{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9960681520314547,
  "eval_steps": 500,
  "global_step": 1524,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001965923984272608,
      "grad_norm": 5.710505962371826,
      "learning_rate": 6.535947712418302e-08,
      "loss": 0.8567,
      "step": 1
    },
    {
      "epoch": 0.003931847968545216,
      "grad_norm": 5.58311653137207,
      "learning_rate": 1.3071895424836603e-07,
      "loss": 0.8409,
      "step": 2
    },
    {
      "epoch": 0.005897771952817824,
      "grad_norm": 5.551272869110107,
      "learning_rate": 1.9607843137254904e-07,
      "loss": 0.8184,
      "step": 3
    },
    {
      "epoch": 0.007863695937090432,
      "grad_norm": 5.5394158363342285,
      "learning_rate": 2.6143790849673207e-07,
      "loss": 0.8196,
      "step": 4
    },
    {
      "epoch": 0.009829619921363041,
      "grad_norm": 5.637419700622559,
      "learning_rate": 3.267973856209151e-07,
      "loss": 0.8381,
      "step": 5
    },
    {
      "epoch": 0.011795543905635648,
      "grad_norm": 5.617730617523193,
      "learning_rate": 3.921568627450981e-07,
      "loss": 0.849,
      "step": 6
    },
    {
      "epoch": 0.013761467889908258,
      "grad_norm": 5.583267688751221,
      "learning_rate": 4.5751633986928105e-07,
      "loss": 0.8528,
      "step": 7
    },
    {
      "epoch": 0.015727391874180863,
      "grad_norm": 5.440237045288086,
      "learning_rate": 5.228758169934641e-07,
      "loss": 0.8135,
      "step": 8
    },
    {
      "epoch": 0.017693315858453473,
      "grad_norm": 5.594210147857666,
      "learning_rate": 5.882352941176471e-07,
      "loss": 0.841,
      "step": 9
    },
    {
      "epoch": 0.019659239842726082,
      "grad_norm": 5.199444770812988,
      "learning_rate": 6.535947712418302e-07,
      "loss": 0.8116,
      "step": 10
    },
    {
      "epoch": 0.02162516382699869,
      "grad_norm": 5.102973937988281,
      "learning_rate": 7.189542483660131e-07,
      "loss": 0.8378,
      "step": 11
    },
    {
      "epoch": 0.023591087811271297,
      "grad_norm": 5.083095550537109,
      "learning_rate": 7.843137254901962e-07,
      "loss": 0.8191,
      "step": 12
    },
    {
      "epoch": 0.025557011795543906,
      "grad_norm": 4.040502071380615,
      "learning_rate": 8.496732026143792e-07,
      "loss": 0.8056,
      "step": 13
    },
    {
      "epoch": 0.027522935779816515,
      "grad_norm": 3.962697982788086,
      "learning_rate": 9.150326797385621e-07,
      "loss": 0.7871,
      "step": 14
    },
    {
      "epoch": 0.02948885976408912,
      "grad_norm": 3.821044921875,
      "learning_rate": 9.80392156862745e-07,
      "loss": 0.8056,
      "step": 15
    },
    {
      "epoch": 0.03145478374836173,
      "grad_norm": 3.665053367614746,
      "learning_rate": 1.0457516339869283e-06,
      "loss": 0.7729,
      "step": 16
    },
    {
      "epoch": 0.033420707732634336,
      "grad_norm": 2.3340535163879395,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.7517,
      "step": 17
    },
    {
      "epoch": 0.035386631716906945,
      "grad_norm": 2.262815475463867,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 0.764,
      "step": 18
    },
    {
      "epoch": 0.037352555701179554,
      "grad_norm": 2.1935653686523438,
      "learning_rate": 1.2418300653594772e-06,
      "loss": 0.7601,
      "step": 19
    },
    {
      "epoch": 0.039318479685452164,
      "grad_norm": 2.0895442962646484,
      "learning_rate": 1.3071895424836604e-06,
      "loss": 0.7611,
      "step": 20
    },
    {
      "epoch": 0.04128440366972477,
      "grad_norm": 1.9166232347488403,
      "learning_rate": 1.3725490196078434e-06,
      "loss": 0.7507,
      "step": 21
    },
    {
      "epoch": 0.04325032765399738,
      "grad_norm": 1.8988736867904663,
      "learning_rate": 1.4379084967320261e-06,
      "loss": 0.7888,
      "step": 22
    },
    {
      "epoch": 0.045216251638269984,
      "grad_norm": 1.5350834131240845,
      "learning_rate": 1.5032679738562091e-06,
      "loss": 0.7509,
      "step": 23
    },
    {
      "epoch": 0.047182175622542594,
      "grad_norm": 2.2456037998199463,
      "learning_rate": 1.5686274509803923e-06,
      "loss": 0.7112,
      "step": 24
    },
    {
      "epoch": 0.0491480996068152,
      "grad_norm": 2.773117780685425,
      "learning_rate": 1.6339869281045753e-06,
      "loss": 0.7437,
      "step": 25
    },
    {
      "epoch": 0.05111402359108781,
      "grad_norm": 2.6303365230560303,
      "learning_rate": 1.6993464052287585e-06,
      "loss": 0.6998,
      "step": 26
    },
    {
      "epoch": 0.05307994757536042,
      "grad_norm": 2.7553372383117676,
      "learning_rate": 1.7647058823529414e-06,
      "loss": 0.7304,
      "step": 27
    },
    {
      "epoch": 0.05504587155963303,
      "grad_norm": 2.382776975631714,
      "learning_rate": 1.8300653594771242e-06,
      "loss": 0.7093,
      "step": 28
    },
    {
      "epoch": 0.05701179554390563,
      "grad_norm": 2.04976487159729,
      "learning_rate": 1.8954248366013072e-06,
      "loss": 0.6828,
      "step": 29
    },
    {
      "epoch": 0.05897771952817824,
      "grad_norm": 1.8414767980575562,
      "learning_rate": 1.96078431372549e-06,
      "loss": 0.7195,
      "step": 30
    },
    {
      "epoch": 0.06094364351245085,
      "grad_norm": 1.3379838466644287,
      "learning_rate": 2.0261437908496734e-06,
      "loss": 0.6714,
      "step": 31
    },
    {
      "epoch": 0.06290956749672345,
      "grad_norm": 1.0512365102767944,
      "learning_rate": 2.0915032679738565e-06,
      "loss": 0.6816,
      "step": 32
    },
    {
      "epoch": 0.06487549148099607,
      "grad_norm": 1.1513341665267944,
      "learning_rate": 2.1568627450980393e-06,
      "loss": 0.6715,
      "step": 33
    },
    {
      "epoch": 0.06684141546526867,
      "grad_norm": 1.170161485671997,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.6749,
      "step": 34
    },
    {
      "epoch": 0.06880733944954129,
      "grad_norm": 1.1509945392608643,
      "learning_rate": 2.2875816993464053e-06,
      "loss": 0.6724,
      "step": 35
    },
    {
      "epoch": 0.07077326343381389,
      "grad_norm": 1.0026973485946655,
      "learning_rate": 2.3529411764705885e-06,
      "loss": 0.6533,
      "step": 36
    },
    {
      "epoch": 0.0727391874180865,
      "grad_norm": 0.8049423098564148,
      "learning_rate": 2.4183006535947716e-06,
      "loss": 0.6687,
      "step": 37
    },
    {
      "epoch": 0.07470511140235911,
      "grad_norm": 0.8071291446685791,
      "learning_rate": 2.4836601307189544e-06,
      "loss": 0.6419,
      "step": 38
    },
    {
      "epoch": 0.07667103538663171,
      "grad_norm": 0.7793786525726318,
      "learning_rate": 2.549019607843137e-06,
      "loss": 0.6659,
