 4.336324990206593e-05,
      "loss": 0.5232,
      "step": 4633
    },
    {
      "epoch": 0.6922617269196295,
      "grad_norm": 0.2437884509563446,
      "learning_rate": 4.3324519588366e-05,
      "loss": 1.0219,
      "step": 4634
    },
    {
      "epoch": 0.6924111144308336,
      "grad_norm": 0.1463748663663864,
      "learning_rate": 4.328580179517738e-05,
      "loss": 1.1065,
      "step": 4635
    },
    {
      "epoch": 0.6925605019420377,
      "grad_norm": 0.13441719114780426,
      "learning_rate": 4.324709653105346e-05,
      "loss": 0.8302,
      "step": 4636
    },
    {
      "epoch": 0.6927098894532417,
      "grad_norm": 0.20794415473937988,
      "learning_rate": 4.320840380454484e-05,
      "loss": 0.4683,
      "step": 4637
    },
    {
      "epoch": 0.6928592769644458,
      "grad_norm": 0.20713651180267334,
      "learning_rate": 4.316972362419931e-05,
      "loss": 1.0673,
      "step": 4638
    },
    {
      "epoch": 0.6930086644756498,
      "grad_norm": 0.12634721398353577,
      "learning_rate": 4.3131055998562006e-05,
      "loss": 0.8815,
      "step": 4639
    },
    {
      "epoch": 0.6931580519868539,
      "grad_norm": 0.14275509119033813,
      "learning_rate": 4.309240093617524e-05,
      "loss": 0.7345,
      "step": 4640
    },
    {
      "epoch": 0.693307439498058,
      "grad_norm": 0.15046358108520508,
      "learning_rate": 4.305375844557843e-05,
      "loss": 0.6326,
      "step": 4641
    },
    {
      "epoch": 0.693456827009262,
      "grad_norm": 0.2537308931350708,
      "learning_rate": 4.301512853530837e-05,
      "loss": 1.1665,
      "step": 4642
    },
    {
      "epoch": 0.6936062145204661,
      "grad_norm": 0.19490928947925568,
      "learning_rate": 4.297651121389901e-05,
      "loss": 0.7825,
      "step": 4643
    },
    {
      "epoch": 0.6937556020316702,
      "grad_norm": 0.12749908864498138,
      "learning_rate": 4.2937906489881516e-05,
      "loss": 0.6161,
      "step": 4644
    },
    {
      "epoch": 0.6939049895428742,
      "grad_norm": 0.15311866998672485,
      "learning_rate": 4.2899314371784305e-05,
      "loss": 0.7838,
      "step": 4645
    },
    {
      "epoch": 0.6940543770540782,
      "grad_norm": 0.12568438053131104,
      "learning_rate": 4.2860734868133e-05,
      "loss": 0.6825,
      "step": 4646
    },
    {
      "epoch": 0.6942037645652823,
      "grad_norm": 0.16030338406562805,
      "learning_rate": 4.282216798745038e-05,
      "loss": 0.8458,
      "step": 4647
    },
    {
      "epoch": 0.6943531520764864,
      "grad_norm": 0.13692030310630798,
      "learning_rate": 4.27836137382565e-05,
      "loss": 0.843,
      "step": 4648
    },
    {
      "epoch": 0.6945025395876905,
      "grad_norm": 0.12696976959705353,
      "learning_rate": 4.2745072129068606e-05,
      "loss": 0.7281,
      "step": 4649
    },
    {
      "epoch": 0.6946519270988946,
      "grad_norm": 0.16177400946617126,
      "learning_rate": 4.270654316840115e-05,
      "loss": 0.6666,
      "step": 4650
    },
    {
      "epoch": 0.6948013146100986,
      "grad_norm": 0.12775912880897522,
      "learning_rate": 4.266802686476579e-05,
      "loss": 0.6794,
      "step": 4651
    },
    {
      "epoch": 0.6949507021213026,
      "grad_norm": 0.2096504271030426,
      "learning_rate": 4.2629523226671375e-05,
      "loss": 0.7933,
      "step": 4652
    },
    {
      "epoch": 0.6951000896325067,
      "grad_norm": 0.12742532789707184,
      "learning_rate": 4.259103226262403e-05,
      "loss": 0.6864,
      "step": 4653
    },
    {
      "epoch": 0.6952494771437108,
      "grad_norm": 0.12906348705291748,
      "learning_rate": 4.25525539811269e-05,
      "loss": 0.5673,
      "step": 4654
    },
    {
      "epoch": 0.6953988646549149,
      "grad_norm": 0.13494399189949036,
      "learning_rate": 4.251408839068054e-05,
      "loss": 0.7522,
      "step": 4655
    },
    {
      "epoch": 0.6955482521661189,
      "grad_norm": 0.15384702384471893,
      "learning_rate": 4.247563549978259e-05,
      "loss": 0.6158,
      "step": 4656
    },
    {
      "epoch": 0.695697639677323,
      "grad_norm": 0.12394435703754425,
      "learning_rate": 4.2437195316927895e-05,
      "loss": 0.6551,
      "step": 4657
    },
    {
      "epoch": 0.6958470271885271,
      "grad_norm": 0.11181367933750153,
      "learning_rate": 4.23987678506085e-05,
      "loss": 0.5762,
      "step": 4658
    },
    {
      "epoch": 0.6959964146997311,
      "grad_norm": 0.17185263335704803,
      "learning_rate": 4.236035310931369e-05,
      "loss": 0.8967,
      "step": 4659
    },
    {
      "epoch": 0.6961458022109351,
      "grad_norm": 0.12710343301296234,
      "learning_rate": 4.23219511015298e-05,
      "loss": 0.7495,
      "step": 4660
    },
    {
      "epoch": 0.6962951897221392,
      "grad_norm": 0.13901861011981964,
      "learning_rate": 4.228356183574048e-05,
      "loss": 0.9328,
      "step": 4661
    },
    {
      "epoch": 0.6964445772333433,
      "grad_norm": 0.12986144423484802,
      "learning_rate": 4.224518532042654e-05,
      "loss": 0.8183,
      "step": 4662
    },
    {
      "epoch": 0.6965939647445474,
      "grad_norm": 0.13648369908332825,
      "learning_rate": 4.2206821564065946e-05,
      "loss": 0.5596,
      "step": 4663
    },
    {
      "epoch": 0.6967433522557515,
      "grad_norm": 0.11775126308202744,
      "learning_rate": 4.2168470575133866e-05,
      "loss": 0.5776,
      "step": 4664
    },
    {
      "epoch": 0.6968927397669555,
      "grad_norm": 0.19849622249603271,
      "learning_rate": 4.213013236210268e-05,
      "loss": 0.5902,
      "step": 4665
    },
    {
      "epoch": 0.6970421272781595,
      "grad_norm": 0.3871475160121918,
      "learning_rate": 4.2091806933441826e-05,
      "loss": 1.1529,
      "step": 4666
    },
    {
      "epoch": 0.6971915147893636,
      "grad_norm": 0.14403188228607178,
      "learning_rate": 4.205349429761805e-05,
      "loss": 0.9964,
      "step": 4667
    },
    {
      "epoch": 0.6973409023005677,
      "grad_norm": 0.12476871907711029,
      "learning_rate": 4.20151944630952e-05,
      "loss": 0.508,
      "step": 4668
    },
    {
      "epoch": 0.6974902898117717,
      "grad_norm": 0.4246998131275177,
      "learning_rate": 4.197690743833431e-05,
      "loss": 1.2671,
      "step": 4669
    },
    {
      "epoch": 0.6976396773229758,
      "grad_norm": 0.23385706543922424,
      "learning_rate": 4.1938633231793624e-05,
      "loss": 1.0746,
      "step": 4670
    },
    {
      "epoch": 0.6977890648341799,
      "grad_norm": 0.1790727823972702,
      "learning_rate": 4.1900371851928496e-05,
      "loss": 0.5776,
      "step": 4671
    },
    {
      "epoch": 0.6979384523453839,
      "grad_norm": 0.23241358995437622,
      "learning_rate": 4.1862123307191484e-05,
      "loss": 0.6734,
      "step": 4672
    },
    {
      "epoch": 0.698087839856588,
      "grad_norm": 0.15196742117404938,
      "learning_rate": 4.182388760603228e-05,
      "loss": 0.6879,
      "step": 4673
    },
    {
      "epoch": 0.698237227367792,
      "grad_norm": 0.1271677315235138,
      "learning_rate": 4.178566475689777e-05,
      "loss": 0.6459,
      "step": 4674
    },
    {
      "epoch": 0.6983866148789961,
      "grad_norm": 0.14158359169960022,
      "learning_rate": 4.174745476823199e-05,
      "loss": 0.7092,
      "step": 4675
    },
    {
      "epoch": 0.6985360023902002,
      "grad_norm": 0.39204293489456177,
      "learning_rate": 4.1709257648476105e-05,
      "loss": 1.1243,
      "step": 4676
    },
    {
      "epoch": 0.6986853899014043,
      "grad_norm": 0.1750764399766922,
      "learning_rate": 4.16710734060685e-05,
      "loss": 0.4715,
      "step": 4677
    },
    {
      "epoch": 0.6988347774126084,
      "grad_norm": 0.12426906079053879,
      "learning_rate": 4.1632902049444686e-05,
      "loss": 0.5197,
      "step": 4678
    },
    {
      "epoch": 0.6989841649238123,
      "grad_norm": 0.12710244953632355,
      "learning_rate": 4.1594743587037276e-05,
      "loss": 0.8483,
      "step": 4679
    },
    {
      "epoch": 0.6991335524350164,
      "grad_norm": 0.20769131183624268,
      "learning_rate": 4.155659802727608e-05,
      "loss": 1.0049,
      "step": 4680
    },
    {
      "epoch": 0.6992829399462205,
      "grad_norm": 0.14490796625614166,
      "learning_rate": 4.151846537858809e-05,
      "loss": 0.5979,
      "step": 4681
    },
    {
      "epoch": 0.6994323274574246,
      "grad_norm": 0.12019822001457214,
      "learning_rate": 4.14803456493974e-05,
      "loss": 0.4748,
      "step": 4682
    },
    {
      "epoch": 0.6995817149686286,
      "grad_norm": 0.15240825712680817,
      "learning_rate": 4.1442238848125245e-05,
      "loss": 0.526,
      "step": 4683
    },
    {
      "epoch": 0.6997311024798327,
      "grad_norm": 0.13235679268836975,
      "learning_rate": 4.1404144983190044e-05,
      "loss": 0.7624,
      "step": 4684
    },
    {
      "epoch": 0.6998804899910368,
      "grad_norm": 0.12137701362371445,
      "learning_rate": 4.136606406300736e-05,
      "loss": 0.7596,
      "step": 4685
    },
    {
      "epoch": 0.7000298775022408,
      "grad_norm": 0.13822999596595764,
      "learning_rate": 4.132799609598981e-05,
      "loss": 0.8379,
      "step": 4686
    },
    {
      "epoch": 0.7001792650134449,
      "grad_norm": 0.18476468324661255,
      "learning_rate": 4.128994109054724e-05,
      "loss": 0.6875,
      "step": 4687
    },
    {
      "epoch": 0.7003286525246489,
      "grad_norm": 0.15190930664539337,
      "learning_rate": 4.1251899055086586e-05,
      "loss": 0.7554,
      "step": 4688
    },
    {
      "epoch": 0.700478040035853,
      "grad_norm": 0.12816409766674042,
      "learning_rate": 4.121386999801195e-05,
      "loss": 0.7609,
      "step": 4689
    },
    {
      "epoch": 0.7006274275470571,
      "grad_norm": 0.3068293333053589,
      "learning_rate": 4.117585392772457e-05,
      "loss": 0.6917,
      "step": 4690
    },
    {
      "epoch": 0.7007768150582612,
      "grad_norm": 0.1662040650844574,
      "learning_rate": 4.113785085262278e-05,
      "loss": 0.789,
      "step": 4691
    },
    {
      "epoch": 0.7009262025694651,
      "grad_norm": 0.14395006000995636,
      "learning_rate": 4.109986078110207e-05,
      "loss": 0.5034,
      "step": 4692
    },
    {
      "epoch": 0.7010755900806692,
      "grad_norm": 0.1381119340658188,
      "learning_rate": 4.106188372155504e-05,
      "loss": 0.9112,
      "step": 4693
    },
    {
      "epoch": 0.7012249775918733,
      "grad_norm": 0.14812889695167542,
      "learning_rate": 4.102391968237143e-05,
      "loss": 0.5825,
      "step": 4694
    },
    {
      "epoch": 0.7013743651030774,
      "grad_norm": 0.13600662350654602,
      "learning_rate": 4.0985968671938094e-05,
      "loss": 0.7681,
      "step": 4695
    },
    {
      "epoch": 0.7015237526142815,
      "grad_norm": 0.2129794955253601,
      "learning_rate": 4.0948030698639006e-05,
      "loss": 0.7743,
      "step": 4696
    },
    {
      "epoch": 0.7016731401254855,
      "grad_norm": 0.18759317696094513,
      "learning_rate": 4.0910105770855277e-05,
      "loss": 0.8777,
      "step": 4697
    },
    {
      "epoch": 0.7018225276366896,
      "grad_norm": 0.084748275578022,
      "learning_rate": 4.087219389696515e-05,
      "loss": 0.31,
      "step": 4698
    },
    {
      "epoch": 0.7019719151478936,
      "grad_norm": 0.14129257202148438,
      "learning_rate": 4.08342950853439e-05,
      "loss": 0.6513,
      "step": 4699
    },
    {
      "epoch": 0.7021213026590977,
      "grad_norm": 0.15457667410373688,
      "learning_rate": 4.0796409344363995e-05,
      "loss": 0.7227,
      "step": 4700
    },
    {
      "epoch": 0.7022706901703017,
      "grad_norm": 0.19534826278686523,
      "learning_rate": 4.0758536682395e-05,
      "loss": 1.0484,
      "step": 4701
    },
    {
      "epoch": 0.7024200776815058,
      "grad_norm": 0.16052408516407013,
      "learning_rate": 4.072067710780359e-05,
      "loss": 0.5476,
      "step": 4702
    },
    {
      "epoch": 0.7025694651927099,
      "grad_norm": 0.173800528049469,
      "learning_rate": 4.068283062895355e-05,
      "loss": 0.701,
      "step": 4703
    },
    {
      "epoch": 0.702718852703914,
      "grad_norm": 0.1300242841243744,
      "learning_rate": 4.064499725420579e-05,
      "loss": 0.737,
      "step": 4704
    },
    {
      "epoch": 0.7028682402151181,
      "grad_norm": 0.1571626365184784,
      "learning_rate": 4.0607176991918225e-05,
      "loss": 0.5822,
      "step": 4705
    },
    {
      "epoch": 0.703017627726322,
      "grad_norm": 0.1941061168909073,
      "learning_rate": 4.0569369850446016e-05,
      "loss": 0.4528,
      "step": 4706
    },
    {
      "epoch": 0.7031670152375261,
      "grad_norm": 0.3267779052257538,
      "learning_rate": 4.053157583814133e-05,
      "loss": 1.0484,
      "step": 4707
    },
    {
      "epoch": 0.7033164027487302,
      "grad_norm": 0.13618910312652588,
      "learning_rate": 4.049379496335347e-05,
      "loss": 0.508,
      "step": 4708
    },
    {
      "epoch": 0.7034657902599343,
      "grad_norm": 0.13677671551704407,
      "learning_rate": 4.045602723442884e-05,
      "loss": 0.9501,
      "step": 4709
    },
    {
      "epoch": 0.7036151777711384,
      "grad_norm": 0.23573485016822815,
      "learning_rate": 4.041827265971091e-05,
      "loss": 0.6417,
      "step": 4710
    },
    {
      "epoch": 0.7037645652823424,
      "grad_norm": 0.16288627684116364,
      "learning_rate": 4.0380531247540285e-05,
      "loss": 0.6865,
      "step": 4711
    },
    {
      "epoch": 0.7039139527935464,
      "grad_norm": 0.13989536464214325,
      "learning_rate": 4.0342803006254624e-05,
      "loss": 0.4992,
      "step": 4712
    },
    {
      "epoch": 0.7040633403047505,
      "grad_norm": 0.12532411515712738,
      "learning_rate": 4.03050879441887e-05,
      "loss": 0.4175,
      "step": 4713
    },
    {
      "epoch": 0.7042127278159546,
      "grad_norm": 0.12677986919879913,
      "learning_rate": 4.0267386069674384e-05,
      "loss": 0.7943,
      "step": 4714
    },
    {
      "epoch": 0.7043621153271586,
      "grad_norm": 0.12706780433654785,
      "learning_rate": 4.022969739104059e-05,
      "loss": 0.7763,
      "step": 4715
    },
    {
      "epoch": 0.7045115028383627,
      "grad_norm": 0.15819714963436127,
      "learning_rate": 4.0192021916613344e-05,
      "loss": 0.5154,
      "step": 4716
    },
    {
      "epoch": 0.7046608903495668,
      "grad_norm": 0.14265893399715424,
      "learning_rate": 4.0154359654715815e-05,
      "loss": 0.7186,
      "step": 4717
    },
    {
      "epoch": 0.7048102778607709,
      "grad_norm": 0.14873388409614563,
      "learning_rate": 4.011671061366811e-05,
      "loss": 0.5109,
      "step": 4718
    },
    {
      "epoch": 0.7049596653719749,
      "grad_norm": 0.18315550684928894,
      "learning_rate": 4.007907480178752e-05,
      "loss": 0.9039,
      "step": 4719
    },
    {
      "epoch": 0.7051090528831789,
      "grad_norm": 0.21747682988643646,
      "learning_rate": 4.004145222738841e-05,
      "loss": 0.6933,
      "step": 4720
    },
    {
      "epoch": 0.705258440394383,
      "grad_norm": 0.24129554629325867,
      "learning_rate": 4.000384289878221e-05,
      "loss": 0.8154,
      "step": 4721
    },
    {
      "epoch": 0.7054078279055871,
      "grad_norm": 0.1393740028142929,
      "learning_rate": 3.996624682427739e-05,
      "loss": 0.5243,
      "step": 4722
    },
    {
      "epoch": 0.7055572154167912,
      "grad_norm": 0.24094797670841217,
      "learning_rate": 3.992866401217956e-05,
      "loss": 1.0619,
      "step": 4723
    },
    {
      "epoch": 0.7057066029279953,
      "grad_norm": 0.1673486828804016,
      "learning_rate": 3.989109447079131e-05,