      "step": 39
    },
    {
      "epoch": 0.07863695937090433,
      "grad_norm": 0.6068173050880432,
      "learning_rate": 2.6143790849673208e-06,
      "loss": 0.6398,
      "step": 40
    },
    {
      "epoch": 0.08060288335517693,
      "grad_norm": 0.589724600315094,
      "learning_rate": 2.6797385620915036e-06,
      "loss": 0.6064,
      "step": 41
    },
    {
      "epoch": 0.08256880733944955,
      "grad_norm": 0.6173176765441895,
      "learning_rate": 2.7450980392156867e-06,
      "loss": 0.648,
      "step": 42
    },
    {
      "epoch": 0.08453473132372215,
      "grad_norm": 0.6198132634162903,
      "learning_rate": 2.8104575163398695e-06,
      "loss": 0.6018,
      "step": 43
    },
    {
      "epoch": 0.08650065530799476,
      "grad_norm": 0.7025796175003052,
      "learning_rate": 2.8758169934640523e-06,
      "loss": 0.6285,
      "step": 44
    },
    {
      "epoch": 0.08846657929226737,
      "grad_norm": 0.49610280990600586,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 0.6006,
      "step": 45
    },
    {
      "epoch": 0.09043250327653997,
      "grad_norm": 0.45890048146247864,
      "learning_rate": 3.0065359477124182e-06,
      "loss": 0.6071,
      "step": 46
    },
    {
      "epoch": 0.09239842726081259,
      "grad_norm": 0.5086219310760498,
      "learning_rate": 3.071895424836602e-06,
      "loss": 0.5908,
      "step": 47
    },
    {
      "epoch": 0.09436435124508519,
      "grad_norm": 0.5073956847190857,
      "learning_rate": 3.1372549019607846e-06,
      "loss": 0.5975,
      "step": 48
    },
    {
      "epoch": 0.0963302752293578,
      "grad_norm": 0.49690189957618713,
      "learning_rate": 3.2026143790849674e-06,
      "loss": 0.5953,
      "step": 49
    },
    {
      "epoch": 0.0982961992136304,
      "grad_norm": 0.49608007073402405,
      "learning_rate": 3.2679738562091506e-06,
      "loss": 0.6189,
      "step": 50
    },
    {
      "epoch": 0.10026212319790301,
      "grad_norm": 0.42027363181114197,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.5825,
      "step": 51
    },
    {
      "epoch": 0.10222804718217562,
      "grad_norm": 0.3997390568256378,
      "learning_rate": 3.398692810457517e-06,
      "loss": 0.5903,
      "step": 52
    },
    {
      "epoch": 0.10419397116644823,
      "grad_norm": 0.3815591335296631,
      "learning_rate": 3.4640522875816997e-06,
      "loss": 0.6104,
      "step": 53
    },
    {
      "epoch": 0.10615989515072084,
      "grad_norm": 0.40735822916030884,
      "learning_rate": 3.529411764705883e-06,
      "loss": 0.5808,
      "step": 54
    },
    {
      "epoch": 0.10812581913499344,
      "grad_norm": 0.4196593165397644,
      "learning_rate": 3.5947712418300657e-06,
      "loss": 0.5827,
      "step": 55
    },
    {
      "epoch": 0.11009174311926606,
      "grad_norm": 0.40242332220077515,
      "learning_rate": 3.6601307189542484e-06,
      "loss": 0.5751,
      "step": 56
    },
    {
      "epoch": 0.11205766710353866,
      "grad_norm": 0.3715681731700897,
      "learning_rate": 3.7254901960784316e-06,
      "loss": 0.5979,
      "step": 57
    },
    {
      "epoch": 0.11402359108781127,
      "grad_norm": 0.34788185358047485,
      "learning_rate": 3.7908496732026144e-06,
      "loss": 0.5718,
      "step": 58
    },
    {
      "epoch": 0.11598951507208388,
      "grad_norm": 0.37109532952308655,
      "learning_rate": 3.856209150326798e-06,
      "loss": 0.5671,
      "step": 59
    },
    {
      "epoch": 0.11795543905635648,
      "grad_norm": 0.35452768206596375,
      "learning_rate": 3.92156862745098e-06,
      "loss": 0.5621,
      "step": 60
    },
    {
      "epoch": 0.1199213630406291,
      "grad_norm": 0.36002615094184875,
      "learning_rate": 3.986928104575164e-06,
      "loss": 0.5881,
      "step": 61
    },
    {
      "epoch": 0.1218872870249017,
      "grad_norm": 0.3341769874095917,
      "learning_rate": 4.052287581699347e-06,
      "loss": 0.5599,
      "step": 62
    },
    {
      "epoch": 0.12385321100917432,
      "grad_norm": 0.34028100967407227,
      "learning_rate": 4.11764705882353e-06,
      "loss": 0.5593,
      "step": 63
    },
    {
      "epoch": 0.1258191349934469,
      "grad_norm": 0.3311789631843567,
      "learning_rate": 4.183006535947713e-06,
      "loss": 0.5632,
      "step": 64
    },
    {
      "epoch": 0.12778505897771952,
      "grad_norm": 0.34246182441711426,
      "learning_rate": 4.2483660130718954e-06,
      "loss": 0.6082,
      "step": 65
    },
    {
      "epoch": 0.12975098296199214,
      "grad_norm": 0.295360803604126,
      "learning_rate": 4.313725490196079e-06,
      "loss": 0.5482,
      "step": 66
    },
    {
      "epoch": 0.13171690694626476,
      "grad_norm": 0.3409435451030731,
      "learning_rate": 4.379084967320262e-06,
      "loss": 0.5675,
      "step": 67
    },
    {
      "epoch": 0.13368283093053734,
      "grad_norm": 0.34775587916374207,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.5521,
      "step": 68
    },
    {
      "epoch": 0.13564875491480996,
      "grad_norm": 0.3222702741622925,
      "learning_rate": 4.509803921568628e-06,
      "loss": 0.5765,
      "step": 69
    },
    {
      "epoch": 0.13761467889908258,
      "grad_norm": 0.2791072130203247,
      "learning_rate": 4.5751633986928105e-06,
      "loss": 0.5511,
      "step": 70
    },
    {
      "epoch": 0.13958060288335516,
      "grad_norm": 0.3028191030025482,
      "learning_rate": 4.640522875816994e-06,
      "loss": 0.5749,
      "step": 71
    },
    {
      "epoch": 0.14154652686762778,
      "grad_norm": 0.28149256110191345,
      "learning_rate": 4.705882352941177e-06,
      "loss": 0.5296,
      "step": 72
    },
    {
      "epoch": 0.1435124508519004,
      "grad_norm": 0.3133794665336609,
      "learning_rate": 4.77124183006536e-06,
      "loss": 0.6033,
      "step": 73
    },
    {
      "epoch": 0.145478374836173,
      "grad_norm": 0.2812778949737549,
      "learning_rate": 4.836601307189543e-06,
      "loss": 0.5632,
      "step": 74
    },
    {
      "epoch": 0.1474442988204456,
      "grad_norm": 0.30257362127304077,
      "learning_rate": 4.901960784313726e-06,
      "loss": 0.5605,
      "step": 75
    },
    {
      "epoch": 0.14941022280471822,
      "grad_norm": 0.25866395235061646,
      "learning_rate": 4.967320261437909e-06,
      "loss": 0.5735,
      "step": 76
    },
    {
      "epoch": 0.15137614678899083,
      "grad_norm": 0.28107836842536926,
      "learning_rate": 5.032679738562092e-06,
      "loss": 0.5742,
      "step": 77
    },
    {
      "epoch": 0.15334207077326342,
      "grad_norm": 0.29915159940719604,
      "learning_rate": 5.098039215686274e-06,
      "loss": 0.5426,
      "step": 78
    },
    {
      "epoch": 0.15530799475753604,
      "grad_norm": 0.2864636778831482,
      "learning_rate": 5.163398692810458e-06,
      "loss": 0.5499,
      "step": 79
    },
    {
      "epoch": 0.15727391874180865,
      "grad_norm": 0.31190046668052673,
      "learning_rate": 5.2287581699346416e-06,
      "loss": 0.547,
      "step": 80
    },
    {
      "epoch": 0.15923984272608127,
      "grad_norm": 0.30313733220100403,
      "learning_rate": 5.294117647058824e-06,
      "loss": 0.5517,
      "step": 81
    },
    {
      "epoch": 0.16120576671035386,
      "grad_norm": 0.2760246694087982,
      "learning_rate": 5.359477124183007e-06,
      "loss": 0.5477,
      "step": 82
    },
    {
      "epoch": 0.16317169069462648,
      "grad_norm": 0.3234259784221649,
      "learning_rate": 5.4248366013071894e-06,
      "loss": 0.5667,
      "step": 83
    },
    {
      "epoch": 0.1651376146788991,
      "grad_norm": 0.32582947611808777,
      "learning_rate": 5.4901960784313735e-06,
      "loss": 0.5612,
      "step": 84
    },
    {
      "epoch": 0.16710353866317168,
      "grad_norm": 0.29849186539649963,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.5855,
      "step": 85
    },
    {
      "epoch": 0.1690694626474443,
      "grad_norm": 0.3021292984485626,
      "learning_rate": 5.620915032679739e-06,
      "loss": 0.5514,
      "step": 86
    },
    {
      "epoch": 0.1710353866317169,
      "grad_norm": 0.30864256620407104,
      "learning_rate": 5.686274509803922e-06,
      "loss": 0.5377,
      "step": 87
    },
    {
      "epoch": 0.17300131061598953,
      "grad_norm": 0.28641265630722046,
      "learning_rate": 5.7516339869281045e-06,
      "loss": 0.5208,
      "step": 88
    },
    {
      "epoch": 0.17496723460026212,
      "grad_norm": 0.307472825050354,
      "learning_rate": 5.816993464052289e-06,
      "loss": 0.5354,
      "step": 89
    },
    {
      "epoch": 0.17693315858453473,
      "grad_norm": 0.3084348440170288,
      "learning_rate": 5.882352941176471e-06,
      "loss": 0.5328,
      "step": 90
    },
    {
      "epoch": 0.17889908256880735,
      "grad_norm": 0.3037461042404175,
      "learning_rate": 5.947712418300654e-06,
      "loss": 0.5287,
      "step": 91
    },
    {
      "epoch": 0.18086500655307994,
      "grad_norm": 0.28996726870536804,
      "learning_rate": 6.0130718954248365e-06,
      "loss": 0.5582,
      "step": 92
    },
    {
      "epoch": 0.18283093053735255,
      "grad_norm": 0.3222200572490692,
      "learning_rate": 6.07843137254902e-06,
      "loss": 0.5666,
      "step": 93
    },
    {
      "epoch": 0.18479685452162517,
      "grad_norm": 0.2793603837490082,
      "learning_rate": 6.143790849673204e-06,
      "loss": 0.5416,
      "step": 94
    },
    {
      "epoch": 0.18676277850589776,
      "grad_norm": 0.2899613380432129,
      "learning_rate": 6.209150326797386e-06,
      "loss": 0.5213,
      "step": 95
    },
    {
      "epoch": 0.18872870249017037,
      "grad_norm": 0.3074812591075897,
      "learning_rate": 6.274509803921569e-06,
      "loss": 0.5353,
      "step": 96
    },
    {
      "epoch": 0.190694626474443,
      "grad_norm": 0.3134816586971283,
      "learning_rate": 6.3398692810457515e-06,
      "loss": 0.5232,
      "step": 97
    },
    {
      "epoch": 0.1926605504587156,
      "grad_norm": 0.3031688630580902,
      "learning_rate": 6.405228758169935e-06,
      "loss": 0.5484,
      "step": 98
    },
    {
      "epoch": 0.1946264744429882,
      "grad_norm": 0.2818717956542969,
      "learning_rate": 6.470588235294119e-06,
      "loss": 0.5255,
      "step": 99
    },
    {
      "epoch": 0.1965923984272608,
      "grad_norm": 0.2932426333427429,
      "learning_rate": 6.535947712418301e-06,
      "loss": 0.5307,
      "step": 100
    },
    {
      "epoch": 0.19855832241153343,
      "grad_norm": 0.2900339663028717,
      "learning_rate": 6.601307189542484e-06,
      "loss": 0.5319,
      "step": 101
    },
    {
      "epoch": 0.20052424639580602,
      "grad_norm": 0.2933003902435303,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.5313,
      "step": 102
    },
    {
      "epoch": 0.20249017038007863,
      "grad_norm": 0.3101281523704529,
      "learning_rate": 6.732026143790851e-06,
      "loss": 0.5337,
      "step": 103
    },
    {
      "epoch": 0.20445609436435125,
      "grad_norm": 0.2710820138454437,
      "learning_rate": 6.797385620915034e-06,
      "loss": 0.5459,
      "step": 104
    },
    {
      "epoch": 0.20642201834862386,
      "grad_norm": 0.31979817152023315,
      "learning_rate": 6.862745098039216e-06,
      "loss": 0.5247,
      "step": 105
    },
    {
      "epoch": 0.20838794233289645,
      "grad_norm": 0.2835678160190582,
      "learning_rate": 6.928104575163399e-06,
      "loss": 0.5182,
      "step": 106
    },
    {
      "epoch": 0.21035386631716907,
      "grad_norm": 0.4005821645259857,
      "learning_rate": 6.993464052287582e-06,
      "loss": 0.5273,
      "step": 107
    },
    {
      "epoch": 0.21231979030144169,
      "grad_norm": 0.30453720688819885,
      "learning_rate": 7.058823529411766e-06,
      "loss": 0.5205,
      "step": 108
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 0.3164799213409424,
      "learning_rate": 7.124183006535948e-06,
      "loss": 0.5297,
      "step": 109
    },
    {
      "epoch": 0.2162516382699869,
      "grad_norm": 0.3417559862136841,
      "learning_rate": 7.189542483660131e-06,
      "loss": 0.5178,
      "step": 110
    },
    {
      "epoch": 0.2182175622542595,
      "grad_norm": 0.28603699803352356,
      "learning_rate": 7.2549019607843145e-06,
      "loss": 0.5394,
      "step": 111
    },
    {
      "epoch": 0.22018348623853212,
      "grad_norm": 0.41378653049468994,
      "learning_rate": 7.320261437908497e-06,
      "loss": 0.5381,
      "step": 112
    },
    {
      "epoch": 0.2221494102228047,
      "grad_norm": 0.31651490926742554,
      "learning_rate": 7.385620915032681e-06,
      "loss": 0.5334,
      "step": 113
    },
    {
      "epoch": 0.22411533420707733,
      "grad_norm": 0.44894519448280334,
      "learning_rate": 7.450980392156863e-06,
      "loss": 0.536,
      "step": 114
    },
    {
      "epoch": 0.22608125819134994,
      "grad_norm": 0.3745504319667816,
      "learning_rate": 7.516339869281046e-06,
      "loss": 0.5289,
      "step": 115
    },
    {
      "epoch": 0.22804718217562253,
      "grad_norm": 0.3731832206249237,
      "learning_rate": 7.581699346405229e-06,
      "loss": 0.5122,
      "step": 116
    },
    {
      "epoch": 0.23001310615989515,
      "grad_norm": 0.37562766671180725,
      "learning_rate": 7.647058823529411e-06,
      "loss": 0.5206,
      "step": 117
    },
    {
      "epoch": 0.23197903014416776,
      "grad_norm": 0.32179543375968933,
      "learning_rate": 7.712418300653596e-06,
      "loss": 0.5253,