      "loss": 0.7475,
      "step": 4724
    },
    {
      "epoch": 0.7058559904391993,
      "grad_norm": 0.1544388383626938,
      "learning_rate": 3.9853538208412366e-05,
      "loss": 0.5791,
      "step": 4725
    },
    {
      "epoch": 0.7060053779504033,
      "grad_norm": 0.19530798494815826,
      "learning_rate": 3.98159952333395e-05,
      "loss": 0.7236,
      "step": 4726
    },
    {
      "epoch": 0.7061547654616074,
      "grad_norm": 0.19490085542201996,
      "learning_rate": 3.977846555386652e-05,
      "loss": 0.7237,
      "step": 4727
    },
    {
      "epoch": 0.7063041529728115,
      "grad_norm": 0.16711673140525818,
      "learning_rate": 3.974094917828438e-05,
      "loss": 0.6483,
      "step": 4728
    },
    {
      "epoch": 0.7064535404840155,
      "grad_norm": 0.23387952148914337,
      "learning_rate": 3.9703446114881035e-05,
      "loss": 0.9943,
      "step": 4729
    },
    {
      "epoch": 0.7066029279952196,
      "grad_norm": 0.22450445592403412,
      "learning_rate": 3.966595637194153e-05,
      "loss": 0.4981,
      "step": 4730
    },
    {
      "epoch": 0.7067523155064237,
      "grad_norm": 0.17931799590587616,
      "learning_rate": 3.9628479957747856e-05,
      "loss": 0.5718,
      "step": 4731
    },
    {
      "epoch": 0.7069017030176278,
      "grad_norm": 0.22777549922466278,
      "learning_rate": 3.9591016880579204e-05,
      "loss": 1.03,
      "step": 4732
    },
    {
      "epoch": 0.7070510905288317,
      "grad_norm": 0.1345868557691574,
      "learning_rate": 3.955356714871175e-05,
      "loss": 0.7802,
      "step": 4733
    },
    {
      "epoch": 0.7072004780400358,
      "grad_norm": 0.3298538625240326,
      "learning_rate": 3.951613077041874e-05,
      "loss": 1.039,
      "step": 4734
    },
    {
      "epoch": 0.7073498655512399,
      "grad_norm": 0.23034991323947906,
      "learning_rate": 3.9478707753970445e-05,
      "loss": 0.5845,
      "step": 4735
    },
    {
      "epoch": 0.707499253062444,
      "grad_norm": 0.12878729403018951,
      "learning_rate": 3.944129810763425e-05,
      "loss": 0.5473,
      "step": 4736
    },
    {
      "epoch": 0.7076486405736481,
      "grad_norm": 0.31970176100730896,
      "learning_rate": 3.9403901839674474e-05,
      "loss": 0.8612,
      "step": 4737
    },
    {
      "epoch": 0.7077980280848521,
      "grad_norm": 0.1613045483827591,
      "learning_rate": 3.9366518958352585e-05,
      "loss": 0.3729,
      "step": 4738
    },
    {
      "epoch": 0.7079474155960561,
      "grad_norm": 0.13906604051589966,
      "learning_rate": 3.932914947192703e-05,
      "loss": 0.6513,
      "step": 4739
    },
    {
      "epoch": 0.7080968031072602,
      "grad_norm": 0.23129820823669434,
      "learning_rate": 3.929179338865333e-05,
      "loss": 0.8687,
      "step": 4740
    },
    {
      "epoch": 0.7082461906184643,
      "grad_norm": 0.11753810942173004,
      "learning_rate": 3.925445071678404e-05,
      "loss": 0.6082,
      "step": 4741
    },
    {
      "epoch": 0.7083955781296684,
      "grad_norm": 0.10983085632324219,
      "learning_rate": 3.9217121464568764e-05,
      "loss": 0.4192,
      "step": 4742
    },
    {
      "epoch": 0.7085449656408724,
      "grad_norm": 0.1299411654472351,
      "learning_rate": 3.9179805640254086e-05,
      "loss": 0.7172,
      "step": 4743
    },
    {
      "epoch": 0.7086943531520765,
      "grad_norm": 0.206137552857399,
      "learning_rate": 3.9142503252083695e-05,
      "loss": 1.0509,
      "step": 4744
    },
    {
      "epoch": 0.7088437406632806,
      "grad_norm": 0.2207917720079422,
      "learning_rate": 3.9105214308298266e-05,
      "loss": 0.4254,
      "step": 4745
    },
    {
      "epoch": 0.7089931281744846,
      "grad_norm": 0.18453793227672577,
      "learning_rate": 3.906793881713552e-05,
      "loss": 0.8574,
      "step": 4746
    },
    {
      "epoch": 0.7091425156856886,
      "grad_norm": 0.21461766958236694,
      "learning_rate": 3.90306767868302e-05,
      "loss": 0.6535,
      "step": 4747
    },
    {
      "epoch": 0.7092919031968927,
      "grad_norm": 0.12043561786413193,
      "learning_rate": 3.899342822561409e-05,
      "loss": 0.5515,
      "step": 4748
    },
    {
      "epoch": 0.7094412907080968,
      "grad_norm": 0.13545720279216766,
      "learning_rate": 3.895619314171601e-05,
      "loss": 0.8255,
      "step": 4749
    },
    {
      "epoch": 0.7095906782193009,
      "grad_norm": 0.27926769852638245,
      "learning_rate": 3.8918971543361714e-05,
      "loss": 0.9725,
      "step": 4750
    },
    {
      "epoch": 0.709740065730505,
      "grad_norm": 0.3345065712928772,
      "learning_rate": 3.8881763438774074e-05,
      "loss": 0.8842,
      "step": 4751
    },
    {
      "epoch": 0.709889453241709,
      "grad_norm": 0.2932509779930115,
      "learning_rate": 3.884456883617296e-05,
      "loss": 0.9278,
      "step": 4752
    },
    {
      "epoch": 0.710038840752913,
      "grad_norm": 0.12356627732515335,
      "learning_rate": 3.8807387743775235e-05,
      "loss": 0.7516,
      "step": 4753
    },
    {
      "epoch": 0.7101882282641171,
      "grad_norm": 0.14359675347805023,
      "learning_rate": 3.87702201697948e-05,
      "loss": 0.7037,
      "step": 4754
    },
    {
      "epoch": 0.7103376157753212,
      "grad_norm": 0.14137418568134308,
      "learning_rate": 3.873306612244256e-05,
      "loss": 0.604,
      "step": 4755
    },
    {
      "epoch": 0.7104870032865253,
      "grad_norm": 0.23085461556911469,
      "learning_rate": 3.8695925609926475e-05,
      "loss": 1.0126,
      "step": 4756
    },
    {
      "epoch": 0.7106363907977293,
      "grad_norm": 0.18453550338745117,
      "learning_rate": 3.8658798640451374e-05,
      "loss": 1.0274,
      "step": 4757
    },
    {
      "epoch": 0.7107857783089334,
      "grad_norm": 0.10957323759794235,
      "learning_rate": 3.862168522221927e-05,
      "loss": 0.5623,
      "step": 4758
    },
    {
      "epoch": 0.7109351658201374,
      "grad_norm": 0.1326921135187149,
      "learning_rate": 3.858458536342907e-05,
      "loss": 0.8087,
      "step": 4759
    },
    {
      "epoch": 0.7110845533313415,
      "grad_norm": 0.2162543535232544,
      "learning_rate": 3.854749907227674e-05,
      "loss": 0.8353,
      "step": 4760
    },
    {
      "epoch": 0.7112339408425455,
      "grad_norm": 0.13952651619911194,
      "learning_rate": 3.851042635695522e-05,
      "loss": 0.5872,
      "step": 4761
    },
    {
      "epoch": 0.7113833283537496,
      "grad_norm": 0.2419928014278412,
      "learning_rate": 3.8473367225654456e-05,
      "loss": 1.2963,
      "step": 4762
    },
    {
      "epoch": 0.7115327158649537,
      "grad_norm": 0.1801142543554306,
      "learning_rate": 3.8436321686561404e-05,
      "loss": 0.6765,
      "step": 4763
    },
    {
      "epoch": 0.7116821033761578,
      "grad_norm": 0.3168342113494873,
      "learning_rate": 3.839928974786001e-05,
      "loss": 0.9169,
      "step": 4764
    },
    {
      "epoch": 0.7118314908873619,
      "grad_norm": 0.3666203022003174,
      "learning_rate": 3.836227141773122e-05,
      "loss": 0.987,
      "step": 4765
    },
    {
      "epoch": 0.7119808783985658,
      "grad_norm": 0.15097633004188538,
      "learning_rate": 3.832526670435297e-05,
      "loss": 0.8164,
      "step": 4766
    },
    {
      "epoch": 0.7121302659097699,
      "grad_norm": 0.31511998176574707,
      "learning_rate": 3.828827561590018e-05,
      "loss": 0.767,
      "step": 4767
    },
    {
      "epoch": 0.712279653420974,
      "grad_norm": 0.22642910480499268,
      "learning_rate": 3.825129816054477e-05,
      "loss": 1.1922,
      "step": 4768
    },
    {
      "epoch": 0.7124290409321781,
      "grad_norm": 0.16663594543933868,
      "learning_rate": 3.821433434645567e-05,
      "loss": 0.5366,
      "step": 4769
    },
    {
      "epoch": 0.7125784284433822,
      "grad_norm": 0.16627927124500275,
      "learning_rate": 3.8177384181798716e-05,
      "loss": 0.575,
      "step": 4770
    },
    {
      "epoch": 0.7127278159545862,
      "grad_norm": 0.12122003734111786,
      "learning_rate": 3.8140447674736824e-05,
      "loss": 0.5168,
      "step": 4771
    },
    {
      "epoch": 0.7128772034657903,
      "grad_norm": 0.13659097254276276,
      "learning_rate": 3.810352483342984e-05,
      "loss": 0.6491,
      "step": 4772
    },
    {
      "epoch": 0.7130265909769943,
      "grad_norm": 0.1572633981704712,
      "learning_rate": 3.806661566603461e-05,
      "loss": 0.6138,
      "step": 4773
    },
    {
      "epoch": 0.7131759784881984,
      "grad_norm": 0.20381632447242737,
      "learning_rate": 3.802972018070495e-05,
      "loss": 0.756,
      "step": 4774
    },
    {
      "epoch": 0.7133253659994024,
      "grad_norm": 0.19783715903759003,
      "learning_rate": 3.79928383855917e-05,
      "loss": 0.7071,
      "step": 4775
    },
    {
      "epoch": 0.7134747535106065,
      "grad_norm": 0.28673315048217773,
      "learning_rate": 3.795597028884256e-05,
      "loss": 0.686,
      "step": 4776
    },
    {
      "epoch": 0.7136241410218106,
      "grad_norm": 0.14014649391174316,
      "learning_rate": 3.79191158986023e-05,
      "loss": 0.6179,
      "step": 4777
    },
    {
      "epoch": 0.7137735285330147,
      "grad_norm": 0.21388134360313416,
      "learning_rate": 3.7882275223012655e-05,
      "loss": 0.8058,
      "step": 4778
    },
    {
      "epoch": 0.7139229160442188,
      "grad_norm": 0.14886732399463654,
      "learning_rate": 3.784544827021231e-05,
      "loss": 0.7641,
      "step": 4779
    },
    {
      "epoch": 0.7140723035554227,
      "grad_norm": 0.14121030271053314,
      "learning_rate": 3.7808635048336916e-05,
      "loss": 0.4326,
      "step": 4780
    },
    {
      "epoch": 0.7142216910666268,
      "grad_norm": 0.23319652676582336,
      "learning_rate": 3.77718355655191e-05,
      "loss": 0.6685,
      "step": 4781
    },
    {
      "epoch": 0.7143710785778309,
      "grad_norm": 0.1756085604429245,
      "learning_rate": 3.773504982988845e-05,
      "loss": 0.3976,
      "step": 4782
    },
    {
      "epoch": 0.714520466089035,
      "grad_norm": 0.12128212302923203,
      "learning_rate": 3.769827784957152e-05,
      "loss": 0.7663,
      "step": 4783
    },
    {
      "epoch": 0.714669853600239,
      "grad_norm": 0.31746405363082886,
      "learning_rate": 3.766151963269182e-05,
      "loss": 0.9114,
      "step": 4784
    },
    {
      "epoch": 0.7148192411114431,
      "grad_norm": 0.35359877347946167,
      "learning_rate": 3.762477518736983e-05,
      "loss": 0.6531,
      "step": 4785
    },
    {
      "epoch": 0.7149686286226471,
      "grad_norm": 0.20281438529491425,
      "learning_rate": 3.7588044521722965e-05,
      "loss": 0.8739,
      "step": 4786
    },
    {
      "epoch": 0.7151180161338512,
      "grad_norm": 0.12364881485700607,
      "learning_rate": 3.755132764386563e-05,
      "loss": 0.4627,
      "step": 4787
    },
    {
      "epoch": 0.7152674036450553,
      "grad_norm": 0.18339090049266815,
      "learning_rate": 3.7514624561909176e-05,
      "loss": 0.9675,
      "step": 4788
    },
    {
      "epoch": 0.7154167911562593,
      "grad_norm": 0.2048814296722412,
      "learning_rate": 3.747793528396185e-05,
      "loss": 0.5416,
      "step": 4789
    },
    {
      "epoch": 0.7155661786674634,
      "grad_norm": 0.13941746950149536,
      "learning_rate": 3.74412598181289e-05,
      "loss": 0.5613,
      "step": 4790
    },
    {
      "epoch": 0.7157155661786675,
      "grad_norm": 0.1659175008535385,
      "learning_rate": 3.740459817251256e-05,
      "loss": 0.7363,
      "step": 4791
    },
    {
      "epoch": 0.7158649536898716,
      "grad_norm": 0.23634079098701477,
      "learning_rate": 3.7367950355211935e-05,
      "loss": 1.0652,
      "step": 4792
    },
    {
      "epoch": 0.7160143412010755,
      "grad_norm": 0.21791531145572662,
      "learning_rate": 3.7331316374323114e-05,
      "loss": 0.5689,
      "step": 4793
    },
    {
      "epoch": 0.7161637287122796,
      "grad_norm": 0.12283717840909958,
      "learning_rate": 3.7294696237939154e-05,
      "loss": 0.7933,
      "step": 4794
    },
    {
      "epoch": 0.7163131162234837,
      "grad_norm": 0.22052957117557526,
      "learning_rate": 3.7258089954149965e-05,
      "loss": 0.7518,
      "step": 4795
    },
    {
      "epoch": 0.7164625037346878,
      "grad_norm": 0.21824215352535248,
      "learning_rate": 3.7221497531042496e-05,
      "loss": 0.7442,
      "step": 4796
    },
    {
      "epoch": 0.7166118912458919,
      "grad_norm": 0.1544213891029358,
      "learning_rate": 3.718491897670058e-05,
      "loss": 0.6335,
      "step": 4797
    },
    {
      "epoch": 0.7167612787570959,
      "grad_norm": 0.15000516176223755,
      "learning_rate": 3.714835429920499e-05,
      "loss": 0.7101,
      "step": 4798
    },
    {
      "epoch": 0.7169106662683,
      "grad_norm": 0.12436383962631226,
      "learning_rate": 3.711180350663346e-05,
      "loss": 0.6837,
      "step": 4799
    },
    {
      "epoch": 0.717060053779504,
      "grad_norm": 0.18402902781963348,
      "learning_rate": 3.7075266607060633e-05,
      "loss": 0.6637,
      "step": 4800
    },
    {
      "epoch": 0.7172094412907081,
      "grad_norm": 0.29761019349098206,
      "learning_rate": 3.703874360855809e-05,
      "loss": 0.9227,
      "step": 4801
    },
    {
      "epoch": 0.7173588288019122,
      "grad_norm": 0.21496152877807617,
      "learning_rate": 3.700223451919434e-05,
      "loss": 1.2139,
      "step": 4802
    },
    {
      "epoch": 0.7175082163131162,
      "grad_norm": 0.14302749931812286,
      "learning_rate": 3.696573934703481e-05,
      "loss": 0.4664,
      "step": 4803
    },
    {
      "epoch": 0.7176576038243203,
      "grad_norm": 0.13339073956012726,
      "learning_rate": 3.6929258100141884e-05,
      "loss": 0.7263,
      "step": 4804
    },
    {
      "epoch": 0.7178069913355244,
      "grad_norm": 0.14082330465316772,
      "learning_rate": 3.6892790786574825e-05,
      "loss": 0.5282,
      "step": 4805
    },
    {
      "epoch": 0.7179563788467284,
      "grad_norm": 0.18513944745063782,
      "learning_rate": 3.685633741438984e-05,
      "loss": 0.8394,
      "step": 4806
    },
    {
      "epoch": 0.7181057663579324,
      "grad_norm": 0.13986468315124512,
      "learning_rate": 3.681989799164011e-05,
      "loss": 0.6052,
      "step": 4807
    },
    {
      "epoch": 0.7182551538691365,
      "grad_norm": 0.565620481967926,
      "learning_rate": 3.6783472526375596e-05,
      "loss": 1.0783,
      "step": 4808
    },
    {
      "epoch": 0.7184045413803406,
      "grad_norm": 0.11753922700881958,
      "learning_rate": 3.674706102664329e-05,
      "loss": 0.579,
      "step": 4809
    },
    {
      "epoch": 0.7185539288915447,
      "grad_norm": 0.14659751951694489,
      "learning_rate": 3.6710663500487096e-05,
      "loss": 0.4559,
      "step": 4810
    },
    {
      "epoch": 0.7187033164027488,
      "grad_norm": 0.15378767251968384,
      "learning_rate": 3.6674279955947776e-05,
      "loss": 0.6418,
      "step": 4811
    },
    {
      "epoch": 0.7188527039139528,
      "grad_norm": 0.1519133597612381,
      "learning_rate": 3.663791040106304e-05,
      "loss": 0.7174,
      "step": 4812
    },
    {
      "epoch": 0.7190020914251568,
      "grad_norm": 0.1260557919740677,
      "learning_rate": 3.66015548438675e-05,
      "loss": 0.6083,
      "step": 4813
    },
    {
      "epoch": 0.7191514789363609,
      "grad_norm": 0.15177126228809357,
      "learning_rate": 3.6565213292392706e-05,
      "loss": 0.8029,
      "step": 4814
    },
    {
      "epoch": 0.719300866447565,
      "grad_norm": 0.21821776032447815,
      "learning_rate": 3.652888575466703e-05,
      "loss": 0.7169,