      "step": 118
    },
    {
      "epoch": 0.23394495412844038,
      "grad_norm": 0.46252328157424927,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.5188,
      "step": 119
    },
    {
      "epoch": 0.23591087811271297,
      "grad_norm": 0.29633471369743347,
      "learning_rate": 7.84313725490196e-06,
      "loss": 0.5129,
      "step": 120
    },
    {
      "epoch": 0.23787680209698558,
      "grad_norm": 0.39629998803138733,
      "learning_rate": 7.908496732026144e-06,
      "loss": 0.5448,
      "step": 121
    },
    {
      "epoch": 0.2398427260812582,
      "grad_norm": 0.30213412642478943,
      "learning_rate": 7.973856209150329e-06,
      "loss": 0.5174,
      "step": 122
    },
    {
      "epoch": 0.2418086500655308,
      "grad_norm": 0.46073678135871887,
      "learning_rate": 8.03921568627451e-06,
      "loss": 0.5264,
      "step": 123
    },
    {
      "epoch": 0.2437745740498034,
      "grad_norm": 0.3474709093570709,
      "learning_rate": 8.104575163398693e-06,
      "loss": 0.5149,
      "step": 124
    },
    {
      "epoch": 0.24574049803407602,
      "grad_norm": 0.37764114141464233,
      "learning_rate": 8.169934640522877e-06,
      "loss": 0.5174,
      "step": 125
    },
    {
      "epoch": 0.24770642201834864,
      "grad_norm": 0.3189232647418976,
      "learning_rate": 8.23529411764706e-06,
      "loss": 0.5257,
      "step": 126
    },
    {
      "epoch": 0.24967234600262123,
      "grad_norm": 0.3779374063014984,
      "learning_rate": 8.300653594771243e-06,
      "loss": 0.5307,
      "step": 127
    },
    {
      "epoch": 0.2516382699868938,
      "grad_norm": 0.2898067235946655,
      "learning_rate": 8.366013071895426e-06,
      "loss": 0.5098,
      "step": 128
    },
    {
      "epoch": 0.25360419397116646,
      "grad_norm": 0.3959129750728607,
      "learning_rate": 8.43137254901961e-06,
      "loss": 0.529,
      "step": 129
    },
    {
      "epoch": 0.25557011795543905,
      "grad_norm": 0.33046963810920715,
      "learning_rate": 8.496732026143791e-06,
      "loss": 0.5165,
      "step": 130
    },
    {
      "epoch": 0.2575360419397117,
      "grad_norm": 0.3353389799594879,
      "learning_rate": 8.562091503267974e-06,
      "loss": 0.5178,
      "step": 131
    },
    {
      "epoch": 0.2595019659239843,
      "grad_norm": 0.36417704820632935,
      "learning_rate": 8.627450980392157e-06,
      "loss": 0.5153,
      "step": 132
    },
    {
      "epoch": 0.26146788990825687,
      "grad_norm": 0.3149842321872711,
      "learning_rate": 8.69281045751634e-06,
      "loss": 0.5339,
      "step": 133
    },
    {
      "epoch": 0.2634338138925295,
      "grad_norm": 0.40793469548225403,
      "learning_rate": 8.758169934640524e-06,
      "loss": 0.5169,
      "step": 134
    },
    {
      "epoch": 0.2653997378768021,
      "grad_norm": 0.3299751877784729,
      "learning_rate": 8.823529411764707e-06,
      "loss": 0.5399,
      "step": 135
    },
    {
      "epoch": 0.2673656618610747,
      "grad_norm": 0.34630492329597473,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.525,
      "step": 136
    },
    {
      "epoch": 0.26933158584534733,
      "grad_norm": 0.3971831500530243,
      "learning_rate": 8.954248366013073e-06,
      "loss": 0.5317,
      "step": 137
    },
    {
      "epoch": 0.2712975098296199,
      "grad_norm": 0.3247168958187103,
      "learning_rate": 9.019607843137256e-06,
      "loss": 0.4959,
      "step": 138
    },
    {
      "epoch": 0.2732634338138925,
      "grad_norm": 0.3679560422897339,
      "learning_rate": 9.084967320261438e-06,
      "loss": 0.5231,
      "step": 139
    },
    {
      "epoch": 0.27522935779816515,
      "grad_norm": 0.37172260880470276,
      "learning_rate": 9.150326797385621e-06,
      "loss": 0.5165,
      "step": 140
    },
    {
      "epoch": 0.27719528178243774,
      "grad_norm": 0.33269190788269043,
      "learning_rate": 9.215686274509804e-06,
      "loss": 0.5179,
      "step": 141
    },
    {
      "epoch": 0.27916120576671033,
      "grad_norm": 0.35121285915374756,
      "learning_rate": 9.281045751633987e-06,
      "loss": 0.5173,
      "step": 142
    },
    {
      "epoch": 0.281127129750983,
      "grad_norm": 0.32511773705482483,
      "learning_rate": 9.34640522875817e-06,
      "loss": 0.5253,
      "step": 143
    },
    {
      "epoch": 0.28309305373525556,
      "grad_norm": 0.34771692752838135,
      "learning_rate": 9.411764705882354e-06,
      "loss": 0.5297,
      "step": 144
    },
    {
      "epoch": 0.2850589777195282,
      "grad_norm": 0.280225932598114,
      "learning_rate": 9.477124183006537e-06,
      "loss": 0.5309,
      "step": 145
    },
    {
      "epoch": 0.2870249017038008,
      "grad_norm": 0.3082393407821655,
      "learning_rate": 9.54248366013072e-06,
      "loss": 0.5029,
      "step": 146
    },
    {
      "epoch": 0.2889908256880734,
      "grad_norm": 0.3622486889362335,
      "learning_rate": 9.607843137254903e-06,
      "loss": 0.5293,
      "step": 147
    },
    {
      "epoch": 0.290956749672346,
      "grad_norm": 0.3383775055408478,
      "learning_rate": 9.673202614379087e-06,
      "loss": 0.5041,
      "step": 148
    },
    {
      "epoch": 0.2929226736566186,
      "grad_norm": 0.427328497171402,
      "learning_rate": 9.738562091503268e-06,
      "loss": 0.5209,
      "step": 149
    },
    {
      "epoch": 0.2948885976408912,
      "grad_norm": 0.29876646399497986,
      "learning_rate": 9.803921568627451e-06,
      "loss": 0.505,
      "step": 150
    },
    {
      "epoch": 0.29685452162516385,
      "grad_norm": 0.4383123815059662,
      "learning_rate": 9.869281045751634e-06,
      "loss": 0.5321,
      "step": 151
    },
    {
      "epoch": 0.29882044560943644,
      "grad_norm": 0.40653786063194275,
      "learning_rate": 9.934640522875818e-06,
      "loss": 0.5166,
      "step": 152
    },
    {
      "epoch": 0.300786369593709,
      "grad_norm": 0.35615018010139465,
      "learning_rate": 1e-05,
      "loss": 0.5018,
      "step": 153
    },
    {
      "epoch": 0.30275229357798167,
      "grad_norm": 0.3174058198928833,
      "learning_rate": 9.999986873025113e-06,
      "loss": 0.4764,
      "step": 154
    },
    {
      "epoch": 0.30471821756225426,
      "grad_norm": 0.363629549741745,
      "learning_rate": 9.999947492169374e-06,
      "loss": 0.5066,
      "step": 155
    },
    {
      "epoch": 0.30668414154652685,
      "grad_norm": 0.3160753846168518,
      "learning_rate": 9.999881857639567e-06,
      "loss": 0.514,
      "step": 156
    },
    {
      "epoch": 0.3086500655307995,
      "grad_norm": 0.3479490876197815,
      "learning_rate": 9.999789969780322e-06,
      "loss": 0.5266,
      "step": 157
    },
    {
      "epoch": 0.3106159895150721,
      "grad_norm": 0.37828707695007324,
      "learning_rate": 9.999671829074127e-06,
      "loss": 0.5034,
      "step": 158
    },
    {
      "epoch": 0.31258191349934467,
      "grad_norm": 0.35347092151641846,
      "learning_rate": 9.999527436141312e-06,
      "loss": 0.5032,
      "step": 159
    },
    {
      "epoch": 0.3145478374836173,
      "grad_norm": 0.3980875611305237,
      "learning_rate": 9.999356791740052e-06,
      "loss": 0.5113,
      "step": 160
    },
    {
      "epoch": 0.3165137614678899,
      "grad_norm": 0.3677966892719269,
      "learning_rate": 9.999159896766368e-06,
      "loss": 0.5131,
      "step": 161
    },
    {
      "epoch": 0.31847968545216254,
      "grad_norm": 0.33659663796424866,
      "learning_rate": 9.998936752254111e-06,
      "loss": 0.4874,
      "step": 162
    },
    {
      "epoch": 0.32044560943643513,
      "grad_norm": 0.3887609541416168,
      "learning_rate": 9.99868735937497e-06,
      "loss": 0.5195,
      "step": 163
    },
    {
      "epoch": 0.3224115334207077,
      "grad_norm": 0.4096239507198334,
      "learning_rate": 9.998411719438451e-06,
      "loss": 0.5047,
      "step": 164
    },
    {
      "epoch": 0.32437745740498036,
      "grad_norm": 0.39740556478500366,
      "learning_rate": 9.998109833891883e-06,
      "loss": 0.5211,
      "step": 165
    },
    {
      "epoch": 0.32634338138925295,
      "grad_norm": 0.3286636173725128,
      "learning_rate": 9.997781704320404e-06,
      "loss": 0.5267,
      "step": 166
    },
    {
      "epoch": 0.32830930537352554,
      "grad_norm": 0.37281227111816406,
      "learning_rate": 9.997427332446953e-06,
      "loss": 0.5271,
      "step": 167
    },
    {
      "epoch": 0.3302752293577982,
      "grad_norm": 0.40283092856407166,
      "learning_rate": 9.997046720132262e-06,
      "loss": 0.5109,
      "step": 168
    },
    {
      "epoch": 0.33224115334207077,
      "grad_norm": 0.3566778302192688,
      "learning_rate": 9.996639869374844e-06,
      "loss": 0.5119,
      "step": 169
    },
    {
      "epoch": 0.33420707732634336,
      "grad_norm": 0.4530947804450989,
      "learning_rate": 9.996206782310993e-06,
      "loss": 0.4852,
      "step": 170
    },
    {
      "epoch": 0.336173001310616,
      "grad_norm": 0.5231772065162659,
      "learning_rate": 9.995747461214752e-06,
      "loss": 0.5207,
      "step": 171
    },
    {
      "epoch": 0.3381389252948886,
      "grad_norm": 0.368431955575943,
      "learning_rate": 9.995261908497924e-06,
      "loss": 0.4918,
      "step": 172
    },
    {
      "epoch": 0.3401048492791612,
      "grad_norm": 0.5586115717887878,
      "learning_rate": 9.994750126710042e-06,
      "loss": 0.5079,
      "step": 173
    },
    {
      "epoch": 0.3420707732634338,
      "grad_norm": 0.39342769980430603,
      "learning_rate": 9.994212118538364e-06,
      "loss": 0.5113,
      "step": 174
    },
    {
      "epoch": 0.3440366972477064,
      "grad_norm": 0.5048351287841797,
      "learning_rate": 9.99364788680786e-06,
      "loss": 0.5225,
      "step": 175
    },
    {
      "epoch": 0.34600262123197906,
      "grad_norm": 0.34715694189071655,
      "learning_rate": 9.99305743448119e-06,
      "loss": 0.5115,
      "step": 176
    },
    {
      "epoch": 0.34796854521625165,
      "grad_norm": 0.48474031686782837,
      "learning_rate": 9.992440764658697e-06,
      "loss": 0.5279,
      "step": 177
    },
    {
      "epoch": 0.34993446920052423,
      "grad_norm": 0.37137722969055176,
      "learning_rate": 9.991797880578386e-06,
      "loss": 0.5229,
      "step": 178
    },
    {
      "epoch": 0.3519003931847969,
      "grad_norm": 0.37795570492744446,
      "learning_rate": 9.991128785615903e-06,
      "loss": 0.5286,
      "step": 179
    },
    {
      "epoch": 0.35386631716906947,
      "grad_norm": 0.3694796562194824,
      "learning_rate": 9.990433483284527e-06,
      "loss": 0.496,
      "step": 180
    },
    {
      "epoch": 0.35583224115334205,
      "grad_norm": 0.326913058757782,
      "learning_rate": 9.989711977235144e-06,
      "loss": 0.5149,
      "step": 181
    },
    {
      "epoch": 0.3577981651376147,
      "grad_norm": 0.3359353840351105,
      "learning_rate": 9.98896427125623e-06,
      "loss": 0.5189,
      "step": 182
    },
    {
      "epoch": 0.3597640891218873,
      "grad_norm": 0.3222823143005371,
      "learning_rate": 9.988190369273834e-06,
      "loss": 0.5131,
      "step": 183
    },
    {
      "epoch": 0.3617300131061599,
      "grad_norm": 0.36565500497817993,
      "learning_rate": 9.987390275351549e-06,
      "loss": 0.4966,
      "step": 184
    },
    {
      "epoch": 0.3636959370904325,
      "grad_norm": 0.37140581011772156,
      "learning_rate": 9.986563993690506e-06,
      "loss": 0.5059,
      "step": 185
    },
    {
      "epoch": 0.3656618610747051,
      "grad_norm": 0.39855438470840454,
      "learning_rate": 9.985711528629332e-06,
      "loss": 0.5185,
      "step": 186
    },
    {
      "epoch": 0.3676277850589777,
      "grad_norm": 0.4295189678668976,
      "learning_rate": 9.984832884644141e-06,
      "loss": 0.4974,
      "step": 187
    },
    {
      "epoch": 0.36959370904325034,
      "grad_norm": 0.32439813017845154,
      "learning_rate": 9.983928066348514e-06,
      "loss": 0.5009,
      "step": 188
    },
    {
      "epoch": 0.37155963302752293,
      "grad_norm": 0.4018779695034027,
      "learning_rate": 9.982997078493457e-06,
      "loss": 0.4995,
      "step": 189
    },
    {
      "epoch": 0.3735255570117955,
      "grad_norm": 0.31894901394844055,
      "learning_rate": 9.98203992596739e-06,
      "loss": 0.4826,
      "step": 190
    },
    {
      "epoch": 0.37549148099606816,
      "grad_norm": 0.3858886957168579,
      "learning_rate": 9.981056613796125e-06,
      "loss": 0.4977,
      "step": 191
    },
    {
      "epoch": 0.37745740498034075,
      "grad_norm": 0.42188769578933716,
      "learning_rate": 9.980047147142824e-06,
      "loss": 0.5067,
      "step": 192
    },
    {
      "epoch": 0.3794233289646134,
      "grad_norm": 0.3689168095588684,
      "learning_rate": 9.979011531307986e-06,
      "loss": 0.4952,
      "step": 193
    },
    {
      "epoch": 0.381389252948886,
      "grad_norm": 0.40118807554244995,
      "learning_rate": 9.977949771729411e-06,
      "loss": 0.4987,
      "step": 194
    },
    {
      "epoch": 0.38335517693315857,
      "grad_norm": 0.33744242787361145,
      "learning_rate": 9.976861873982177e-06,
      "loss": 0.5017,
      "step": 195
    },
    {
      "epoch": 0.3853211009174312,
      "grad_norm": 0.3832445740699768,
      "learning_rate": 9.975747843778606e-06,
      "loss": 0.4903,
      "step": 196
    },
    {
      "epoch": 0.3872870249017038,
      "grad_norm": 0.3638055622577667,
      "learning_rate": 9.974607686968239e-06,
      "loss": 0.499,