      "step": 4815
    },
    {
      "epoch": 0.719450253958769,
      "grad_norm": 0.15493890643119812,
      "learning_rate": 3.6492572238715806e-05,
      "loss": 0.6885,
      "step": 4816
    },
    {
      "epoch": 0.7195996414699731,
      "grad_norm": 0.14351004362106323,
      "learning_rate": 3.645627275256126e-05,
      "loss": 0.7623,
      "step": 4817
    },
    {
      "epoch": 0.7197490289811772,
      "grad_norm": 0.1219073086977005,
      "learning_rate": 3.641998730422257e-05,
      "loss": 0.5804,
      "step": 4818
    },
    {
      "epoch": 0.7198984164923813,
      "grad_norm": 0.1784246861934662,
      "learning_rate": 3.6383715901715744e-05,
      "loss": 0.7603,
      "step": 4819
    },
    {
      "epoch": 0.7200478040035853,
      "grad_norm": 0.13119448721408844,
      "learning_rate": 3.6347458553053726e-05,
      "loss": 0.6619,
      "step": 4820
    },
    {
      "epoch": 0.7201971915147893,
      "grad_norm": 0.15528130531311035,
      "learning_rate": 3.63112152662463e-05,
      "loss": 0.7521,
      "step": 4821
    },
    {
      "epoch": 0.7203465790259934,
      "grad_norm": 0.3417429029941559,
      "learning_rate": 3.6274986049300186e-05,
      "loss": 1.3162,
      "step": 4822
    },
    {
      "epoch": 0.7204959665371975,
      "grad_norm": 0.15710623562335968,
      "learning_rate": 3.623877091021901e-05,
      "loss": 0.7327,
      "step": 4823
    },
    {
      "epoch": 0.7206453540484016,
      "grad_norm": 0.2467455118894577,
      "learning_rate": 3.620256985700328e-05,
      "loss": 0.7573,
      "step": 4824
    },
    {
      "epoch": 0.7207947415596057,
      "grad_norm": 0.14444081485271454,
      "learning_rate": 3.616638289765035e-05,
      "loss": 0.734,
      "step": 4825
    },
    {
      "epoch": 0.7209441290708096,
      "grad_norm": 0.19203653931617737,
      "learning_rate": 3.613021004015452e-05,
      "loss": 0.631,
      "step": 4826
    },
    {
      "epoch": 0.7210935165820137,
      "grad_norm": 0.1447475254535675,
      "learning_rate": 3.609405129250697e-05,
      "loss": 0.4203,
      "step": 4827
    },
    {
      "epoch": 0.7212429040932178,
      "grad_norm": 0.14933067560195923,
      "learning_rate": 3.605790666269566e-05,
      "loss": 0.436,
      "step": 4828
    },
    {
      "epoch": 0.7213922916044219,
      "grad_norm": 0.14689955115318298,
      "learning_rate": 3.602177615870558e-05,
      "loss": 0.5561,
      "step": 4829
    },
    {
      "epoch": 0.721541679115626,
      "grad_norm": 0.16088725626468658,
      "learning_rate": 3.5985659788518486e-05,
      "loss": 0.7142,
      "step": 4830
    },
    {
      "epoch": 0.72169106662683,
      "grad_norm": 0.2595069110393524,
      "learning_rate": 3.59495575601131e-05,
      "loss": 0.6447,
      "step": 4831
    },
    {
      "epoch": 0.7218404541380341,
      "grad_norm": 0.1803315281867981,
      "learning_rate": 3.591346948146495e-05,
      "loss": 0.8007,
      "step": 4832
    },
    {
      "epoch": 0.7219898416492381,
      "grad_norm": 0.15082429349422455,
      "learning_rate": 3.5877395560546465e-05,
      "loss": 0.7846,
      "step": 4833
    },
    {
      "epoch": 0.7221392291604422,
      "grad_norm": 0.12936636805534363,
      "learning_rate": 3.584133580532696e-05,
      "loss": 0.7282,
      "step": 4834
    },
    {
      "epoch": 0.7222886166716462,
      "grad_norm": 0.21296890079975128,
      "learning_rate": 3.580529022377258e-05,
      "loss": 0.5514,
      "step": 4835
    },
    {
      "epoch": 0.7224380041828503,
      "grad_norm": 0.13324841856956482,
      "learning_rate": 3.57692588238464e-05,
      "loss": 0.8157,
      "step": 4836
    },
    {
      "epoch": 0.7225873916940544,
      "grad_norm": 1.0075340270996094,
      "learning_rate": 3.573324161350829e-05,
      "loss": 1.8935,
      "step": 4837
    },
    {
      "epoch": 0.7227367792052585,
      "grad_norm": 0.1315302848815918,
      "learning_rate": 3.569723860071505e-05,
      "loss": 0.622,
      "step": 4838
    },
    {
      "epoch": 0.7228861667164626,
      "grad_norm": 0.1604132354259491,
      "learning_rate": 3.5661249793420306e-05,
      "loss": 0.8162,
      "step": 4839
    },
    {
      "epoch": 0.7230355542276665,
      "grad_norm": 0.17565181851387024,
      "learning_rate": 3.5625275199574594e-05,
      "loss": 0.6824,
      "step": 4840
    },
    {
      "epoch": 0.7231849417388706,
      "grad_norm": 0.557249903678894,
      "learning_rate": 3.5589314827125186e-05,
      "loss": 1.1108,
      "step": 4841
    },
    {
      "epoch": 0.7233343292500747,
      "grad_norm": 0.1486799716949463,
      "learning_rate": 3.555336868401635e-05,
      "loss": 0.9465,
      "step": 4842
    },
    {
      "epoch": 0.7234837167612788,
      "grad_norm": 0.18214502930641174,
      "learning_rate": 3.5517436778189164e-05,
      "loss": 0.763,
      "step": 4843
    },
    {
      "epoch": 0.7236331042724828,
      "grad_norm": 0.2883506715297699,
      "learning_rate": 3.5481519117581544e-05,
      "loss": 0.9877,
      "step": 4844
    },
    {
      "epoch": 0.7237824917836869,
      "grad_norm": 0.27589353919029236,
      "learning_rate": 3.544561571012826e-05,
      "loss": 0.7554,
      "step": 4845
    },
    {
      "epoch": 0.723931879294891,
      "grad_norm": 0.14624963700771332,
      "learning_rate": 3.540972656376099e-05,
      "loss": 0.7684,
      "step": 4846
    },
    {
      "epoch": 0.724081266806095,
      "grad_norm": 0.16719317436218262,
      "learning_rate": 3.5373851686408154e-05,
      "loss": 0.6884,
      "step": 4847
    },
    {
      "epoch": 0.724230654317299,
      "grad_norm": 0.17736296355724335,
      "learning_rate": 3.533799108599509e-05,
      "loss": 0.8527,
      "step": 4848
    },
    {
      "epoch": 0.7243800418285031,
      "grad_norm": 0.15044468641281128,
      "learning_rate": 3.5302144770444e-05,
      "loss": 0.5644,
      "step": 4849
    },
    {
      "epoch": 0.7245294293397072,
      "grad_norm": 0.22550426423549652,
      "learning_rate": 3.526631274767389e-05,
      "loss": 0.7893,
      "step": 4850
    },
    {
      "epoch": 0.7246788168509113,
      "grad_norm": 0.13661430776119232,
      "learning_rate": 3.523049502560063e-05,
      "loss": 0.7625,
      "step": 4851
    },
    {
      "epoch": 0.7248282043621154,
      "grad_norm": 0.2950759828090668,
      "learning_rate": 3.51946916121369e-05,
      "loss": 0.745,
      "step": 4852
    },
    {
      "epoch": 0.7249775918733193,
      "grad_norm": 0.43481260538101196,
      "learning_rate": 3.515890251519227e-05,
      "loss": 0.9223,
      "step": 4853
    },
    {
      "epoch": 0.7251269793845234,
      "grad_norm": 0.14166009426116943,
      "learning_rate": 3.512312774267309e-05,
      "loss": 0.5891,
      "step": 4854
    },
    {
      "epoch": 0.7252763668957275,
      "grad_norm": 0.146379292011261,
      "learning_rate": 3.5087367302482596e-05,
      "loss": 0.674,
      "step": 4855
    },
    {
      "epoch": 0.7254257544069316,
      "grad_norm": 0.11837875843048096,
      "learning_rate": 3.505162120252083e-05,
      "loss": 0.5529,
      "step": 4856
    },
    {
      "epoch": 0.7255751419181357,
      "grad_norm": 0.1523013412952423,
      "learning_rate": 3.501588945068467e-05,
      "loss": 0.6864,
      "step": 4857
    },
    {
      "epoch": 0.7257245294293397,
      "grad_norm": 0.21290011703968048,
      "learning_rate": 3.4980172054867824e-05,
      "loss": 0.4669,
      "step": 4858
    },
    {
      "epoch": 0.7258739169405438,
      "grad_norm": 0.15386582911014557,
      "learning_rate": 3.494446902296086e-05,
      "loss": 0.7396,
      "step": 4859
    },
    {
      "epoch": 0.7260233044517478,
      "grad_norm": 0.19065417349338531,
      "learning_rate": 3.490878036285109e-05,
      "loss": 0.9365,
      "step": 4860
    },
    {
      "epoch": 0.7261726919629519,
      "grad_norm": 0.1286652535200119,
      "learning_rate": 3.487310608242272e-05,
      "loss": 0.7746,
      "step": 4861
    },
    {
      "epoch": 0.726322079474156,
      "grad_norm": 0.11794201284646988,
      "learning_rate": 3.483744618955678e-05,
      "loss": 0.5505,
      "step": 4862
    },
    {
      "epoch": 0.72647146698536,
      "grad_norm": 0.1947849541902542,
      "learning_rate": 3.4801800692131095e-05,
      "loss": 0.5924,
      "step": 4863
    },
    {
      "epoch": 0.7266208544965641,
      "grad_norm": 0.2685377299785614,
      "learning_rate": 3.4766169598020326e-05,
      "loss": 1.2776,
      "step": 4864
    },
    {
      "epoch": 0.7267702420077682,
      "grad_norm": 0.13499918580055237,
      "learning_rate": 3.473055291509597e-05,
      "loss": 0.6081,
      "step": 4865
    },
    {
      "epoch": 0.7269196295189723,
      "grad_norm": 0.25589507818222046,
      "learning_rate": 3.469495065122627e-05,
      "loss": 0.4479,
      "step": 4866
    },
    {
      "epoch": 0.7270690170301762,
      "grad_norm": 0.15981222689151764,
      "learning_rate": 3.465936281427634e-05,
      "loss": 0.7699,
      "step": 4867
    },
    {
      "epoch": 0.7272184045413803,
      "grad_norm": 0.1838883012533188,
      "learning_rate": 3.462378941210811e-05,
      "loss": 0.8428,
      "step": 4868
    },
    {
      "epoch": 0.7273677920525844,
      "grad_norm": 0.12275431305170059,
      "learning_rate": 3.458823045258029e-05,
      "loss": 0.613,
      "step": 4869
    },
    {
      "epoch": 0.7275171795637885,
      "grad_norm": 0.15354964137077332,
      "learning_rate": 3.4552685943548446e-05,
      "loss": 0.808,
      "step": 4870
    },
    {
      "epoch": 0.7276665670749926,
      "grad_norm": 0.132765531539917,
      "learning_rate": 3.4517155892864916e-05,
      "loss": 0.7607,
      "step": 4871
    },
    {
      "epoch": 0.7278159545861966,
      "grad_norm": 0.15481601655483246,
      "learning_rate": 3.4481640308378826e-05,
      "loss": 0.7648,
      "step": 4872
    },
    {
      "epoch": 0.7279653420974006,
      "grad_norm": 0.2896997034549713,
      "learning_rate": 3.4446139197936166e-05,
      "loss": 0.6322,
      "step": 4873
    },
    {
      "epoch": 0.7281147296086047,
      "grad_norm": 0.2003454566001892,
      "learning_rate": 3.441065256937966e-05,
      "loss": 0.9804,
      "step": 4874
    },
    {
      "epoch": 0.7282641171198088,
      "grad_norm": 0.136475071310997,
      "learning_rate": 3.43751804305489e-05,
      "loss": 0.6631,
      "step": 4875
    },
    {
      "epoch": 0.7284135046310128,
      "grad_norm": 0.1172839105129242,
      "learning_rate": 3.4339722789280214e-05,
      "loss": 0.4526,
      "step": 4876
    },
    {
      "epoch": 0.7285628921422169,
      "grad_norm": 0.16096670925617218,
      "learning_rate": 3.430427965340678e-05,
      "loss": 0.6765,
      "step": 4877
    },
    {
      "epoch": 0.728712279653421,
      "grad_norm": 0.14083969593048096,
      "learning_rate": 3.4268851030758564e-05,
      "loss": 0.713,
      "step": 4878
    },
    {
      "epoch": 0.7288616671646251,
      "grad_norm": 0.1572013795375824,
      "learning_rate": 3.4233436929162246e-05,
      "loss": 0.6359,
      "step": 4879
    },
    {
      "epoch": 0.729011054675829,
      "grad_norm": 0.2318340539932251,
      "learning_rate": 3.4198037356441406e-05,
      "loss": 1.1381,
      "step": 4880
    },
    {
      "epoch": 0.7291604421870331,
      "grad_norm": 0.29885709285736084,
      "learning_rate": 3.4162652320416355e-05,
      "loss": 1.1139,
      "step": 4881
    },
    {
      "epoch": 0.7293098296982372,
      "grad_norm": 0.10630807280540466,
      "learning_rate": 3.412728182890422e-05,
      "loss": 0.4659,
      "step": 4882
    },
    {
      "epoch": 0.7294592172094413,
      "grad_norm": 0.15514931082725525,
      "learning_rate": 3.40919258897189e-05,
      "loss": 0.7648,
      "step": 4883
    },
    {
      "epoch": 0.7296086047206454,
      "grad_norm": 0.15156708657741547,
      "learning_rate": 3.405658451067107e-05,
      "loss": 0.6752,
      "step": 4884
    },
    {
      "epoch": 0.7297579922318495,
      "grad_norm": 0.19219937920570374,
      "learning_rate": 3.4021257699568267e-05,
      "loss": 0.9351,
      "step": 4885
    },
    {
      "epoch": 0.7299073797430535,
      "grad_norm": 0.18557752668857574,
      "learning_rate": 3.3985945464214644e-05,
      "loss": 0.5118,
      "step": 4886
    },
    {
      "epoch": 0.7300567672542575,
      "grad_norm": 0.19660314917564392,
      "learning_rate": 3.395064781241126e-05,
      "loss": 0.5078,
      "step": 4887
    },
    {
      "epoch": 0.7302061547654616,
      "grad_norm": 0.11988287419080734,
      "learning_rate": 3.391536475195597e-05,
      "loss": 0.541,
      "step": 4888
    },
    {
      "epoch": 0.7303555422766657,
      "grad_norm": 0.10480613261461258,
      "learning_rate": 3.388009629064331e-05,
      "loss": 0.6314,
      "step": 4889
    },
    {
      "epoch": 0.7305049297878697,
      "grad_norm": 0.17043368518352509,
      "learning_rate": 3.3844842436264645e-05,
      "loss": 0.7069,
      "step": 4890
    },
    {
      "epoch": 0.7306543172990738,
      "grad_norm": 0.1603437066078186,
      "learning_rate": 3.380960319660819e-05,
      "loss": 0.5169,
      "step": 4891
    },
    {
      "epoch": 0.7308037048102779,
      "grad_norm": 0.1630362868309021,
      "learning_rate": 3.3774378579458756e-05,
      "loss": 0.4691,
      "step": 4892
    },
    {
      "epoch": 0.730953092321482,
      "grad_norm": 0.14402790367603302,
      "learning_rate": 3.3739168592598056e-05,
      "loss": 0.5427,
      "step": 4893
    },
    {
      "epoch": 0.731102479832686,
      "grad_norm": 0.2103247195482254,
      "learning_rate": 3.370397324380453e-05,
      "loss": 0.8591,
      "step": 4894
    },
    {
      "epoch": 0.73125186734389,
      "grad_norm": 0.18361686170101166,
      "learning_rate": 3.366879254085339e-05,
      "loss": 0.7155,
      "step": 4895
    },
    {
      "epoch": 0.7314012548550941,
      "grad_norm": 0.2822781503200531,
      "learning_rate": 3.363362649151661e-05,
      "loss": 1.1073,
      "step": 4896
    },
    {
      "epoch": 0.7315506423662982,
      "grad_norm": 0.1566934734582901,
      "learning_rate": 3.359847510356293e-05,
      "loss": 0.6806,
      "step": 4897
    },
    {
      "epoch": 0.7317000298775023,
      "grad_norm": 0.23482957482337952,
      "learning_rate": 3.356333838475788e-05,
      "loss": 0.7034,
      "step": 4898
    },
    {
      "epoch": 0.7318494173887063,
      "grad_norm": 0.39208146929740906,
      "learning_rate": 3.3528216342863635e-05,
      "loss": 1.0448,
      "step": 4899
    },
    {
      "epoch": 0.7319988048999103,
      "grad_norm": 0.48120588064193726,
      "learning_rate": 3.349310898563928e-05,
      "loss": 0.7724,
      "step": 4900
    },
    {
      "epoch": 0.7321481924111144,
      "grad_norm": 0.13316498696804047,
      "learning_rate": 3.345801632084056e-05,
      "loss": 0.2074,
      "step": 4901
    },
    {
      "epoch": 0.7322975799223185,
      "grad_norm": 0.1250278353691101,
      "learning_rate": 3.342293835621999e-05,
      "loss": 0.7238,
      "step": 4902
    },
    {
      "epoch": 0.7324469674335226,
      "grad_norm": 0.1801614761352539,
      "learning_rate": 3.338787509952688e-05,
      "loss": 0.7776,
      "step": 4903
    },
    {
      "epoch": 0.7325963549447266,
      "grad_norm": 0.17948834598064423,
      "learning_rate": 3.335282655850727e-05,
      "loss": 0.7433,
      "step": 4904
    },
    {
      "epoch": 0.7327457424559307,
      "grad_norm": 0.269188791513443,
      "learning_rate": 3.3317792740903866e-05,
      "loss": 0.6755,
      "step": 4905
    },
    {
      "epoch": 0.7328951299671348,
      "grad_norm": 0.16531093418598175,
      "learning_rate": 3.328277365445621e-05,
      "loss": 0.5296,
      "step": 4906
    },
    {