      "step": 197
    },
    {
      "epoch": 0.3892529488859764,
      "grad_norm": 0.3829182982444763,
      "learning_rate": 9.973441409537795e-06,
      "loss": 0.5031,
      "step": 198
    },
    {
      "epoch": 0.39121887287024903,
      "grad_norm": 0.3455500304698944,
      "learning_rate": 9.972249017611153e-06,
      "loss": 0.5183,
      "step": 199
    },
    {
      "epoch": 0.3931847968545216,
      "grad_norm": 0.36498287320137024,
      "learning_rate": 9.971030517449314e-06,
      "loss": 0.5076,
      "step": 200
    },
    {
      "epoch": 0.3951507208387942,
      "grad_norm": 0.3443746864795685,
      "learning_rate": 9.969785915450368e-06,
      "loss": 0.4911,
      "step": 201
    },
    {
      "epoch": 0.39711664482306686,
      "grad_norm": 0.3428663909435272,
      "learning_rate": 9.968515218149455e-06,
      "loss": 0.5154,
      "step": 202
    },
    {
      "epoch": 0.39908256880733944,
      "grad_norm": 0.34340935945510864,
      "learning_rate": 9.967218432218742e-06,
      "loss": 0.4961,
      "step": 203
    },
    {
      "epoch": 0.40104849279161203,
      "grad_norm": 0.3325728178024292,
      "learning_rate": 9.965895564467381e-06,
      "loss": 0.5166,
      "step": 204
    },
    {
      "epoch": 0.4030144167758847,
      "grad_norm": 0.29527515172958374,
      "learning_rate": 9.96454662184147e-06,
      "loss": 0.4952,
      "step": 205
    },
    {
      "epoch": 0.40498034076015726,
      "grad_norm": 0.3524538576602936,
      "learning_rate": 9.963171611424022e-06,
      "loss": 0.4947,
      "step": 206
    },
    {
      "epoch": 0.4069462647444299,
      "grad_norm": 0.32383114099502563,
      "learning_rate": 9.961770540434931e-06,
      "loss": 0.472,
      "step": 207
    },
    {
      "epoch": 0.4089121887287025,
      "grad_norm": 0.3288062810897827,
      "learning_rate": 9.960343416230926e-06,
      "loss": 0.5072,
      "step": 208
    },
    {
      "epoch": 0.4108781127129751,
      "grad_norm": 0.34645599126815796,
      "learning_rate": 9.958890246305534e-06,
      "loss": 0.5052,
      "step": 209
    },
    {
      "epoch": 0.41284403669724773,
      "grad_norm": 0.3500404953956604,
      "learning_rate": 9.95741103828905e-06,
      "loss": 0.4891,
      "step": 210
    },
    {
      "epoch": 0.4148099606815203,
      "grad_norm": 0.3329164683818817,
      "learning_rate": 9.95590579994848e-06,
      "loss": 0.4768,
      "step": 211
    },
    {
      "epoch": 0.4167758846657929,
      "grad_norm": 0.34561702609062195,
      "learning_rate": 9.954374539187516e-06,
      "loss": 0.5009,
      "step": 212
    },
    {
      "epoch": 0.41874180865006555,
      "grad_norm": 0.30753183364868164,
      "learning_rate": 9.952817264046486e-06,
      "loss": 0.5008,
      "step": 213
    },
    {
      "epoch": 0.42070773263433814,
      "grad_norm": 0.30183136463165283,
      "learning_rate": 9.951233982702315e-06,
      "loss": 0.4982,
      "step": 214
    },
    {
      "epoch": 0.4226736566186107,
      "grad_norm": 0.3075961768627167,
      "learning_rate": 9.949624703468482e-06,
      "loss": 0.4795,
      "step": 215
    },
    {
      "epoch": 0.42463958060288337,
      "grad_norm": 0.32029989361763,
      "learning_rate": 9.947989434794973e-06,
      "loss": 0.5324,
      "step": 216
    },
    {
      "epoch": 0.42660550458715596,
      "grad_norm": 0.32862916588783264,
      "learning_rate": 9.946328185268242e-06,
      "loss": 0.4994,
      "step": 217
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.3346083462238312,
      "learning_rate": 9.944640963611156e-06,
      "loss": 0.4903,
      "step": 218
    },
    {
      "epoch": 0.4305373525557012,
      "grad_norm": 0.33467158675193787,
      "learning_rate": 9.942927778682968e-06,
      "loss": 0.4751,
      "step": 219
    },
    {
      "epoch": 0.4325032765399738,
      "grad_norm": 0.2937348186969757,
      "learning_rate": 9.94118863947925e-06,
      "loss": 0.4885,
      "step": 220
    },
    {
      "epoch": 0.43446920052424637,
      "grad_norm": 0.3598921000957489,
      "learning_rate": 9.939423555131856e-06,
      "loss": 0.5071,
      "step": 221
    },
    {
      "epoch": 0.436435124508519,
      "grad_norm": 0.2920673191547394,
      "learning_rate": 9.937632534908872e-06,
      "loss": 0.485,
      "step": 222
    },
    {
      "epoch": 0.4384010484927916,
      "grad_norm": 0.32336753606796265,
      "learning_rate": 9.935815588214573e-06,
      "loss": 0.4877,
      "step": 223
    },
    {
      "epoch": 0.44036697247706424,
      "grad_norm": 0.30055752396583557,
      "learning_rate": 9.93397272458936e-06,
      "loss": 0.4767,
      "step": 224
    },
    {
      "epoch": 0.44233289646133683,
      "grad_norm": 0.30270230770111084,
      "learning_rate": 9.932103953709724e-06,
      "loss": 0.5157,
      "step": 225
    },
    {
      "epoch": 0.4442988204456094,
      "grad_norm": 0.30203357338905334,
      "learning_rate": 9.93020928538819e-06,
      "loss": 0.4799,
      "step": 226
    },
    {
      "epoch": 0.44626474442988207,
      "grad_norm": 0.32589325308799744,
      "learning_rate": 9.928288729573264e-06,
      "loss": 0.4841,
      "step": 227
    },
    {
      "epoch": 0.44823066841415465,
      "grad_norm": 0.2911665141582489,
      "learning_rate": 9.926342296349378e-06,
      "loss": 0.4773,
      "step": 228
    },
    {
      "epoch": 0.45019659239842724,
      "grad_norm": 0.3272002041339874,
      "learning_rate": 9.924369995936846e-06,
      "loss": 0.4848,
      "step": 229
    },
    {
      "epoch": 0.4521625163826999,
      "grad_norm": 0.2837628722190857,
      "learning_rate": 9.922371838691803e-06,
      "loss": 0.4823,
      "step": 230
    },
    {
      "epoch": 0.4541284403669725,
      "grad_norm": 0.306533545255661,
      "learning_rate": 9.920347835106152e-06,
      "loss": 0.4891,
      "step": 231
    },
    {
      "epoch": 0.45609436435124506,
      "grad_norm": 0.3979865610599518,
      "learning_rate": 9.918297995807511e-06,
      "loss": 0.4827,
      "step": 232
    },
    {
      "epoch": 0.4580602883355177,
      "grad_norm": 0.28756773471832275,
      "learning_rate": 9.916222331559158e-06,
      "loss": 0.5067,
      "step": 233
    },
    {
      "epoch": 0.4600262123197903,
      "grad_norm": 0.44354233145713806,
      "learning_rate": 9.914120853259968e-06,
      "loss": 0.4736,
      "step": 234
    },
    {
      "epoch": 0.4619921363040629,
      "grad_norm": 0.3389144241809845,
      "learning_rate": 9.911993571944361e-06,
      "loss": 0.4674,
      "step": 235
    },
    {
      "epoch": 0.4639580602883355,
      "grad_norm": 0.40141353011131287,
      "learning_rate": 9.909840498782246e-06,
      "loss": 0.4988,
      "step": 236
    },
    {
      "epoch": 0.4659239842726081,
      "grad_norm": 0.3574008345603943,
      "learning_rate": 9.90766164507896e-06,
      "loss": 0.4823,
      "step": 237
    },
    {
      "epoch": 0.46788990825688076,
      "grad_norm": 0.32266223430633545,
      "learning_rate": 9.9054570222752e-06,
      "loss": 0.4858,
      "step": 238
    },
    {
      "epoch": 0.46985583224115335,
      "grad_norm": 0.34144648909568787,
      "learning_rate": 9.903226641946982e-06,
      "loss": 0.4908,
      "step": 239
    },
    {
      "epoch": 0.47182175622542594,
      "grad_norm": 0.30944961309432983,
      "learning_rate": 9.900970515805564e-06,
      "loss": 0.4907,
      "step": 240
    },
    {
      "epoch": 0.4737876802096986,
      "grad_norm": 0.38036513328552246,
      "learning_rate": 9.89868865569739e-06,
      "loss": 0.5004,
      "step": 241
    },
    {
      "epoch": 0.47575360419397117,
      "grad_norm": 0.3423393964767456,
      "learning_rate": 9.89638107360403e-06,
      "loss": 0.4859,
      "step": 242
    },
    {
      "epoch": 0.47771952817824376,
      "grad_norm": 0.38603687286376953,
      "learning_rate": 9.89404778164211e-06,
      "loss": 0.4858,
      "step": 243
    },
    {
      "epoch": 0.4796854521625164,
      "grad_norm": 0.2823747992515564,
      "learning_rate": 9.891688792063254e-06,
      "loss": 0.481,
      "step": 244
    },
    {
      "epoch": 0.481651376146789,
      "grad_norm": 0.39549919962882996,
      "learning_rate": 9.889304117254027e-06,
      "loss": 0.5326,
      "step": 245
    },
    {
      "epoch": 0.4836173001310616,
      "grad_norm": 0.3132837116718292,
      "learning_rate": 9.886893769735852e-06,
      "loss": 0.4933,
      "step": 246
    },
    {
      "epoch": 0.4855832241153342,
      "grad_norm": 0.36471930146217346,
      "learning_rate": 9.884457762164958e-06,
      "loss": 0.4974,
      "step": 247
    },
    {
      "epoch": 0.4875491480996068,
      "grad_norm": 0.32906249165534973,
      "learning_rate": 9.881996107332307e-06,
      "loss": 0.4802,
      "step": 248
    },
    {
      "epoch": 0.4895150720838794,
      "grad_norm": 0.3176659047603607,
      "learning_rate": 9.879508818163536e-06,
      "loss": 0.4793,
      "step": 249
    },
    {
      "epoch": 0.49148099606815204,
      "grad_norm": 0.32096564769744873,
      "learning_rate": 9.876995907718875e-06,
      "loss": 0.492,
      "step": 250
    },
    {
      "epoch": 0.49344692005242463,
      "grad_norm": 0.3479171693325043,
      "learning_rate": 9.87445738919309e-06,
      "loss": 0.4985,
      "step": 251
    },
    {
      "epoch": 0.4954128440366973,
      "grad_norm": 0.3736298978328705,
      "learning_rate": 9.871893275915408e-06,
      "loss": 0.4944,
      "step": 252
    },
    {
      "epoch": 0.49737876802096986,
      "grad_norm": 0.2852229177951813,
      "learning_rate": 9.869303581349448e-06,
      "loss": 0.4613,
      "step": 253
    },
    {
      "epoch": 0.49934469200524245,
      "grad_norm": 0.3456667363643646,
      "learning_rate": 9.866688319093155e-06,
      "loss": 0.4977,
      "step": 254
    },
    {
      "epoch": 0.5013106159895151,
      "grad_norm": 0.27879470586776733,
      "learning_rate": 9.864047502878717e-06,
      "loss": 0.4823,
      "step": 255
    },
    {
      "epoch": 0.5032765399737876,
      "grad_norm": 0.3423708975315094,
      "learning_rate": 9.86138114657251e-06,
      "loss": 0.4771,
      "step": 256
    },
    {
      "epoch": 0.5052424639580603,
      "grad_norm": 0.3409838080406189,
      "learning_rate": 9.858689264175012e-06,
      "loss": 0.4648,
      "step": 257
    },
    {
      "epoch": 0.5072083879423329,
      "grad_norm": 0.32440969347953796,
      "learning_rate": 9.855971869820726e-06,
      "loss": 0.4696,
      "step": 258
    },
    {
      "epoch": 0.5091743119266054,
      "grad_norm": 0.32682913541793823,
      "learning_rate": 9.853228977778125e-06,
      "loss": 0.4853,
      "step": 259
    },
    {
      "epoch": 0.5111402359108781,
      "grad_norm": 0.38436365127563477,
      "learning_rate": 9.850460602449555e-06,
      "loss": 0.4838,
      "step": 260
    },
    {
      "epoch": 0.5131061598951507,
      "grad_norm": 0.33962368965148926,
      "learning_rate": 9.847666758371175e-06,
      "loss": 0.5017,
      "step": 261
    },
    {
      "epoch": 0.5150720838794234,
      "grad_norm": 0.3401210606098175,
      "learning_rate": 9.844847460212875e-06,
      "loss": 0.4689,
      "step": 262
    },
    {
      "epoch": 0.5170380078636959,
      "grad_norm": 0.3694901168346405,
      "learning_rate": 9.842002722778191e-06,
      "loss": 0.4848,
      "step": 263
    },
    {
      "epoch": 0.5190039318479686,
      "grad_norm": 0.3879911005496979,
      "learning_rate": 9.83913256100425e-06,
      "loss": 0.4919,
      "step": 264
    },
    {
      "epoch": 0.5209698558322412,
      "grad_norm": 0.3491765856742859,
      "learning_rate": 9.836236989961665e-06,
      "loss": 0.5028,
      "step": 265
    },
    {
      "epoch": 0.5229357798165137,
      "grad_norm": 0.384046345949173,
      "learning_rate": 9.83331602485447e-06,
      "loss": 0.4799,
      "step": 266
    },
    {
      "epoch": 0.5249017038007864,
      "grad_norm": 0.35017964243888855,
      "learning_rate": 9.830369681020043e-06,
      "loss": 0.4887,
      "step": 267
    },
    {
      "epoch": 0.526867627785059,
      "grad_norm": 0.3668372929096222,
      "learning_rate": 9.827397973929012e-06,
      "loss": 0.4781,
      "step": 268
    },
    {
      "epoch": 0.5288335517693316,
      "grad_norm": 0.3488873243331909,
      "learning_rate": 9.82440091918519e-06,
      "loss": 0.4976,
      "step": 269
    },
    {
      "epoch": 0.5307994757536042,
      "grad_norm": 0.3492840528488159,
      "learning_rate": 9.821378532525479e-06,
      "loss": 0.4939,
      "step": 270
    },
    {
      "epoch": 0.5327653997378768,
      "grad_norm": 0.3887472450733185,
      "learning_rate": 9.8183308298198e-06,
      "loss": 0.5051,
      "step": 271
    },
    {
      "epoch": 0.5347313237221494,
      "grad_norm": 0.41635629534721375,
      "learning_rate": 9.815257827071e-06,
      "loss": 0.4955,
      "step": 272
    },
    {
      "epoch": 0.536697247706422,
      "grad_norm": 0.31790298223495483,
      "learning_rate": 9.812159540414766e-06,
      "loss": 0.4727,
      "step": 273
    },
    {
      "epoch": 0.5386631716906947,
      "grad_norm": 0.35766008496284485,
      "learning_rate": 9.809035986119553e-06,
      "loss": 0.4765,
      "step": 274
    },
    {
      "epoch": 0.5406290956749672,
      "grad_norm": 0.34865373373031616,
      "learning_rate": 9.80588718058649e-06,
      "loss": 0.4889,
      "step": 275
    },
    {
      "epoch": 0.5425950196592398,
      "grad_norm": 0.32705819606781006,
      "learning_rate": 9.802713140349294e-06,
      "loss": 0.5019,