      "epoch": 0.7330445174783388,
      "grad_norm": 0.1237042024731636,
      "learning_rate": 3.324776930690064e-05,
      "loss": 0.6364,
      "step": 4907
    },
    {
      "epoch": 0.7331939049895428,
      "grad_norm": 0.1580040603876114,
      "learning_rate": 3.321277970597013e-05,
      "loss": 0.5614,
      "step": 4908
    },
    {
      "epoch": 0.7333432925007469,
      "grad_norm": 0.1564781665802002,
      "learning_rate": 3.3177804859394435e-05,
      "loss": 0.8509,
      "step": 4909
    },
    {
      "epoch": 0.733492680011951,
      "grad_norm": 0.32889047265052795,
      "learning_rate": 3.314284477490005e-05,
      "loss": 1.0221,
      "step": 4910
    },
    {
      "epoch": 0.7336420675231551,
      "grad_norm": 0.12401048094034195,
      "learning_rate": 3.310789946021026e-05,
      "loss": 0.5775,
      "step": 4911
    },
    {
      "epoch": 0.7337914550343592,
      "grad_norm": 0.11835158616304398,
      "learning_rate": 3.307296892304496e-05,
      "loss": 0.6394,
      "step": 4912
    },
    {
      "epoch": 0.7339408425455632,
      "grad_norm": 0.20171450078487396,
      "learning_rate": 3.3038053171120885e-05,
      "loss": 0.5331,
      "step": 4913
    },
    {
      "epoch": 0.7340902300567672,
      "grad_norm": 0.6921703219413757,
      "learning_rate": 3.300315221215149e-05,
      "loss": 1.4942,
      "step": 4914
    },
    {
      "epoch": 0.7342396175679713,
      "grad_norm": 0.13117076456546783,
      "learning_rate": 3.296826605384695e-05,
      "loss": 0.7428,
      "step": 4915
    },
    {
      "epoch": 0.7343890050791754,
      "grad_norm": 0.1458209604024887,
      "learning_rate": 3.293339470391416e-05,
      "loss": 0.544,
      "step": 4916
    },
    {
      "epoch": 0.7345383925903795,
      "grad_norm": 0.2634068727493286,
      "learning_rate": 3.2898538170056813e-05,
      "loss": 0.7872,
      "step": 4917
    },
    {
      "epoch": 0.7346877801015835,
      "grad_norm": 0.16177213191986084,
      "learning_rate": 3.286369645997517e-05,
      "loss": 0.7515,
      "step": 4918
    },
    {
      "epoch": 0.7348371676127876,
      "grad_norm": 0.16289487481117249,
      "learning_rate": 3.2828869581366364e-05,
      "loss": 0.6469,
      "step": 4919
    },
    {
      "epoch": 0.7349865551239916,
      "grad_norm": 0.19163528084754944,
      "learning_rate": 3.279405754192419e-05,
      "loss": 0.5429,
      "step": 4920
    },
    {
      "epoch": 0.7351359426351957,
      "grad_norm": 0.13361147046089172,
      "learning_rate": 3.275926034933922e-05,
      "loss": 0.7462,
      "step": 4921
    },
    {
      "epoch": 0.7352853301463997,
      "grad_norm": 0.27693939208984375,
      "learning_rate": 3.2724478011298655e-05,
      "loss": 0.9758,
      "step": 4922
    },
    {
      "epoch": 0.7354347176576038,
      "grad_norm": 0.13949550688266754,
      "learning_rate": 3.26897105354865e-05,
      "loss": 0.8089,
      "step": 4923
    },
    {
      "epoch": 0.7355841051688079,
      "grad_norm": 0.2877447009086609,
      "learning_rate": 3.265495792958341e-05,
      "loss": 0.9768,
      "step": 4924
    },
    {
      "epoch": 0.735733492680012,
      "grad_norm": 0.151119202375412,
      "learning_rate": 3.262022020126682e-05,
      "loss": 0.5853,
      "step": 4925
    },
    {
      "epoch": 0.7358828801912161,
      "grad_norm": 0.13426688313484192,
      "learning_rate": 3.2585497358210816e-05,
      "loss": 0.851,
      "step": 4926
    },
    {
      "epoch": 0.73603226770242,
      "grad_norm": 0.15394486486911774,
      "learning_rate": 3.255078940808625e-05,
      "loss": 0.7609,
      "step": 4927
    },
    {
      "epoch": 0.7361816552136241,
      "grad_norm": 0.17044860124588013,
      "learning_rate": 3.2516096358560635e-05,
      "loss": 0.7128,
      "step": 4928
    },
    {
      "epoch": 0.7363310427248282,
      "grad_norm": 0.1305260807275772,
      "learning_rate": 3.248141821729822e-05,
      "loss": 0.7324,
      "step": 4929
    },
    {
      "epoch": 0.7364804302360323,
      "grad_norm": 0.12767207622528076,
      "learning_rate": 3.244675499196e-05,
      "loss": 0.6067,
      "step": 4930
    },
    {
      "epoch": 0.7366298177472363,
      "grad_norm": 0.25985705852508545,
      "learning_rate": 3.241210669020356e-05,
      "loss": 0.7031,
      "step": 4931
    },
    {
      "epoch": 0.7367792052584404,
      "grad_norm": 0.17194299399852753,
      "learning_rate": 3.2377473319683284e-05,
      "loss": 0.626,
      "step": 4932
    },
    {
      "epoch": 0.7369285927696445,
      "grad_norm": 0.17602917551994324,
      "learning_rate": 3.234285488805023e-05,
      "loss": 0.9724,
      "step": 4933
    },
    {
      "epoch": 0.7370779802808485,
      "grad_norm": 0.2732759714126587,
      "learning_rate": 3.2308251402952184e-05,
      "loss": 0.7364,
      "step": 4934
    },
    {
      "epoch": 0.7372273677920526,
      "grad_norm": 0.16593991219997406,
      "learning_rate": 3.227366287203357e-05,
      "loss": 0.5544,
      "step": 4935
    },
    {
      "epoch": 0.7373767553032566,
      "grad_norm": 0.19833271205425262,
      "learning_rate": 3.22390893029356e-05,
      "loss": 0.8524,
      "step": 4936
    },
    {
      "epoch": 0.7375261428144607,
      "grad_norm": 0.2168133705854416,
      "learning_rate": 3.2204530703296045e-05,
      "loss": 1.0758,
      "step": 4937
    },
    {
      "epoch": 0.7376755303256648,
      "grad_norm": 0.19049033522605896,
      "learning_rate": 3.216998708074948e-05,
      "loss": 0.389,
      "step": 4938
    },
    {
      "epoch": 0.7378249178368689,
      "grad_norm": 0.09984315931797028,
      "learning_rate": 3.213545844292714e-05,
      "loss": 0.4281,
      "step": 4939
    },
    {
      "epoch": 0.737974305348073,
      "grad_norm": 0.07858629524707794,
      "learning_rate": 3.2100944797456946e-05,
      "loss": 0.3187,
      "step": 4940
    },
    {
      "epoch": 0.7381236928592769,
      "grad_norm": 0.15625493228435516,
      "learning_rate": 3.206644615196351e-05,
      "loss": 0.3592,
      "step": 4941
    },
    {
      "epoch": 0.738273080370481,
      "grad_norm": 0.1579444408416748,
      "learning_rate": 3.2031962514068135e-05,
      "loss": 0.7601,
      "step": 4942
    },
    {
      "epoch": 0.7384224678816851,
      "grad_norm": 0.22950637340545654,
      "learning_rate": 3.19974938913888e-05,
      "loss": 0.5273,
      "step": 4943
    },
    {
      "epoch": 0.7385718553928892,
      "grad_norm": 0.37070295214653015,
      "learning_rate": 3.196304029154017e-05,
      "loss": 0.8437,
      "step": 4944
    },
    {
      "epoch": 0.7387212429040932,
      "grad_norm": 0.18260644376277924,
      "learning_rate": 3.192860172213359e-05,
      "loss": 0.7486,
      "step": 4945
    },
    {
      "epoch": 0.7388706304152973,
      "grad_norm": 0.33533430099487305,
      "learning_rate": 3.189417819077708e-05,
      "loss": 0.8253,
      "step": 4946
    },
    {
      "epoch": 0.7390200179265013,
      "grad_norm": 0.12721547484397888,
      "learning_rate": 3.1859769705075346e-05,
      "loss": 0.8251,
      "step": 4947
    },
    {
      "epoch": 0.7391694054377054,
      "grad_norm": 0.17424146831035614,
      "learning_rate": 3.182537627262977e-05,
      "loss": 0.8914,
      "step": 4948
    },
    {
      "epoch": 0.7393187929489095,
      "grad_norm": 0.13731816411018372,
      "learning_rate": 3.1790997901038446e-05,
      "loss": 0.6597,
      "step": 4949
    },
    {
      "epoch": 0.7394681804601135,
      "grad_norm": 0.13299672305583954,
      "learning_rate": 3.175663459789602e-05,
      "loss": 0.8984,
      "step": 4950
    },
    {
      "epoch": 0.7396175679713176,
      "grad_norm": 0.2328134924173355,
      "learning_rate": 3.172228637079394e-05,
      "loss": 0.9186,
      "step": 4951
    },
    {
      "epoch": 0.7397669554825217,
      "grad_norm": 0.14470314979553223,
      "learning_rate": 3.1687953227320255e-05,
      "loss": 0.7466,
      "step": 4952
    },
    {
      "epoch": 0.7399163429937258,
      "grad_norm": 0.22423890233039856,
      "learning_rate": 3.1653635175059715e-05,
      "loss": 0.7771,
      "step": 4953
    },
    {
      "epoch": 0.7400657305049297,
      "grad_norm": 0.3554167151451111,
      "learning_rate": 3.161933222159371e-05,
      "loss": 1.021,
      "step": 4954
    },
    {
      "epoch": 0.7402151180161338,
      "grad_norm": 0.20676332712173462,
      "learning_rate": 3.1585044374500314e-05,
      "loss": 0.636,
      "step": 4955
    },
    {
      "epoch": 0.7403645055273379,
      "grad_norm": 0.15013669431209564,
      "learning_rate": 3.155077164135428e-05,
      "loss": 0.8258,
      "step": 4956
    },
    {
      "epoch": 0.740513893038542,
      "grad_norm": 0.13174167275428772,
      "learning_rate": 3.151651402972694e-05,
      "loss": 0.9268,
      "step": 4957
    },
    {
      "epoch": 0.7406632805497461,
      "grad_norm": 0.17627792060375214,
      "learning_rate": 3.148227154718638e-05,
      "loss": 0.8157,
      "step": 4958
    },
    {
      "epoch": 0.7408126680609501,
      "grad_norm": 0.1449446976184845,
      "learning_rate": 3.1448044201297287e-05,
      "loss": 0.8371,
      "step": 4959
    },
    {
      "epoch": 0.7409620555721542,
      "grad_norm": 0.18143129348754883,
      "learning_rate": 3.141383199962106e-05,
      "loss": 0.4589,
      "step": 4960
    },
    {
      "epoch": 0.7411114430833582,
      "grad_norm": 0.20145520567893982,
      "learning_rate": 3.1379634949715675e-05,
      "loss": 0.8748,
      "step": 4961
    },
    {
      "epoch": 0.7412608305945623,
      "grad_norm": 0.2033759504556656,
      "learning_rate": 3.134545305913582e-05,
      "loss": 0.457,
      "step": 4962
    },
    {
      "epoch": 0.7414102181057663,
      "grad_norm": 0.20394885540008545,
      "learning_rate": 3.131128633543282e-05,
      "loss": 0.7754,
      "step": 4963
    },
    {
      "epoch": 0.7415596056169704,
      "grad_norm": 0.15353569388389587,
      "learning_rate": 3.1277134786154635e-05,
      "loss": 0.7836,
      "step": 4964
    },
    {
      "epoch": 0.7417089931281745,
      "grad_norm": 0.23631766438484192,
      "learning_rate": 3.12429984188459e-05,
      "loss": 0.8268,
      "step": 4965
    },
    {
      "epoch": 0.7418583806393786,
      "grad_norm": 0.17601190507411957,
      "learning_rate": 3.120887724104786e-05,
      "loss": 0.786,
      "step": 4966
    },
    {
      "epoch": 0.7420077681505826,
      "grad_norm": 0.24534234404563904,
      "learning_rate": 3.117477126029844e-05,
      "loss": 1.1769,
      "step": 4967
    },
    {
      "epoch": 0.7421571556617866,
      "grad_norm": 0.13882935047149658,
      "learning_rate": 3.114068048413218e-05,
      "loss": 0.7714,
      "step": 4968
    },
    {
      "epoch": 0.7423065431729907,
      "grad_norm": 0.18392275273799896,
      "learning_rate": 3.1106604920080305e-05,
      "loss": 0.9732,
      "step": 4969
    },
    {
      "epoch": 0.7424559306841948,
      "grad_norm": 0.8148640394210815,
      "learning_rate": 3.107254457567059e-05,
      "loss": 1.7701,
      "step": 4970
    },
    {
      "epoch": 0.7426053181953989,
      "grad_norm": 0.26277825236320496,
      "learning_rate": 3.103849945842754e-05,
      "loss": 0.7356,
      "step": 4971
    },
    {
      "epoch": 0.742754705706603,
      "grad_norm": 0.12942837178707123,
      "learning_rate": 3.100446957587224e-05,
      "loss": 0.522,
      "step": 4972
    },
    {
      "epoch": 0.742904093217807,
      "grad_norm": 0.32177719473838806,
      "learning_rate": 3.097045493552245e-05,
      "loss": 0.8224,
      "step": 4973
    },
    {
      "epoch": 0.743053480729011,
      "grad_norm": 0.17378588020801544,
      "learning_rate": 3.093645554489254e-05,
      "loss": 0.9629,
      "step": 4974
    },
    {
      "epoch": 0.7432028682402151,
      "grad_norm": 0.14082741737365723,
      "learning_rate": 3.090247141149354e-05,
      "loss": 0.8499,
      "step": 4975
    },
    {
      "epoch": 0.7433522557514192,
      "grad_norm": 0.14403241872787476,
      "learning_rate": 3.0868502542833014e-05,
      "loss": 0.8945,
      "step": 4976
    },
    {
      "epoch": 0.7435016432626232,
      "grad_norm": 0.4742143750190735,
      "learning_rate": 3.0834548946415264e-05,
      "loss": 0.986,
      "step": 4977
    },
    {
      "epoch": 0.7436510307738273,
      "grad_norm": 0.15638861060142517,
      "learning_rate": 3.080061062974119e-05,
      "loss": 0.5808,
      "step": 4978
    },
    {
      "epoch": 0.7438004182850314,
      "grad_norm": 0.1463359296321869,
      "learning_rate": 3.0766687600308244e-05,
      "loss": 0.3574,
      "step": 4979
    },
    {
      "epoch": 0.7439498057962355,
      "grad_norm": 0.14899824559688568,
      "learning_rate": 3.073277986561064e-05,
      "loss": 0.6442,
      "step": 4980
    },
    {
      "epoch": 0.7440991933074395,
      "grad_norm": 0.14941129088401794,
      "learning_rate": 3.06988874331391e-05,
      "loss": 0.6961,
      "step": 4981
    },
    {
      "epoch": 0.7442485808186435,
      "grad_norm": 0.14615993201732635,
      "learning_rate": 3.066501031038104e-05,
      "loss": 0.5963,
      "step": 4982
    },
    {
      "epoch": 0.7443979683298476,
      "grad_norm": 0.16326816380023956,
      "learning_rate": 3.0631148504820364e-05,
      "loss": 0.7108,
      "step": 4983
    },
    {
      "epoch": 0.7445473558410517,
      "grad_norm": 0.5542786717414856,
      "learning_rate": 3.059730202393773e-05,
      "loss": 1.1031,
      "step": 4984
    },
    {
      "epoch": 0.7446967433522558,
      "grad_norm": 0.38216614723205566,
      "learning_rate": 3.056347087521038e-05,
      "loss": 0.8559,
      "step": 4985
    },
    {
      "epoch": 0.7448461308634599,
      "grad_norm": 0.21998490393161774,
      "learning_rate": 3.052965506611212e-05,
      "loss": 0.7615,
      "step": 4986
    },
    {
      "epoch": 0.7449955183746638,
      "grad_norm": 0.15500688552856445,
      "learning_rate": 3.0495854604113406e-05,
      "loss": 0.5483,
      "step": 4987
    },
    {
      "epoch": 0.7451449058858679,
      "grad_norm": 0.16193658113479614,
      "learning_rate": 3.0462069496681333e-05,
      "loss": 1.0433,
      "step": 4988
    },
    {
      "epoch": 0.745294293397072,
      "grad_norm": 0.12404018640518188,
      "learning_rate": 3.04282997512795e-05,
      "loss": 0.6508,
      "step": 4989
    },
    {
      "epoch": 0.7454436809082761,
      "grad_norm": 0.1350613534450531,
      "learning_rate": 3.0394545375368212e-05,
      "loss": 0.7947,
      "step": 4990
    },
    {
      "epoch": 0.7455930684194801,
      "grad_norm": 0.17499637603759766,
      "learning_rate": 3.036080637640435e-05,
      "loss": 0.6111,
      "step": 4991
    },
    {
      "epoch": 0.7457424559306842,
      "grad_norm": 0.14929813146591187,
      "learning_rate": 3.0327082761841376e-05,
      "loss": 0.5512,
      "step": 4992
    },
    {
      "epoch": 0.7458918434418883,
      "grad_norm": 0.1146111935377121,
      "learning_rate": 3.0293374539129382e-05,
      "loss": 0.6288,
      "step": 4993
    },
    {
      "epoch": 0.7460412309530923,
      "grad_norm": 0.1721353381872177,
      "learning_rate": 3.0259681715715094e-05,
      "loss": 0.7352,
      "step": 4994
    },
    {
      "epoch": 0.7461906184642964,
      "grad_norm": 0.17522495985031128,
      "learning_rate": 3.0226004299041664e-05,
      "loss": 0.6711,
      "step": 4995
    },
    {
      "epoch": 0.7463400059755004,
      "grad_norm": 0.1915367841720581,
      "learning_rate": 3.019234229654909e-05,
      "loss": 1.0126,
      "step": 4996
    },
    {
      "epoch": 0.7464893934867045,
      "grad_norm": 0.158406063914299,
      "learning_rate": 3.015869571567379e-05,
      "loss": 0.6488,
      "step": 4997
    },
    {
      "epoch": 0.7466387809979086,
      "grad_norm": 0.13220737874507904,
      "learning_rate": 3.012506456384885e-05,
      "loss": 0.5635,
      "step": 4998
    },
    {
      "epoch": 0.7467881685091127,
      "grad_norm": 0.11290343850851059,
      "learning_rate": 3.00914488485039e-05,
      "loss": 0.5701,
      "step": 4999
    },
    {
      "epoch": 0.7469375560203168,
      "grad_norm": 0.12224692106246948,
      "learning_rate": 3.0057848577065194e-05,
      "loss": 0.7427,
      "step": 5000
    },
    {
      "epoch": 0.7470869435315207,
      "grad_norm": 0.15156102180480957,
      "learning_rate": 3.0024263756955596e-05,
      "loss": 0.8542,
      "step": 5001
    },
    {
      "epoch": 0.7472363310427248,
      "grad_norm": 0.12842871248722076,
      "learning_rate": 2.9990694395594454e-05,
      "loss": 0.6182,
      "step": 5002
    },
    {
      "epoch": 0.7473857185539289,
      "grad_norm": 0.14256559312343597,
      "learning_rate": 2.9957140500397816e-05,
      "loss": 1.0457,
      "step": 5003
    },
    {
      "epoch": 0.747535106065133,
      "grad_norm": 0.18850605189800262,
      "learning_rate": 2.9923602078778267e-05,
      "loss": 0.5922,
      "step": 5004
    },
    {
      "epoch": 0.747684493576337,
      "grad_norm": 0.1620553433895111,
      "learning_rate": 2.9890079138144968e-05,
      "loss": 0.744,
      "step": 5005
    },
    {
      "epoch": 0.7478338810875411,
      "grad_norm": 0.13287848234176636,
      "learning_rate": 2.9856571685903678e-05,
      "loss": 0.7904,
      "step": 5006
    },
    {
      "epoch": 0.7479832685987452,
      "grad_norm": 0.18404249846935272,
      "learning_rate": 2.982307972945674e-05,
      "loss": 0.6677,
      "step": 5007
    },
    {
      "epoch": 0.7481326561099492,
      "grad_norm": 0.2042563110589981,
      "learning_rate": 2.9789603276203006e-05,
      "loss": 0.9404,
      "step": 5008
    },
    {
      "epoch": 0.7482820436211532,
      "grad_norm": 0.1849035620689392,
      "learning_rate": 2.9756142333537983e-05,
      "loss": 0.8404,
      "step": 5009
    },
    {
      "epoch": 0.7484314311323573,
      "grad_norm": 0.24479633569717407,
      "learning_rate": 2.972269690885372e-05,
      "loss": 0.8957,
      "step": 5010
    },
    {
      "epoch": 0.7485808186435614,
      "grad_norm": 0.14913740754127502,
      "learning_rate": 2.9689267009538845e-05,
      "loss": 0.3532,
      "step": 5011
    },
    {
      "epoch": 0.7487302061547655,
      "grad_norm": 0.20600877702236176,
      "learning_rate": 2.9655852642978567e-05,
      "loss": 0.6256,
      "step": 5012
    },
    {
      "epoch": 0.7488795936659696,
      "grad_norm": 0.17830181121826172,
      "learning_rate": 2.962245381655462e-05,
      "loss": 0.6759,
      "step": 5013
    },
    {
      "epoch": 0.7490289811771735,
      "grad_norm": 0.15848462283611298,
      "learning_rate": 2.9589070537645346e-05,
      "loss": 0.6309,
      "step": 5014
    },
    {
      "epoch": 0.7491783686883776,
      "grad_norm": 0.1965680569410324,
      "learning_rate": 2.955570281362564e-05,
      "loss": 0.63,
      "step": 5015
    },
    {
      "epoch": 0.7493277561995817,
      "grad_norm": 0.1585129201412201,
      "learning_rate": 2.952235065186697e-05,
      "loss": 0.8364,
      "step": 5016
    },
    {
      "epoch": 0.7494771437107858,
      "grad_norm": 0.24089588224887848,
      "learning_rate": 2.948901405973735e-05,
      "loss": 0.8318,
      "step": 5017
    },
    {
      "epoch": 0.7496265312219899,
      "grad_norm": 0.25036704540252686,
      "learning_rate": 2.945569304460136e-05,
      "loss": 0.7252,
      "step": 5018
    },
    {
      "epoch": 0.7497759187331939,
      "grad_norm": 0.1384117603302002,
      "learning_rate": 2.9422387613820134e-05,
      "loss": 0.5636,
      "step": 5019
    },
    {
      "epoch": 0.749925306244398,
      "grad_norm": 0.254249632358551,
      "learning_rate": 2.9389097774751416e-05,
      "loss": 0.5831,
      "step": 5020
    },
    {
      "epoch": 0.750074693755602,
      "grad_norm": 0.13221363723278046,
      "learning_rate": 2.9355823534749383e-05,
      "loss": 0.4262,
      "step": 5021
    },
    {
      "epoch": 0.7502240812668061,
      "grad_norm": 0.13553647696971893,
      "learning_rate": 2.9322564901164872e-05,
      "loss": 0.7986,
      "step": 5022
    },
    {
      "epoch": 0.7503734687780101,
      "grad_norm": 0.17351311445236206,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 0.727,
      "step": 5023
    },
    {
      "epoch": 0.7505228562892142,
      "grad_norm": 0.2165457159280777,
      "learning_rate": 2.9256094482634433e-05,
      "loss": 1.1298,
      "step": 5024
    },
    {
      "epoch": 0.7506722438004183,
      "grad_norm": 0.2038213610649109,
      "learning_rate": 2.9222882712372857e-05,
      "loss": 0.6127,
      "step": 5025
    },
    {
      "epoch": 0.7508216313116224,
      "grad_norm": 0.16468964517116547,
      "learning_rate": 2.9189686577897547e-05,
      "loss": 0.8799,
      "step": 5026
    },
    {
      "epoch": 0.7509710188228265,
      "grad_norm": 0.2717529833316803,
      "learning_rate": 2.915650608654209e-05,
      "loss": 1.1108,
      "step": 5027
    },
    {
      "epoch": 0.7511204063340304,
      "grad_norm": 0.16647474467754364,
      "learning_rate": 2.9123341245636494e-05,
      "loss": 0.81,
      "step": 5028
    },
    {
      "epoch": 0.7512697938452345,
      "grad_norm": 0.22329993546009064,
      "learning_rate": 2.909019206250746e-05,
      "loss": 0.7655,
      "step": 5029
    },
    {
      "epoch": 0.7514191813564386,
      "grad_norm": 0.16364027559757233,
      "learning_rate": 2.9057058544478144e-05,
      "loss": 0.6906,
      "step": 5030
    },
    {
      "epoch": 0.7515685688676427,
      "grad_norm": 0.10985849797725677,
      "learning_rate": 2.9023940698868268e-05,
      "loss": 0.6275,
      "step": 5031
    },
    {
      "epoch": 0.7517179563788468,
      "grad_norm": 0.20116040110588074,
      "learning_rate": 2.8990838532994104e-05,
      "loss": 0.8575,
      "step": 5032
    },
    {
      "epoch": 0.7518673438900508,
      "grad_norm": 0.12416386604309082,
      "learning_rate": 2.895775205416843e-05,
      "loss": 0.6485,
      "step": 5033
    },
    {
      "epoch": 0.7520167314012548,
      "grad_norm": 0.12661822140216827,
      "learning_rate": 2.8924681269700582e-05,
      "loss": 0.6243,
      "step": 5034
    },
    {
      "epoch": 0.7521661189124589,
      "grad_norm": 0.16459856927394867,
      "learning_rate": 2.8891626186896415e-05,
      "loss": 0.3861,
      "step": 5035
    },
    {
      "epoch": 0.752315506423663,
      "grad_norm": 0.1190391331911087,
      "learning_rate": 2.885858681305832e-05,
      "loss": 0.6817,
      "step": 5036
    },
    {
      "epoch": 0.752464893934867,
      "grad_norm": 0.16120287775993347,
      "learning_rate": 2.8825563155485223e-05,
      "loss": 0.7667,
      "step": 5037
    },
    {
      "epoch": 0.7526142814460711,
      "grad_norm": 0.26034191250801086,
      "learning_rate": 2.8792555221472573e-05,
      "loss": 0.9749,
      "step": 5038
    },
    {
      "epoch": 0.7527636689572752,
      "grad_norm": 0.15702778100967407,
      "learning_rate": 2.875956301831233e-05,
      "loss": 0.7079,
      "step": 5039
    },
    {
      "epoch": 0.7529130564684793,
      "grad_norm": 0.17792493104934692,
      "learning_rate": 2.8726586553293043e-05,
      "loss": 0.66,
      "step": 5040
    },
    {
      "epoch": 0.7530624439796832,
      "grad_norm": 0.17464792728424072,
      "learning_rate": 2.8693625833699677e-05,
      "loss": 0.3282,
      "step": 5041
    },
    {
      "epoch": 0.7532118314908873,
      "grad_norm": 0.19540861248970032,
      "learning_rate": 2.8660680866813782e-05,
      "loss": 0.893,
      "step": 5042
    },
    {
      "epoch": 0.7533612190020914,
      "grad_norm": 0.37740805745124817,
      "learning_rate": 2.862775165991346e-05,
      "loss": 0.8963,
      "step": 5043
    },
    {
      "epoch": 0.7535106065132955,
      "grad_norm": 0.1216905266046524,
      "learning_rate": 2.8594838220273256e-05,
      "loss": 0.5721,
      "step": 5044
    },
    {
      "epoch": 0.7536599940244996,
      "grad_norm": 0.14260949194431305,
      "learning_rate": 2.85619405551643e-05,
      "loss": 0.8457,
      "step": 5045
    },
    {
      "epoch": 0.7538093815357036,
      "grad_norm": 0.17160159349441528,
      "learning_rate": 2.8529058671854224e-05,
      "loss": 0.7581,
      "step": 5046
    },
    {
      "epoch": 0.7539587690469077,
      "grad_norm": 0.1480349749326706,
      "learning_rate": 2.8496192577607096e-05,
      "loss": 0.5601,
      "step": 5047
    },
    {
      "epoch": 0.7541081565581117,
      "grad_norm": 0.4231443703174591,
      "learning_rate": 2.846334227968359e-05,
      "loss": 1.2129,
      "step": 5048
    },
    {
      "epoch": 0.7542575440693158,
      "grad_norm": 0.13939209282398224,
      "learning_rate": 2.8430507785340855e-05,
      "loss": 0.5927,
      "step": 5049
    },
    {
      "epoch": 0.7544069315805199,
      "grad_norm": 0.4725044071674347,
      "learning_rate": 2.8397689101832558e-05,
      "loss": 1.0795,
      "step": 5050
    },
    {
      "epoch": 0.7545563190917239,
      "grad_norm": 0.12156057357788086,
      "learning_rate": 2.836488623640885e-05,
      "loss": 0.6953,
      "step": 5051
    },
    {
      "epoch": 0.754705706602928,
      "grad_norm": 0.15424707531929016,
      "learning_rate": 2.8332099196316386e-05,
      "loss": 0.6856,
      "step": 5052
    },
    {
      "epoch": 0.7548550941141321,
      "grad_norm": 0.2183186113834381,
      "learning_rate": 2.8299327988798442e-05,
      "loss": 0.791,
      "step": 5053
    },
    {
      "epoch": 0.7550044816253362,
      "grad_norm": 0.1216680184006691,
      "learning_rate": 2.8266572621094588e-05,
      "loss": 0.7444,
      "step": 5054
    },
    {
      "epoch": 0.7551538691365401,
      "grad_norm": 0.11926071345806122,
      "learning_rate": 2.8233833100441042e-05,
      "loss": 0.7031,
      "step": 5055
    },
    {
      "epoch": 0.7553032566477442,
      "grad_norm": 0.14066322147846222,
      "learning_rate": 2.8201109434070482e-05,
      "loss": 0.6756,
      "step": 5056
    },
    {
      "epoch": 0.7554526441589483,
      "grad_norm": 0.2866126298904419,
      "learning_rate": 2.816840162921208e-05,
      "loss": 0.8306,
      "step": 5057
    },
    {
      "epoch": 0.7556020316701524,
      "grad_norm": 0.2710447311401367,
      "learning_rate": 2.8135709693091516e-05,
      "loss": 0.479,
      "step": 5058
    },
    {
      "epoch": 0.7557514191813565,
      "grad_norm": 0.22824953496456146,
      "learning_rate": 2.8103033632930976e-05,
      "loss": 0.9723,
      "step": 5059
    },
    {
      "epoch": 0.7559008066925605,
      "grad_norm": 0.12431114166975021,
      "learning_rate": 2.807037345594907e-05,
      "loss": 0.6896,
      "step": 5060
    },
    {
      "epoch": 0.7560501942037645,
      "grad_norm": 0.14889276027679443,
      "learning_rate": 2.8037729169360972e-05,
      "loss": 0.904,
      "step": 5061
    },
    {
      "epoch": 0.7561995817149686,
      "grad_norm": 0.25204998254776,
      "learning_rate": 2.8005100780378323e-05,
      "loss": 0.7539,
      "step": 5062
    },
    {
      "epoch": 0.7563489692261727,
      "grad_norm": 0.2646719515323639,
      "learning_rate": 2.7972488296209266e-05,
      "loss": 0.9927,
      "step": 5063
    },
    {
      "epoch": 0.7564983567373768,
      "grad_norm": 0.18368102610111237,
      "learning_rate": 2.793989172405839e-05,
      "loss": 0.8017,
      "step": 5064
    },
    {
      "epoch": 0.7566477442485808,
      "grad_norm": 0.14722304046154022,
      "learning_rate": 2.790731107112683e-05,
      "loss": 0.5435,
      "step": 5065
    },
    {
      "epoch": 0.7567971317597849,
      "grad_norm": 0.1308070868253708,
      "learning_rate": 2.7874746344612114e-05,
      "loss": 0.9281,
      "step": 5066
    },
    {
      "epoch": 0.756946519270989,
      "grad_norm": 0.16848422586917877,
      "learning_rate": 2.784219755170835e-05,
      "loss": 0.6476,
      "step": 5067
    },
    {
      "epoch": 0.757095906782193,
      "grad_norm": 0.15280523896217346,
      "learning_rate": 2.780966469960602e-05,
      "loss": 0.4845,
      "step": 5068
    },
    {
      "epoch": 0.757245294293397,
      "grad_norm": 0.14833232760429382,
      "learning_rate": 2.777714779549223e-05,
      "loss": 0.6475,
      "step": 5069
    },
    {
      "epoch": 0.7573946818046011,
      "grad_norm": 0.19221706688404083,
      "learning_rate": 2.7744646846550448e-05,
      "loss": 0.709,
      "step": 5070
    },
    {
      "epoch": 0.7575440693158052,
      "grad_norm": 0.10112302005290985,
      "learning_rate": 2.771216185996064e-05,
      "loss": 0.5177,
      "step": 5071
    },
    {
      "epoch": 0.7576934568270093,
      "grad_norm": 0.13837836682796478,
      "learning_rate": 2.7679692842899284e-05,
      "loss": 0.8419,
      "step": 5072
    },
    {
      "epoch": 0.7578428443382134,
      "grad_norm": 0.15775489807128906,
      "learning_rate": 2.764723980253925e-05,
      "loss": 0.4381,
      "step": 5073
    },
    {
      "epoch": 0.7579922318494174,
      "grad_norm": 0.13899023830890656,
      "learning_rate": 2.7614802746049938e-05,
      "loss": 0.9004,
      "step": 5074
    },
    {
      "epoch": 0.7581416193606214,
      "grad_norm": 0.16276732087135315,
      "learning_rate": 2.7582381680597234e-05,
      "loss": 0.7165,
      "step": 5075
    },
    {
      "epoch": 0.7582910068718255,
      "grad_norm": 0.13076668977737427,
      "learning_rate": 2.7549976613343452e-05,
      "loss": 0.5737,
      "step": 5076
    },
    {
      "epoch": 0.7584403943830296,
      "grad_norm": 0.20137935876846313,
      "learning_rate": 2.7517587551447387e-05,
      "loss": 0.7752,
      "step": 5077
    },
    {
      "epoch": 0.7585897818942337,
      "grad_norm": 0.16680040955543518,
      "learning_rate": 2.7485214502064316e-05,
      "loss": 0.6651,
      "step": 5078
    },
    {
      "epoch": 0.7587391694054377,
      "grad_norm": 0.15458594262599945,
      "learning_rate": 2.7452857472345916e-05,
      "loss": 0.529,
      "step": 5079
    },
    {
      "epoch": 0.7588885569166418,
      "grad_norm": 0.14064569771289825,
      "learning_rate": 2.7420516469440384e-05,
      "loss": 0.9084,
      "step": 5080
    },
    {
      "epoch": 0.7590379444278458,
      "grad_norm": 0.22690708935260773,
      "learning_rate": 2.7388191500492356e-05,
      "loss": 0.6977,
      "step": 5081
    },
    {
      "epoch": 0.7591873319390499,
      "grad_norm": 0.15593421459197998,
      "learning_rate": 2.7355882572642944e-05,
      "loss": 0.6772,
      "step": 5082
    },
    {
      "epoch": 0.7593367194502539,
      "grad_norm": 0.21101237833499908,
      "learning_rate": 2.7323589693029693e-05,
      "loss": 0.5738,
      "step": 5083
    },
    {
      "epoch": 0.759486106961458,
      "grad_norm": 0.2022375911474228,
      "learning_rate": 2.7291312868786624e-05,
      "loss": 0.9959,
      "step": 5084
    },
    {
      "epoch": 0.7596354944726621,
      "grad_norm": 0.12789630889892578,
      "learning_rate": 2.7259052107044168e-05,
      "loss": 0.8434,
      "step": 5085
    },
    {
      "epoch": 0.7597848819838662,
      "grad_norm": 0.3459501266479492,
      "learning_rate": 2.7226807414929278e-05,
      "loss": 0.9457,
      "step": 5086
    },
    {
      "epoch": 0.7599342694950703,
      "grad_norm": 0.1309424489736557,
      "learning_rate": 2.7194578799565284e-05,
      "loss": 0.4725,
      "step": 5087
    },
    {
      "epoch": 0.7600836570062742,
      "grad_norm": 0.21480251848697662,
      "learning_rate": 2.7162366268072026e-05,
      "loss": 0.773,
      "step": 5088
    },
    {
      "epoch": 0.7602330445174783,
      "grad_norm": 0.16052570939064026,
      "learning_rate": 2.7130169827565733e-05,
      "loss": 0.7111,
      "step": 5089
    },
    {
      "epoch": 0.7603824320286824,
      "grad_norm": 0.1763329803943634,
      "learning_rate": 2.7097989485159137e-05,
      "loss": 0.4306,
      "step": 5090
    },
    {
      "epoch": 0.7605318195398865,
      "grad_norm": 0.40726637840270996,
      "learning_rate": 2.7065825247961387e-05,
      "loss": 0.7217,
      "step": 5091
    },
    {
      "epoch": 0.7606812070510905,
      "grad_norm": 0.13224910199642181,
      "learning_rate": 2.703367712307804e-05,
      "loss": 0.9143,
      "step": 5092
    },
    {
      "epoch": 0.7608305945622946,
      "grad_norm": 0.13229981064796448,
      "learning_rate": 2.7001545117611137e-05,
      "loss": 0.853,
      "step": 5093
    },
    {
      "epoch": 0.7609799820734987,
      "grad_norm": 0.180668905377388,
      "learning_rate": 2.696942923865915e-05,
      "loss": 0.5079,
      "step": 5094
    },
    {
      "epoch": 0.7611293695847027,
      "grad_norm": 0.28195956349372864,
      "learning_rate": 2.6937329493317e-05,
      "loss": 0.8496,
      "step": 5095
    },
    {
      "epoch": 0.7612787570959068,
      "grad_norm": 0.28360915184020996,
      "learning_rate": 2.6905245888676012e-05,
      "loss": 0.9196,
      "step": 5096
    },
    {
      "epoch": 0.7614281446071108,
      "grad_norm": 0.15810814499855042,
      "learning_rate": 2.6873178431823965e-05,
      "loss": 0.6509,
      "step": 5097
    },
    {
      "epoch": 0.7615775321183149,
      "grad_norm": 0.117258720099926,
      "learning_rate": 2.68411271298451e-05,
      "loss": 0.7053,
      "step": 5098
    },
    {
      "epoch": 0.761726919629519,
      "grad_norm": 0.1758682131767273,
      "learning_rate": 2.680909198982e-05,
      "loss": 0.7691,
      "step": 5099
    },
    {
      "epoch": 0.7618763071407231,
      "grad_norm": 0.11233513057231903,
      "learning_rate": 2.6777073018825772e-05,
      "loss": 0.5313,
      "step": 5100
    },
    {
      "epoch": 0.762025694651927,
      "grad_norm": 0.15412203967571259,
      "learning_rate": 2.6745070223935898e-05,
      "loss": 0.5094,
      "step": 5101
    },
    {
      "epoch": 0.7621750821631311,
      "grad_norm": 0.16605402529239655,
      "learning_rate": 2.6713083612220314e-05,
      "loss": 0.3882,
      "step": 5102
    },
    {
      "epoch": 0.7623244696743352,
      "grad_norm": 0.19472065567970276,
      "learning_rate": 2.6681113190745355e-05,
      "loss": 0.8825,
      "step": 5103
    },
    {
      "epoch": 0.7624738571855393,
      "grad_norm": 0.26733240485191345,
      "learning_rate": 2.6649158966573817e-05,
      "loss": 0.7177,
      "step": 5104
    },
    {
      "epoch": 0.7626232446967434,
      "grad_norm": 0.21994580328464508,
      "learning_rate": 2.6617220946764874e-05,
      "loss": 0.6592,
      "step": 5105
    },
    {
      "epoch": 0.7627726322079474,
      "grad_norm": 0.1439264863729477,
      "learning_rate": 2.6585299138374143e-05,
      "loss": 0.5055,
      "step": 5106
    },
    {
      "epoch": 0.7629220197191515,
      "grad_norm": 0.14561933279037476,
      "learning_rate": 2.6553393548453664e-05,
      "loss": 1.0097,
      "step": 5107
    },
    {
      "epoch": 0.7630714072303555,
      "grad_norm": 0.40974053740501404,
      "learning_rate": 2.6521504184051892e-05,
      "loss": 1.2461,
      "step": 5108
    },
    {
      "epoch": 0.7632207947415596,
      "grad_norm": 0.24454110860824585,
      "learning_rate": 2.6489631052213672e-05,
      "loss": 0.4965,
      "step": 5109
    },
    {
      "epoch": 0.7633701822527637,
      "grad_norm": 0.16711658239364624,
      "learning_rate": 2.6457774159980307e-05,
      "loss": 0.858,
      "step": 5110
    },
    {
      "epoch": 0.7635195697639677,
      "grad_norm": 0.1509416103363037,
      "learning_rate": 2.64259335143895e-05,
      "loss": 0.6802,
      "step": 5111
    },
    {
      "epoch": 0.7636689572751718,
      "grad_norm": 0.1979948729276657,
      "learning_rate": 2.639410912247531e-05,
      "loss": 1.0477,
      "step": 5112
    },
    {
      "epoch": 0.7638183447863759,
      "grad_norm": 0.1650601327419281,
      "learning_rate": 2.6362300991268276e-05,
      "loss": 0.4915,
      "step": 5113
    },
    {
      "epoch": 0.76396773229758,
      "grad_norm": 0.36566370725631714,
      "learning_rate": 2.63305091277953e-05,
      "loss": 0.9499,
      "step": 5114
    },
    {
      "epoch": 0.7641171198087839,
      "grad_norm": 0.18968114256858826,
      "learning_rate": 2.6298733539079733e-05,
      "loss": 0.581,
      "step": 5115
    },
    {
      "epoch": 0.764266507319988,
      "grad_norm": 0.21526873111724854,
      "learning_rate": 2.6266974232141285e-05,
      "loss": 0.7202,
      "step": 5116
    },
    {
      "epoch": 0.7644158948311921,
      "grad_norm": 0.20729364454746246,
      "learning_rate": 2.6235231213996136e-05,
      "loss": 0.9346,
      "step": 5117
    },
    {
      "epoch": 0.7645652823423962,
      "grad_norm": 0.12529149651527405,
      "learning_rate": 2.620350449165676e-05,
      "loss": 0.4966,
      "step": 5118
    },
    {
      "epoch": 0.7647146698536003,
      "grad_norm": 0.2524208426475525,
      "learning_rate": 2.6171794072132116e-05,
      "loss": 0.8387,
      "step": 5119
    },
    {
      "epoch": 0.7648640573648043,
      "grad_norm": 0.13706377148628235,
      "learning_rate": 2.6140099962427533e-05,
      "loss": 0.6079,
      "step": 5120
    },
    {
      "epoch": 0.7650134448760084,
      "grad_norm": 0.14956916868686676,
      "learning_rate": 2.6108422169544757e-05,
      "loss": 0.7591,
      "step": 5121
    },
    {
      "epoch": 0.7651628323872124,
      "grad_norm": 0.13669827580451965,
      "learning_rate": 2.6076760700481893e-05,
      "loss": 0.7305,
      "step": 5122
    },
    {
      "epoch": 0.7653122198984165,
      "grad_norm": 0.16515716910362244,
      "learning_rate": 2.604511556223349e-05,
      "loss": 0.6412,
      "step": 5123
    },
    {
      "epoch": 0.7654616074096205,
      "grad_norm": 0.19598056375980377,
      "learning_rate": 2.6013486761790427e-05,
      "loss": 1.0272,
      "step": 5124
    },
    {
      "epoch": 0.7656109949208246,
      "grad_norm": 0.14166109263896942,
      "learning_rate": 2.598187430614004e-05,
      "loss": 0.6906,
      "step": 5125
    },
    {
      "epoch": 0.7657603824320287,
      "grad_norm": 0.23055726289749146,
      "learning_rate": 2.5950278202265997e-05,
      "loss": 0.6188,
      "step": 5126
    },
    {
      "epoch": 0.7659097699432328,
      "grad_norm": 0.29765239357948303,
      "learning_rate": 2.5918698457148392e-05,
      "loss": 0.8884,
      "step": 5127
    },
    {
      "epoch": 0.7660591574544368,
      "grad_norm": 0.14606207609176636,
      "learning_rate": 2.588713507776368e-05,
      "loss": 0.6376,
      "step": 5128
    },
    {
      "epoch": 0.7662085449656408,
      "grad_norm": 0.1387065351009369,
      "learning_rate": 2.5855588071084713e-05,
      "loss": 0.5662,
      "step": 5129
    },
    {
      "epoch": 0.7663579324768449,
      "grad_norm": 0.2773972451686859,
      "learning_rate": 2.582405744408076e-05,
      "loss": 0.8767,
      "step": 5130
    },
    {
      "epoch": 0.766507319988049,
      "grad_norm": 0.229660302400589,
      "learning_rate": 2.579254320371738e-05,
      "loss": 0.8365,
      "step": 5131
    },
    {
      "epoch": 0.7666567074992531,
      "grad_norm": 0.17010971903800964,
      "learning_rate": 2.5761045356956593e-05,
      "loss": 0.8837,
      "step": 5132
    },
    {
      "epoch": 0.7668060950104572,
      "grad_norm": 0.32914021611213684,
      "learning_rate": 2.572956391075676e-05,
      "loss": 0.9372,
      "step": 5133
    },
    {
      "epoch": 0.7669554825216612,
      "grad_norm": 0.16494086384773254,
      "learning_rate": 2.5698098872072652e-05,
      "loss": 0.5333,
      "step": 5134
    },
    {
      "epoch": 0.7671048700328652,
      "grad_norm": 0.18340666592121124,
      "learning_rate": 2.5666650247855385e-05,
      "loss": 0.7873,
      "step": 5135
    },
    {
      "epoch": 0.7672542575440693,
      "grad_norm": 0.19218583405017853,
      "learning_rate": 2.5635218045052477e-05,
      "loss": 0.6675,
      "step": 5136
    },
    {
      "epoch": 0.7674036450552734,
      "grad_norm": 0.22187642753124237,
      "learning_rate": 2.5603802270607758e-05,
      "loss": 0.6774,
      "step": 5137
    },
    {
      "epoch": 0.7675530325664774,
      "grad_norm": 0.1874605417251587,
      "learning_rate": 2.5572402931461493e-05,
      "loss": 0.8474,
      "step": 5138
    },
    {
      "epoch": 0.7677024200776815,
      "grad_norm": 0.23976771533489227,
      "learning_rate": 2.554102003455029e-05,
      "loss": 1.2521,
      "step": 5139
    },
    {
      "epoch": 0.7678518075888856,
      "grad_norm": 0.12778620421886444,
      "learning_rate": 2.5509653586807125e-05,
      "loss": 0.5907,
      "step": 5140
    },
    {
      "epoch": 0.7680011951000897,
      "grad_norm": 0.2607424259185791,
      "learning_rate": 2.5478303595161335e-05,
      "loss": 0.9782,
      "step": 5141
    },
    {
      "epoch": 0.7681505826112937,
      "grad_norm": 0.18479058146476746,
      "learning_rate": 2.5446970066538656e-05,
      "loss": 0.7662,
      "step": 5142
    },
    {
      "epoch": 0.7682999701224977,
      "grad_norm": 0.1848878711462021,
      "learning_rate": 2.5415653007861183e-05,
      "loss": 0.5662,
      "step": 5143
    },
    {
      "epoch": 0.7684493576337018,
      "grad_norm": 0.16110166907310486,
      "learning_rate": 2.53843524260473e-05,
      "loss": 0.7528,
      "step": 5144
    },
    {
      "epoch": 0.7685987451449059,
      "grad_norm": 0.2282876819372177,
      "learning_rate": 2.5353068328011796e-05,
      "loss": 0.704,
      "step": 5145
    },
    {
      "epoch": 0.76874813265611,
      "grad_norm": 0.16270409524440765,
      "learning_rate": 2.5321800720665856e-05,
      "loss": 0.5344,
      "step": 5146
    },
    {
      "epoch": 0.768897520167314,
      "grad_norm": 0.15513327717781067,
      "learning_rate": 2.529054961091697e-05,
      "loss": 0.7261,
      "step": 5147
    },
    {
      "epoch": 0.769046907678518,
      "grad_norm": 0.10703502595424652,
      "learning_rate": 2.5259315005669027e-05,
      "loss": 0.5737,
      "step": 5148
    },
    {
      "epoch": 0.7691962951897221,
      "grad_norm": 0.24275276064872742,
      "learning_rate": 2.5228096911822242e-05,
      "loss": 0.7815,
      "step": 5149
    },
    {
      "epoch": 0.7693456827009262,
      "grad_norm": 0.14641046524047852,
      "learning_rate": 2.5196895336273163e-05,
      "loss": 0.8073,
      "step": 5150
    },
    {
      "epoch": 0.7694950702121303,
      "grad_norm": 0.14365264773368835,
      "learning_rate": 2.5165710285914714e-05,
      "loss": 0.8219,
      "step": 5151
    },
    {
      "epoch": 0.7696444577233343,
      "grad_norm": 0.13880550861358643,
      "learning_rate": 2.513454176763618e-05,
      "loss": 0.6761,
      "step": 5152
    },
    {
      "epoch": 0.7697938452345384,
      "grad_norm": 0.14273954927921295,
      "learning_rate": 2.5103389788323173e-05,
      "loss": 0.8194,
      "step": 5153
    },
    {
      "epoch": 0.7699432327457425,
      "grad_norm": 0.1752299815416336,
      "learning_rate": 2.507225435485766e-05,
      "loss": 0.7974,
      "step": 5154
    },
    {
      "epoch": 0.7700926202569465,
      "grad_norm": 0.34338995814323425,
      "learning_rate": 2.504113547411796e-05,
      "loss": 0.7461,
      "step": 5155
    },
    {
      "epoch": 0.7702420077681505,
      "grad_norm": 0.1982237845659256,
      "learning_rate": 2.501003315297875e-05,
      "loss": 0.9637,
      "step": 5156
    },
    {
      "epoch": 0.7703913952793546,
      "grad_norm": 0.31925347447395325,
      "learning_rate": 2.4978947398310936e-05,
      "loss": 1.0552,
      "step": 5157
    },
    {
      "epoch": 0.7705407827905587,
      "grad_norm": 0.5262662172317505,
      "learning_rate": 2.4947878216981945e-05,
      "loss": 1.1285,
      "step": 5158
    },
    {
      "epoch": 0.7706901703017628,
      "grad_norm": 0.13540929555892944,
      "learning_rate": 2.491682561585542e-05,
      "loss": 0.791,
      "step": 5159
    },
    {
      "epoch": 0.7708395578129669,
      "grad_norm": 0.17319613695144653,
      "learning_rate": 2.4885789601791364e-05,
      "loss": 1.1156,
      "step": 5160
    },
    {
      "epoch": 0.770988945324171,
      "grad_norm": 0.14595972001552582,
      "learning_rate": 2.4854770181646147e-05,
      "loss": 0.6963,
      "step": 5161
    },
    {
      "epoch": 0.7711383328353749,
      "grad_norm": 0.22043079137802124,
      "learning_rate": 2.4823767362272455e-05,
      "loss": 0.6215,
      "step": 5162
    },
    {
      "epoch": 0.771287720346579,
      "grad_norm": 0.2202375829219818,
      "learning_rate": 2.479278115051925e-05,
      "loss": 0.6789,
      "step": 5163
    },
    {
      "epoch": 0.7714371078577831,
      "grad_norm": 0.2979015111923218,
      "learning_rate": 2.4761811553231916e-05,
      "loss": 0.8857,
      "step": 5164
    },
    {
      "epoch": 0.7715864953689872,
      "grad_norm": 0.16109201312065125,
      "learning_rate": 2.4730858577252126e-05,
      "loss": 0.8636,
      "step": 5165
    },
    {
      "epoch": 0.7717358828801912,
      "grad_norm": 0.15835662186145782,
      "learning_rate": 2.469992222941787e-05,
      "loss": 0.5455,
      "step": 5166
    },
    {
      "epoch": 0.7718852703913953,
      "grad_norm": 0.13439109921455383,
      "learning_rate": 2.466900251656349e-05,
      "loss": 0.6949,
      "step": 5167
    },
    {
      "epoch": 0.7720346579025994,
      "grad_norm": 0.1309661865234375,
      "learning_rate": 2.4638099445519636e-05,
      "loss": 0.5049,
      "step": 5168
    },
    {
      "epoch": 0.7721840454138034,
      "grad_norm": 0.1377398520708084,
      "learning_rate": 2.460721302311332e-05,
      "loss": 0.8167,
      "step": 5169
    },
    {
      "epoch": 0.7723334329250074,
      "grad_norm": 0.26614993810653687,
      "learning_rate": 2.4576343256167766e-05,
      "loss": 0.9529,
      "step": 5170
    },
    {
      "epoch": 0.7724828204362115,
      "grad_norm": 0.155905082821846,
      "learning_rate": 2.4545490151502637e-05,
      "loss": 0.6286,
      "step": 5171
    },
    {
      "epoch": 0.7726322079474156,
      "grad_norm": 0.23148490488529205,
      "learning_rate": 2.4514653715933876e-05,
      "loss": 0.8397,
      "step": 5172
    },
    {
      "epoch": 0.7727815954586197,
      "grad_norm": 0.11777471750974655,
      "learning_rate": 2.4483833956273727e-05,
      "loss": 0.5558,
      "step": 5173
    },
    {
      "epoch": 0.7729309829698238,
      "grad_norm": 0.16988354921340942,
      "learning_rate": 2.4453030879330784e-05,
      "loss": 0.7409,
      "step": 5174
    },
    {
      "epoch": 0.7730803704810277,
      "grad_norm": 0.13410671055316925,
      "learning_rate": 2.4422244491909917e-05,
      "loss": 0.7973,
      "step": 5175
    },
    {
      "epoch": 0.7732297579922318,
      "grad_norm": 0.17080184817314148,
      "learning_rate": 2.4391474800812332e-05,
      "loss": 0.7237,
      "step": 5176
    },
    {
      "epoch": 0.7733791455034359,
      "grad_norm": 0.22488288581371307,
      "learning_rate": 2.436072181283553e-05,
      "loss": 0.4184,
      "step": 5177
    },
    {
      "epoch": 0.77352853301464,
      "grad_norm": 0.13836485147476196,
      "learning_rate": 2.4329985534773358e-05,
      "loss": 0.726,
      "step": 5178
    },
    {
      "epoch": 0.773677920525844,
      "grad_norm": 0.24351896345615387,
      "learning_rate": 2.429926597341593e-05,
      "loss": 0.9226,
      "step": 5179
    },
    {
      "epoch": 0.7738273080370481,
      "grad_norm": 0.3098902404308319,
      "learning_rate": 2.4268563135549682e-05,
      "loss": 1.26,
      "step": 5180
    },
    {
      "epoch": 0.7739766955482522,
      "grad_norm": 0.13766397535800934,
      "learning_rate": 2.4237877027957357e-05,
      "loss": 0.7987,
      "step": 5181
    },
    {
      "epoch": 0.7741260830594562,
      "grad_norm": 0.15389464795589447,
      "learning_rate": 2.4207207657418042e-05,
      "loss": 0.7006,
      "step": 5182
    },
    {
      "epoch": 0.7742754705706603,
      "grad_norm": 0.19149081408977509,
      "learning_rate": 2.4176555030707026e-05,
      "loss": 0.5554,
      "step": 5183
    },
    {
      "epoch": 0.7744248580818643,
      "grad_norm": 0.12880487740039825,
      "learning_rate": 2.4145919154595975e-05,
      "loss": 0.9374,
      "step": 5184
    },
    {
      "epoch": 0.7745742455930684,
      "grad_norm": 0.30477726459503174,
      "learning_rate": 2.411530003585286e-05,
      "loss": 0.6865,
      "step": 5185
    },
    {
      "epoch": 0.7747236331042725,
      "grad_norm": 0.32513925433158875,
      "learning_rate": 2.4084697681241906e-05,
      "loss": 0.7726,
      "step": 5186
    },
    {
      "epoch": 0.7748730206154766,
      "grad_norm": 0.1700081080198288,
      "learning_rate": 2.4054112097523684e-05,
      "loss": 0.7069,
      "step": 5187
    },
    {
      "epoch": 0.7750224081266807,
      "grad_norm": 0.1529083400964737,
      "learning_rate": 2.402354329145504e-05,
      "loss": 0.5411,
      "step": 5188
    },
    {
      "epoch": 0.7751717956378846,
      "grad_norm": 0.234103724360466,
      "learning_rate": 2.3992991269789068e-05,
      "loss": 1.1678,
      "step": 5189
    },
    {
      "epoch": 0.7753211831490887,
      "grad_norm": 0.20433910191059113,
      "learning_rate": 2.3962456039275206e-05,
      "loss": 0.9571,
      "step": 5190
    },
    {
      "epoch": 0.7754705706602928,
      "grad_norm": 0.1230558529496193,
      "learning_rate": 2.393193760665917e-05,
      "loss": 0.6836,
      "step": 5191
    },
    {
      "epoch": 0.7756199581714969,
      "grad_norm": 0.21065321564674377,
      "learning_rate": 2.3901435978682986e-05,
      "loss": 0.6115,
      "step": 5192
    },
    {
      "epoch": 0.775769345682701,
      "grad_norm": 0.11225999891757965,
      "learning_rate": 2.387095116208493e-05,
      "loss": 0.5141,
      "step": 5193
    },
    {
      "epoch": 0.775918733193905,
      "grad_norm": 0.11128906160593033,
      "learning_rate": 2.3840483163599582e-05,
      "loss": 0.715,
      "step": 5194
    },
    {
      "epoch": 0.776068120705109,
      "grad_norm": 0.17481008172035217,
      "learning_rate": 2.3810031989957803e-05,
      "loss": 0.7704,
      "step": 5195
    },
    {
      "epoch": 0.7762175082163131,
      "grad_norm": 0.13672198355197906,
      "learning_rate": 2.3779597647886753e-05,
      "loss": 0.8391,
      "step": 5196
    },
    {
      "epoch": 0.7763668957275172,
      "grad_norm": 0.1717597395181656,
      "learning_rate": 2.374918014410985e-05,
      "loss": 0.6559,
      "step": 5197
    },
    {
      "epoch": 0.7765162832387212,
      "grad_norm": 0.12831032276153564,
      "learning_rate": 2.37187794853468e-05,
      "loss": 0.7115,
      "step": 5198
    },
    {
      "epoch": 0.7766656707499253,
      "grad_norm": 0.15792694687843323,
      "learning_rate": 2.368839567831359e-05,
      "loss": 0.3384,
      "step": 5199
    },
    {
      "epoch": 0.7768150582611294,
      "grad_norm": 0.13054996728897095,
      "learning_rate": 2.3658028729722502e-05,
      "loss": 0.6383,
      "step": 5200
    },
    {
      "epoch": 0.7769644457723335,
      "grad_norm": 0.2442476451396942,
      "learning_rate": 2.362767864628208e-05,
      "loss": 0.9469,
      "step": 5201
    },
    {
      "epoch": 0.7771138332835374,
      "grad_norm": 0.2185201495885849,
      "learning_rate": 2.3597345434697093e-05,
      "loss": 0.4848,
      "step": 5202
    },
    {
      "epoch": 0.7772632207947415,
      "grad_norm": 0.15739725530147552,
      "learning_rate": 2.356702910166866e-05,
      "loss": 0.6661,
      "step": 5203
    },
    {
      "epoch": 0.7774126083059456,
      "grad_norm": 0.16388019919395447,
      "learning_rate": 2.3536729653894118e-05,
      "loss": 0.6237,
      "step": 5204
    },
    {
      "epoch": 0.7775619958171497,
      "grad_norm": 0.39650067687034607,
      "learning_rate": 2.350644709806713e-05,
      "loss": 1.3265,
      "step": 5205
    },
    {
      "epoch": 0.7777113833283538,
      "grad_norm": 0.19836650788784027,
      "learning_rate": 2.3476181440877564e-05,
      "loss": 0.9356,
      "step": 5206
    },
    {
      "epoch": 0.7778607708395578,
      "grad_norm": 0.12467662245035172,
      "learning_rate": 2.344593268901162e-05,
      "loss": 0.5258,
      "step": 5207
    },
    {
      "epoch": 0.7780101583507619,
      "grad_norm": 0.14233796298503876,
      "learning_rate": 2.341570084915168e-05,
      "loss": 0.764,
      "step": 5208
    },
    {
      "epoch": 0.7781595458619659,
      "grad_norm": 0.1997152417898178,
      "learning_rate": 2.3385485927976446e-05,
      "loss": 0.3713,
      "step": 5209
    },
    {
      "epoch": 0.77830893337317,
      "grad_norm": 0.18228153884410858,
      "learning_rate": 2.3355287932160884e-05,
      "loss": 0.8753,
      "step": 5210
    },
    {
      "epoch": 0.778458320884374,
      "grad_norm": 0.15264804661273956,
      "learning_rate": 2.33251068683762e-05,
      "loss": 0.9123,
      "step": 5211
    },
    {
      "epoch": 0.7786077083955781,
      "grad_norm": 0.12046895176172256,
      "learning_rate": 2.329494274328988e-05,
      "loss": 0.7284,
      "step": 5212
    },
    {
      "epoch": 0.7787570959067822,
      "grad_norm": 0.1352137327194214,
      "learning_rate": 2.3264795563565668e-05,
      "loss": 0.6907,
      "step": 5213
    },
    {
      "epoch": 0.7789064834179863,
      "grad_norm": 0.17324700951576233,
      "learning_rate": 2.3234665335863526e-05,
      "loss": 0.4594,
      "step": 5214
    },
    {
      "epoch": 0.7790558709291904,
      "grad_norm": 0.23618392646312714,
      "learning_rate": 2.3204552066839712e-05,
      "loss": 0.424,
      "step": 5215
    },
    {
      "epoch": 0.7792052584403943,
      "grad_norm": 0.17923210561275482,
      "learning_rate": 2.3174455763146717e-05,
      "loss": 0.7902,
      "step": 5216
    },
    {
      "epoch": 0.7793546459515984,
      "grad_norm": 0.1630912721157074,
      "learning_rate": 2.3144376431433314e-05,
      "loss": 0.7237,
      "step": 5217
    },
    {
      "epoch": 0.7795040334628025,
      "grad_norm": 0.19822578132152557,
      "learning_rate": 2.3114314078344478e-05,
      "loss": 0.552,
      "step": 5218
    },
    {
      "epoch": 0.7796534209740066,
      "grad_norm": 0.1707557737827301,
      "learning_rate": 2.3084268710521463e-05,
      "loss": 0.8263,
      "step": 5219
    },
    {
      "epoch": 0.7798028084852107,
      "grad_norm": 0.12634268403053284,
      "learning_rate": 2.3054240334601805e-05,
      "loss": 0.7531,
      "step": 5220
    },
    {
      "epoch": 0.7799521959964147,
      "grad_norm": 0.26757070422172546,
      "learning_rate": 2.3024228957219197e-05,
      "loss": 0.862,
      "step": 5221
    },
    {
      "epoch": 0.7801015835076187,
      "grad_norm": 0.12678813934326172,
      "learning_rate": 2.2994234585003638e-05,
      "loss": 0.3108,
      "step": 5222
    },
    {
      "epoch": 0.7802509710188228,
      "grad_norm": 0.14095012843608856,
      "learning_rate": 2.2964257224581366e-05,
      "loss": 0.9452,
      "step": 5223
    },
    {
      "epoch": 0.7804003585300269,
      "grad_norm": 0.15837283432483673,
      "learning_rate": 2.2934296882574847e-05,
      "loss": 0.7871,
      "step": 5224
    },
    {
      "epoch": 0.780549746041231,
      "grad_norm": 0.22808951139450073,
      "learning_rate": 2.290435356560282e-05,
      "loss": 0.5606,
      "step": 5225
    },
    {
      "epoch": 0.780699133552435,
      "grad_norm": 0.17145612835884094,
      "learning_rate": 2.287442728028021e-05,
      "loss": 0.6108,
      "step": 5226
    },
    {
      "epoch": 0.7808485210636391,
      "grad_norm": 0.20653286576271057,
      "learning_rate": 2.2844518033218254e-05,
      "loss": 1.1661,
      "step": 5227
    },
    {
      "epoch": 0.7809979085748432,
      "grad_norm": 0.12662354111671448,
      "learning_rate": 2.2814625831024318e-05,
      "loss": 0.7967,
      "step": 5228
    },
    {
      "epoch": 0.7811472960860472,
      "grad_norm": 0.12491440027952194,
      "learning_rate": 2.2784750680302093e-05,
      "loss": 0.4305,
      "step": 5229
    },
    {
      "epoch": 0.7812966835972512,
      "grad_norm": 0.1404963731765747,
      "learning_rate": 2.2754892587651434e-05,
      "loss": 0.4858,
      "step": 5230
    },
    {
      "epoch": 0.7814460711084553,
      "grad_norm": 0.23742114007472992,
      "learning_rate": 2.2725051559668532e-05,
      "loss": 0.735,
      "step": 5231
    },
    {
      "epoch": 0.7815954586196594,
      "grad_norm": 0.13371920585632324,
      "learning_rate": 2.2695227602945702e-05,
      "loss": 0.694,
      "step": 5232
    },
    {
      "epoch": 0.7817448461308635,
      "grad_norm": 0.5834226608276367,
      "learning_rate": 2.2665420724071583e-05,
      "loss": 1.1741,
      "step": 5233
    },
    {
      "epoch": 0.7818942336420676,
      "grad_norm": 0.12345851957798004,
      "learning_rate": 2.2635630929630904e-05,
      "loss": 0.6491,
      "step": 5234
    },
    {
      "epoch": 0.7820436211532716,
      "grad_norm": 0.14365260303020477,
      "learning_rate": 2.2605858226204745e-05,
      "loss": 0.5771,
      "step": 5235
    },
    {
      "epoch": 0.7821930086644756,
      "grad_norm": 0.1705927550792694,
      "learning_rate": 2.2576102620370364e-05,
      "loss": 0.5232,
      "step": 5236
    },
    {
      "epoch": 0.7823423961756797,
      "grad_norm": 0.32067492604255676,
      "learning_rate": 2.2546364118701245e-05,
      "loss": 0.9158,
      "step": 5237
    },
    {
      "epoch": 0.7824917836868838,
      "grad_norm": 0.16069325804710388,
      "learning_rate": 2.251664272776709e-05,
      "loss": 0.583,
      "step": 5238
    },
    {
      "epoch": 0.7826411711980878,
      "grad_norm": 0.14831659197807312,
      "learning_rate": 2.2486938454133832e-05,
      "loss": 0.7355,
      "step": 5239
    },
    {
      "epoch": 0.7827905587092919,
      "grad_norm": 0.41942569613456726,
      "learning_rate": 2.2457251304363646e-05,
      "loss": 0.7556,
      "step": 5240
    },
    {
      "epoch": 0.782939946220496,
      "grad_norm": 0.18932239711284637,
      "learning_rate": 2.242758128501482e-05,
      "loss": 0.5326,
      "step": 5241
    },
    {
      "epoch": 0.7830893337317,
      "grad_norm": 0.15653546154499054,
      "learning_rate": 2.2397928402641988e-05,
      "loss": 0.481,
      "step": 5242
    },
    {
      "epoch": 0.7832387212429041,
      "grad_norm": 0.2667771577835083,
      "learning_rate": 2.236829266379592e-05,
      "loss": 1.02,
      "step": 5243
    },
    {
      "epoch": 0.7833881087541081,
      "grad_norm": 0.22265301644802094,
      "learning_rate": 2.2338674075023615e-05,
      "loss": 0.4846,
      "step": 5244
    },
    {
      "epoch": 0.7835374962653122,
      "grad_norm": 0.12368563562631607,
      "learning_rate": 2.2309072642868313e-05,
      "loss": 0.6198,
      "step": 5245
    },
    {
      "epoch": 0.7836868837765163,
      "grad_norm": 0.14013154804706573,
      "learning_rate": 2.2279488373869416e-05,
      "loss": 0.4184,
      "step": 5246
    },
    {
      "epoch": 0.7838362712877204,
      "grad_norm": 0.140771746635437,
      "learning_rate": 2.2249921274562568e-05,
      "loss": 0.6675,
      "step": 5247
    },
    {
      "epoch": 0.7839856587989245,
      "grad_norm": 0.17905528843402863,
      "learning_rate": 2.2220371351479607e-05,
      "loss": 0.5137,
      "step": 5248
    },
    {
      "epoch": 0.7841350463101284,
      "grad_norm": 0.1287696808576584,
      "learning_rate": 2.2190838611148578e-05,
      "loss": 0.6749,
      "step": 5249
    },
    {
      "epoch": 0.7842844338213325,
      "grad_norm": 0.1631043255329132,
      "learning_rate": 2.2161323060093742e-05,
      "loss": 0.6744,
      "step": 5250
    },
    {
      "epoch": 0.7844338213325366,
      "grad_norm": 0.1754007339477539,
      "learning_rate": 2.2131824704835536e-05,
      "loss": 0.723,
      "step": 5251
    },
    {
      "epoch": 0.7845832088437407,
      "grad_norm": 0.24155062437057495,
      "learning_rate": 2.2102343551890627e-05,
      "loss": 0.6506,
      "step": 5252
    },
    {
      "epoch": 0.7847325963549447,
      "grad_norm": 0.14358638226985931,
      "learning_rate": 2.207287960777188e-05,
      "loss": 0.6456,
      "step": 5253
    },
    {
      "epoch": 0.7848819838661488,
      "grad_norm": 0.25092267990112305,
      "learning_rate": 2.2043432878988313e-05,
      "loss": 0.81,
      "step": 5254
    },
    {
      "epoch": 0.7850313713773529,
      "grad_norm": 0.14415691792964935,
      "learning_rate": 2.201400337204519e-05,
      "loss": 0.7501,
      "step": 5255
    },
    {
      "epoch": 0.7851807588885569,
      "grad_norm": 0.1455066055059433,
      "learning_rate": 2.198459109344395e-05,
      "loss": 0.688,
      "step": 5256
    },
    {
      "epoch": 0.785330146399761,
      "grad_norm": 0.12997117638587952,
      "learning_rate": 2.1955196049682235e-05,
      "loss": 0.4462,
      "step": 5257
    },
    {
      "epoch": 0.785479533910965,
      "grad_norm": 0.15429700911045074,
      "learning_rate": 2.1925818247253893e-05,
      "loss": 0.8282,
      "step": 5258
    },
    {
      "epoch": 0.7856289214221691,
      "grad_norm": 0.14819805324077606,
      "learning_rate": 2.1896457692648943e-05,
      "loss": 0.5724,
      "step": 5259
    },
    {
      "epoch": 0.7857783089333732,
      "grad_norm": 0.32732897996902466,
      "learning_rate": 2.186711439235356e-05,
      "loss": 0.7226,
      "step": 5260
    },
    {
      "epoch": 0.7859276964445773,
      "grad_norm": 0.1415766328573227,
      "learning_rate": 2.1837788352850175e-05,
      "loss": 0.4802,
      "step": 5261
    },
    {
      "epoch": 0.7860770839557812,
      "grad_norm": 0.15946222841739655,
      "learning_rate": 2.180847958061737e-05,
      "loss": 0.6933,
      "step": 5262
    },
    {
      "epoch": 0.7862264714669853,
      "grad_norm": 0.16171157360076904,
      "learning_rate": 2.1779188082129907e-05,
      "loss": 0.5153,
      "step": 5263
    },
    {
      "epoch": 0.7863758589781894,
      "grad_norm": 0.2069016695022583,
      "learning_rate": 2.174991386385876e-05,
      "loss": 0.6292,
      "step": 5264
    },
    {
      "epoch": 0.7865252464893935,
      "grad_norm": 0.16005486249923706,
      "learning_rate": 2.1720656932271055e-05,
      "loss": 0.6626,
      "step": 5265
    },
    {
      "epoch": 0.7866746340005976,
      "grad_norm": 0.17377814650535583,
      "learning_rate": 2.169141729383011e-05,
      "loss": 0.8427,
      "step": 5266
    },
    {
      "epoch": 0.7868240215118016,
      "grad_norm": 0.12980499863624573,
      "learning_rate": 2.166219495499543e-05,
      "loss": 0.5545,
      "step": 5267
    },
    {
      "epoch": 0.7869734090230057,
      "grad_norm": 0.1729169338941574,
      "learning_rate": 2.163298992222269e-05,
      "loss": 0.5274,
      "step": 5268
    },
    {
      "epoch": 0.7871227965342097,
      "grad_norm": 0.09537654370069504,
      "learning_rate": 2.160380220196375e-05,
      "loss": 0.2936,
      "step": 5269
    },
    {
      "epoch": 0.7872721840454138,
      "grad_norm": 0.3166316747665405,
      "learning_rate": 2.1574631800666635e-05,
      "loss": 0.9623,
      "step": 5270
    },
    {
      "epoch": 0.7874215715566178,
      "grad_norm": 0.19380927085876465,
      "learning_rate": 2.1545478724775538e-05,
      "loss": 0.6129,
      "step": 5271
    },
    {
      "epoch": 0.7875709590678219,
      "grad_norm": 0.19988813996315002,
      "learning_rate": 2.1516342980730885e-05,
      "loss": 0.8165,
      "step": 5272
    },
    {
      "epoch": 0.787720346579026,
      "grad_norm": 0.14266125857830048,
      "learning_rate": 2.1487224574969155e-05,
      "loss": 0.4703,
      "step": 5273
    },
    {
      "epoch": 0.7878697340902301,
      "grad_norm": 0.13862362504005432,
      "learning_rate": 2.145812351392309e-05,
      "loss": 0.7926,
      "step": 5274
    },
    {
      "epoch": 0.7880191216014342,
      "grad_norm": 0.1776328980922699,
      "learning_rate": 2.1429039804021588e-05,
      "loss": 0.6817,
      "step": 5275
    },
    {
      "epoch": 0.7881685091126381,
      "grad_norm": 0.2444443702697754,
      "learning_rate": 2.1399973451689682e-05,
      "loss": 0.9492,
      "step": 5276
    },
    {
      "epoch": 0.7883178966238422,
      "grad_norm": 0.13802656531333923,
      "learning_rate": 2.1370924463348606e-05,
      "loss": 0.821,
      "step": 5277
    },
    {
      "epoch": 0.7884672841350463,
      "grad_norm": 0.12073404341936111,
      "learning_rate": 2.1341892845415766e-05,
      "loss": 0.5257,
      "step": 5278
    },
    {
      "epoch": 0.7886166716462504,
      "grad_norm": 0.1868436634540558,
      "learning_rate": 2.1312878604304652e-05,
      "loss": 0.943,
      "step": 5279
    },
    {
      "epoch": 0.7887660591574545,
      "grad_norm": 0.23814302682876587,
      "learning_rate": 2.1283881746424982e-05,
      "loss": 0.7913,
      "step": 5280
    },
    {
      "epoch": 0.7889154466686585,
      "grad_norm": 0.2047875076532364,
      "learning_rate": 2.1254902278182643e-05,
      "loss": 0.7433,
      "step": 5281
    },
    {
      "epoch": 0.7890648341798626,
      "grad_norm": 0.17805886268615723,
      "learning_rate": 2.1225940205979645e-05,
      "loss": 0.4809,
      "step": 5282
    },
    {
      "epoch": 0.7892142216910666,
      "grad_norm": 0.17691335082054138,
      "learning_rate": 2.119699553621417e-05,
      "loss": 0.8618,
      "step": 5283
    },
    {
      "epoch": 0.7893636092022707,
      "grad_norm": 0.14697951078414917,
      "learning_rate": 2.1168068275280562e-05,
      "loss": 0.8464,
      "step": 5284
    },
    {
      "epoch": 0.7895129967134747,
      "grad_norm": 0.15303760766983032,
      "learning_rate": 2.1139158429569304e-05,
      "loss": 0.7501,
      "step": 5285
    },
    {
      "epoch": 0.7896623842246788,
      "grad_norm": 0.1326155960559845,
      "learning_rate": 2.111026600546704e-05,
      "loss": 0.778,
      "step": 5286
    },
    {
      "epoch": 0.7898117717358829,
      "grad_norm": 0.12192168831825256,
      "learning_rate": 2.1081391009356566e-05,
      "loss": 0.5006,
      "step": 5287
    },
    {
      "epoch": 0.789961159247087,
      "grad_norm": 0.1505921632051468,
      "learning_rate": 2.1052533447616817e-05,
      "loss": 0.7354,
      "step": 5288
    },
    {
      "epoch": 0.790110546758291,
      "grad_norm": 0.23057404160499573,
      "learning_rate": 2.1023693326622895e-05,
      "loss": 0.7355,
      "step": 5289
    },
    {
      "epoch": 0.790259934269495,
      "grad_norm": 0.1398073434829712,
      "learning_rate": 2.0994870652746045e-05,
      "loss": 1.0309,
      "step": 5290
    },
    {
      "epoch": 0.7904093217806991,
      "grad_norm": 0.13761861622333527,
      "learning_rate": 2.0966065432353667e-05,
      "loss": 0.6886,
      "step": 5291
    },
    {
      "epoch": 0.7905587092919032,
      "grad_norm": 0.15954484045505524,
      "learning_rate": 2.093727767180923e-05,
      "loss": 0.3804,
      "step": 5292
    },
    {
      "epoch": 0.7907080968031073,
      "grad_norm": 0.14123547077178955,
      "learning_rate": 2.090850737747245e-05,
      "loss": 0.7245,
      "step": 5293
    },
    {
      "epoch": 0.7908574843143114,
      "grad_norm": 0.1482001394033432,
      "learning_rate": 2.087975455569915e-05,
      "loss": 0.6888,
      "step": 5294
    },
    {
      "epoch": 0.7910068718255154,
      "grad_norm": 0.1445704996585846,
      "learning_rate": 2.085101921284125e-05,
      "loss": 0.6093,
      "step": 5295
    },
    {
      "epoch": 0.7911562593367194,
      "grad_norm": 0.16729885339736938,
      "learning_rate": 2.0822301355246877e-05,
      "loss": 0.8377,
      "step": 5296
    },
    {
      "epoch": 0.7913056468479235,
      "grad_norm": 0.15319351851940155,
      "learning_rate": 2.0793600989260232e-05,
      "loss": 0.506,
      "step": 5297
    },
    {
      "epoch": 0.7914550343591276,
      "grad_norm": 0.1801481992006302,
      "learning_rate": 2.0764918121221722e-05,
      "loss": 0.6218,
      "step": 5298
    },
    {
      "epoch": 0.7916044218703316,
      "grad_norm": 0.2480681836605072,
      "learning_rate": 2.073625275746779e-05,
      "loss": 0.7721,
      "step": 5299
    },
    {
      "epoch": 0.7917538093815357,
      "grad_norm": 0.21741868555545807,
      "learning_rate": 2.0707604904331103e-05,
      "loss": 0.8283,
      "step": 5300
    },
    {
      "epoch": 0.7919031968927398,
      "grad_norm": 0.28018736839294434,
      "learning_rate": 2.067897456814041e-05,
      "loss": 0.7265,
      "step": 5301
    },
    {
      "epoch": 0.7920525844039439,
      "grad_norm": 0.16300749778747559,
      "learning_rate": 2.0650361755220625e-05,
      "loss": 0.9235,
      "step": 5302
    },
    {
      "epoch": 0.7922019719151479,
      "grad_norm": 0.119844950735569,
      "learning_rate": 2.0621766471892755e-05,
      "loss": 0.4212,
      "step": 5303
    },
    {
      "epoch": 0.7923513594263519,
      "grad_norm": 0.4314511716365814,
      "learning_rate": 2.0593188724473956e-05,
      "loss": 1.141,
      "step": 5304
    },
    {
      "epoch": 0.792500746937556,
      "grad_norm": 0.15806156396865845,
      "learning_rate": 2.0564628519277505e-05,
      "loss": 0.6626,
      "step": 5305
    },
    {
      "epoch": 0.7926501344487601,
      "grad_norm": 0.5810412764549255,
      "learning_rate": 2.053608586261282e-05,
      "loss": 1.5437,
      "step": 5306
    },
    {
      "epoch": 0.7927995219599642,
      "grad_norm": 0.2102261185646057,
      "learning_rate": 2.0507560760785393e-05,
      "loss": 0.6665,
      "step": 5307
    },
    {
      "epoch": 0.7929489094711683,
      "grad_norm": 0.12895916402339935,
      "learning_rate": 2.04790532200969e-05,
      "loss": 1.0067,
      "step": 5308
    },
    {
      "epoch": 0.7930982969823722,
      "grad_norm": 0.14141923189163208,
      "learning_rate": 2.0450563246845102e-05,
      "loss": 0.9321,
      "step": 5309
    },
    {
      "epoch": 0.7932476844935763,
      "grad_norm": 0.3165157735347748,
      "learning_rate": 2.042209084732387e-05,
      "loss": 0.7519,
      "step": 5310
    },
    {
      "epoch": 0.7933970720047804,
      "grad_norm": 0.12752245366573334,
      "learning_rate": 2.039363602782325e-05,
      "loss": 0.8171,
      "step": 5311
    },
    {
      "epoch": 0.7935464595159845,
      "grad_norm": 0.17814117670059204,
      "learning_rate": 2.0365198794629303e-05,
      "loss": 0.7055,
      "step": 5312
    },
    {
      "epoch": 0.7936958470271885,
      "grad_norm": 0.14469845592975616,
      "learning_rate": 2.03367791540243e-05,
      "loss": 0.6685,
      "step": 5313
    },
    {
      "epoch": 0.7938452345383926,
      "grad_norm": 0.20674848556518555,
      "learning_rate": 2.030837711228657e-05,
      "loss": 0.7583,
      "step": 5314
    },
    {
      "epoch": 0.7939946220495967,
      "grad_norm": 0.1621209979057312,
      "learning_rate": 2.0279992675690597e-05,
      "loss": 0.415,
      "step": 5315
    },
    {
      "epoch": 0.7941440095608007,
      "grad_norm": 0.22147023677825928,
      "learning_rate": 2.0251625850506927e-05,
      "loss": 0.683,
      "step": 5316
    },
    {
      "epoch": 0.7942933970720047,
      "grad_norm": 0.2192467898130417,
      "learning_rate": 2.0223276643002298e-05,
      "loss": 0.806,
      "step": 5317
    },
    {
      "epoch": 0.7944427845832088,
      "grad_norm": 0.2327844649553299,
      "learning_rate": 2.0194945059439417e-05,
      "loss": 0.9715,
      "step": 5318
    },
    {
      "epoch": 0.7945921720944129,
      "grad_norm": 0.1621626913547516,
      "learning_rate": 2.016663110607723e-05,
      "loss": 0.7152,
      "step": 5319
    },
    {
      "epoch": 0.794741559605617,
      "grad_norm": 0.15661302208900452,
      "learning_rate": 2.01383347891707e-05,
      "loss": 0.8897,
      "step": 5320
    },
    {
      "epoch": 0.7948909471168211,
      "grad_norm": 0.1622706949710846,
      "learning_rate": 2.0110056114970976e-05,
      "loss": 0.3876,
      "step": 5321
    },
    {
      "epoch": 0.7950403346280251,
      "grad_norm": 0.3006645143032074,
      "learning_rate": 2.0081795089725253e-05,
      "loss": 1.015,
      "step": 5322
    },
    {
      "epoch": 0.7951897221392291,
      "grad_norm": 0.4093431234359741,
      "learning_rate": 2.0053551719676834e-05,
      "loss": 0.9866,
      "step": 5323
    },
    {
      "epoch": 0.7953391096504332,
      "grad_norm": 0.13412240147590637,
      "learning_rate": 2.0025326011065148e-05,
      "loss": 0.4136,
      "step": 5324
    },
    {
      "epoch": 0.7954884971616373,
      "grad_norm": 0.3148360550403595,
      "learning_rate": 1.9997117970125657e-05,
      "loss": 0.8149,
      "step": 5325
    },
    {
      "epoch": 0.7956378846728414,
      "grad_norm": 0.2047261744737625,
      "learning_rate": 1.996892760308998e-05,
      "loss": 0.7814,
      "step": 5326
    },
    {
      "epoch": 0.7957872721840454,
      "grad_norm": 0.14328084886074066,
      "learning_rate": 1.9940754916185833e-05,
      "loss": 0.9543,
      "step": 5327
    },
    {
      "epoch": 0.7959366596952495,
      "grad_norm": 0.17345598340034485,
      "learning_rate": 1.9912599915637e-05,
      "loss": 0.6357,
      "step": 5328
    },
    {
      "epoch": 0.7960860472064536,
      "grad_norm": 0.12172969430685043,
      "learning_rate": 1.9884462607663356e-05,
      "loss": 0.5103,
      "step": 5329
    },
    {
      "epoch": 0.7962354347176576,
      "grad_norm": 0.11952207237482071,
      "learning_rate": 1.9856342998480913e-05,
      "loss": 0.6461,
      "step": 5330
    },
    {
      "epoch": 0.7963848222288616,
      "grad_norm": 0.12935063242912292,
      "learning_rate": 1.9828241094301682e-05,
      "loss": 0.8368,
      "step": 5331
    },
    {
      "epoch": 0.7965342097400657,
      "grad_norm": 0.1379585862159729,
      "learning_rate": 1.9800156901333855e-05,
      "loss": 0.538,
      "step": 5332
    },
    {
      "epoch": 0.7966835972512698,
      "grad_norm": 0.13809721171855927,
      "learning_rate": 1.9772090425781663e-05,
      "loss": 0.9884,
      "step": 5333
    },
    {
      "epoch": 0.7968329847624739,
      "grad_norm": 0.14485864341259003,
      "learning_rate": 1.9744041673845448e-05,
      "loss": 0.6174,
      "step": 5334
    },
    {
      "epoch": 0.796982372273678,
      "grad_norm": 0.15824495255947113,
      "learning_rate": 1.9716010651721617e-05,
      "loss": 0.8428,
      "step": 5335
    },
    {
      "epoch": 0.7971317597848819,
      "grad_norm": 0.17472563683986664,
      "learning_rate": 1.9687997365602663e-05,
      "loss": 0.777,
      "step": 5336
    },
    {
      "epoch": 0.797281147296086,
      "grad_norm": 0.13931693136692047,
      "learning_rate": 1.9660001821677156e-05,
      "loss": 0.5304,
      "step": 5337
    },
    {
      "epoch": 0.7974305348072901,
      "grad_norm": 0.12112227082252502,
      "learning_rate": 1.963202402612978e-05,
      "loss": 0.6447,
      "step": 5338
    },
    {
      "epoch": 0.7975799223184942,
      "grad_norm": 0.16521169245243073,
      "learning_rate": 1.9604063985141252e-05,
      "loss": 1.0061,
      "step": 5339
    },
    {
      "epoch": 0.7977293098296983,
      "grad_norm": 0.15004034340381622,
      "learning_rate": 1.9576121704888396e-05,
      "loss": 0.8712,
      "step": 5340
    },
    {
      "epoch": 0.7978786973409023,
      "grad_norm": 0.3711937665939331,
      "learning_rate": 1.9548197191544116e-05,
      "loss": 0.7065,
      "step": 5341
    },
    {
      "epoch": 0.7980280848521064,
      "grad_norm": 0.19882605969905853,
      "learning_rate": 1.9520290451277358e-05,
      "loss": 0.714,
      "step": 5342
    },
    {
      "epoch": 0.7981774723633104,
      "grad_norm": 0.11923779547214508,
      "learning_rate": 1.9492401490253186e-05,
      "loss": 0.7009,
      "step": 5343
    },
    {
      "epoch": 0.7983268598745145,
      "grad_norm": 0.16307072341442108,
      "learning_rate": 1.946453031463269e-05,
      "loss": 0.7133,
      "step": 5344
    },
    {
      "epoch": 0.7984762473857185,
      "grad_norm": 0.15806059539318085,
      "learning_rate": 1.943667693057306e-05,
      "loss": 0.8278,
      "step": 5345
    },
    {
      "epoch": 0.7986256348969226,
      "grad_norm": 0.3282451927661896,
      "learning_rate": 1.9408841344227547e-05,
      "loss": 0.7688,
      "step": 5346
    },
    {
      "epoch": 0.7987750224081267,
      "grad_norm": 0.13596604764461517,
      "learning_rate": 1.938102356174548e-05,
      "loss": 0.8522,
      "step": 5347
    },
    {
      "epoch": 0.7989244099193308,
      "grad_norm": 0.20291773974895477,
      "learning_rate": 1.9353223589272252e-05,
      "loss": 0.8937,
      "step": 5348
    },
    {
      "epoch": 0.7990737974305349,
      "grad_norm": 0.16487884521484375,
      "learning_rate": 1.9325441432949343e-05,
      "loss": 0.8992,
      "step": 5349
    },
    {
      "epoch": 0.7992231849417388,
      "grad_norm": 0.12905290722846985,
      "learning_rate": 1.92976770989142e-05,
      "loss": 0.8551,
      "step": 5350
    },
    {
      "epoch": 0.7993725724529429,
      "grad_norm": 0.19457148015499115,
      "learning_rate": 1.9269930593300456e-05,
      "loss": 0.5808,
      "step": 5351
    },
    {
      "epoch": 0.799521959964147,
      "grad_norm": 0.3203955292701721,
      "learning_rate": 1.9242201922237736e-05,
      "loss": 0.682,
      "step": 5352
    },
    {
      "epoch": 0.7996713474753511,
      "grad_norm": 0.3002904951572418,
      "learning_rate": 1.921449109185176e-05,
      "loss": 0.7558,
      "step": 5353
    },
    {
      "epoch": 0.7998207349865551,
      "grad_norm": 0.2331896722316742,
      "learning_rate": 1.918679810826427e-05,
      "loss": 0.9767,
      "step": 5354
    },
    {
      "epoch": 0.7999701224977592,
      "grad_norm": 0.17402228713035583,
      "learning_rate": 1.9159122977593092e-05,
      "loss": 0.692,
      "step": 5355
    },
    {
      "epoch": 0.8001195100089632,
      "grad_norm": 0.14816094934940338,
      "learning_rate": 1.91314657059521e-05,
      "loss": 0.9056,
      "step": 5356
    },
    {
      "epoch": 0.8002688975201673,
      "grad_norm": 0.18688000738620758,
      "learning_rate": 1.9103826299451222e-05,
      "loss": 0.6324,
      "step": 5357
    },
    {
      "epoch": 0.8004182850313714,
      "grad_norm": 0.2019314169883728,
      "learning_rate": 1.907620476419645e-05,
      "loss": 1.1324,
      "step": 5358
    },
    {
      "epoch": 0.8005676725425754,
      "grad_norm": 0.2464497834444046,
      "learning_rate": 1.904860110628981e-05,
      "loss": 1.2232,
      "step": 5359
    },
    {
      "epoch": 0.8007170600537795,
      "grad_norm": 0.17911460995674133,
      "learning_rate": 1.9021015331829396e-05,
      "loss": 0.7772,
      "step": 5360
    }
  ],
  "logging_steps": 1,
  "max_steps": 6694,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 670,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.2341211063504077e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}