      "step": 276
    },
    {
      "epoch": 0.5445609436435125,
      "grad_norm": 0.32511982321739197,
      "learning_rate": 9.799513882074182e-06,
      "loss": 0.4805,
      "step": 277
    },
    {
      "epoch": 0.546526867627785,
      "grad_norm": 0.3440779149532318,
      "learning_rate": 9.796289422559788e-06,
      "loss": 0.4611,
      "step": 278
    },
    {
      "epoch": 0.5484927916120577,
      "grad_norm": 0.3183061182498932,
      "learning_rate": 9.79303977873707e-06,
      "loss": 0.4733,
      "step": 279
    },
    {
      "epoch": 0.5504587155963303,
      "grad_norm": 0.33301064372062683,
      "learning_rate": 9.789764967669225e-06,
      "loss": 0.4995,
      "step": 280
    },
    {
      "epoch": 0.5524246395806028,
      "grad_norm": 0.3093130588531494,
      "learning_rate": 9.786465006551601e-06,
      "loss": 0.4795,
      "step": 281
    },
    {
      "epoch": 0.5543905635648755,
      "grad_norm": 0.31512442231178284,
      "learning_rate": 9.783139912711597e-06,
      "loss": 0.4864,
      "step": 282
    },
    {
      "epoch": 0.5563564875491481,
      "grad_norm": 0.3267965614795685,
      "learning_rate": 9.779789703608586e-06,
      "loss": 0.4891,
      "step": 283
    },
    {
      "epoch": 0.5583224115334207,
      "grad_norm": 0.30504485964775085,
      "learning_rate": 9.776414396833811e-06,
      "loss": 0.4658,
      "step": 284
    },
    {
      "epoch": 0.5602883355176933,
      "grad_norm": 0.33676230907440186,
      "learning_rate": 9.773014010110298e-06,
      "loss": 0.5032,
      "step": 285
    },
    {
      "epoch": 0.562254259501966,
      "grad_norm": 0.33661526441574097,
      "learning_rate": 9.769588561292762e-06,
      "loss": 0.4791,
      "step": 286
    },
    {
      "epoch": 0.5642201834862385,
      "grad_norm": 0.310126394033432,
      "learning_rate": 9.766138068367519e-06,
      "loss": 0.5007,
      "step": 287
    },
    {
      "epoch": 0.5661861074705111,
      "grad_norm": 0.35674116015434265,
      "learning_rate": 9.76266254945238e-06,
      "loss": 0.4751,
      "step": 288
    },
    {
      "epoch": 0.5681520314547838,
      "grad_norm": 0.3606201112270355,
      "learning_rate": 9.759162022796566e-06,
      "loss": 0.49,
      "step": 289
    },
    {
      "epoch": 0.5701179554390564,
      "grad_norm": 0.445045530796051,
      "learning_rate": 9.755636506780606e-06,
      "loss": 0.4909,
      "step": 290
    },
    {
      "epoch": 0.572083879423329,
      "grad_norm": 0.36715635657310486,
      "learning_rate": 9.752086019916246e-06,
      "loss": 0.5007,
      "step": 291
    },
    {
      "epoch": 0.5740498034076016,
      "grad_norm": 0.3720010221004486,
      "learning_rate": 9.748510580846344e-06,
      "loss": 0.5019,
      "step": 292
    },
    {
      "epoch": 0.5760157273918742,
      "grad_norm": 0.38443723320961,
      "learning_rate": 9.744910208344783e-06,
      "loss": 0.4956,
      "step": 293
    },
    {
      "epoch": 0.5779816513761468,
      "grad_norm": 0.3365614414215088,
      "learning_rate": 9.74128492131636e-06,
      "loss": 0.4902,
      "step": 294
    },
    {
      "epoch": 0.5799475753604194,
      "grad_norm": 0.35834208130836487,
      "learning_rate": 9.737634738796698e-06,
      "loss": 0.4822,
      "step": 295
    },
    {
      "epoch": 0.581913499344692,
      "grad_norm": 0.34508103132247925,
      "learning_rate": 9.733959679952136e-06,
      "loss": 0.4851,
      "step": 296
    },
    {
      "epoch": 0.5838794233289646,
      "grad_norm": 0.4581654965877533,
      "learning_rate": 9.730259764079636e-06,
      "loss": 0.502,
      "step": 297
    },
    {
      "epoch": 0.5858453473132372,
      "grad_norm": 0.34432968497276306,
      "learning_rate": 9.726535010606684e-06,
      "loss": 0.4866,
      "step": 298
    },
    {
      "epoch": 0.5878112712975099,
      "grad_norm": 0.43571075797080994,
      "learning_rate": 9.722785439091172e-06,
      "loss": 0.4901,
      "step": 299
    },
    {
      "epoch": 0.5897771952817824,
      "grad_norm": 0.3575364053249359,
      "learning_rate": 9.719011069221316e-06,
      "loss": 0.4866,
      "step": 300
    },
    {
      "epoch": 0.591743119266055,
      "grad_norm": 0.3271622061729431,
      "learning_rate": 9.715211920815538e-06,
      "loss": 0.4897,
      "step": 301
    },
    {
      "epoch": 0.5937090432503277,
      "grad_norm": 0.38964205980300903,
      "learning_rate": 9.711388013822369e-06,
      "loss": 0.4778,
      "step": 302
    },
    {
      "epoch": 0.5956749672346002,
      "grad_norm": 0.3184320330619812,
      "learning_rate": 9.70753936832034e-06,
      "loss": 0.5051,
      "step": 303
    },
    {
      "epoch": 0.5976408912188729,
      "grad_norm": 0.370286226272583,
      "learning_rate": 9.703666004517884e-06,
      "loss": 0.5047,
      "step": 304
    },
    {
      "epoch": 0.5996068152031455,
      "grad_norm": 0.32215046882629395,
      "learning_rate": 9.699767942753216e-06,
      "loss": 0.4933,
      "step": 305
    },
    {
      "epoch": 0.601572739187418,
      "grad_norm": 0.3631575107574463,
      "learning_rate": 9.695845203494242e-06,
      "loss": 0.4689,
      "step": 306
    },
    {
      "epoch": 0.6035386631716907,
      "grad_norm": 0.322794109582901,
      "learning_rate": 9.691897807338444e-06,
      "loss": 0.4796,
      "step": 307
    },
    {
      "epoch": 0.6055045871559633,
      "grad_norm": 0.3302377760410309,
      "learning_rate": 9.687925775012764e-06,
      "loss": 0.4847,
      "step": 308
    },
    {
      "epoch": 0.6074705111402359,
      "grad_norm": 0.3229452669620514,
      "learning_rate": 9.683929127373514e-06,
      "loss": 0.4777,
      "step": 309
    },
    {
      "epoch": 0.6094364351245085,
      "grad_norm": 0.29505693912506104,
      "learning_rate": 9.679907885406253e-06,
      "loss": 0.4902,
      "step": 310
    },
    {
      "epoch": 0.6114023591087812,
      "grad_norm": 0.37603944540023804,
      "learning_rate": 9.675862070225671e-06,
      "loss": 0.4877,
      "step": 311
    },
    {
      "epoch": 0.6133682830930537,
      "grad_norm": 0.3594703674316406,
      "learning_rate": 9.671791703075502e-06,
      "loss": 0.4912,
      "step": 312
    },
    {
      "epoch": 0.6153342070773263,
      "grad_norm": 0.32903850078582764,
      "learning_rate": 9.667696805328384e-06,
      "loss": 0.4944,
      "step": 313
    },
    {
      "epoch": 0.617300131061599,
      "grad_norm": 0.3076882064342499,
      "learning_rate": 9.663577398485765e-06,
      "loss": 0.4784,
      "step": 314
    },
    {
      "epoch": 0.6192660550458715,
      "grad_norm": 0.34379473328590393,
      "learning_rate": 9.659433504177786e-06,
      "loss": 0.4872,
      "step": 315
    },
    {
      "epoch": 0.6212319790301442,
      "grad_norm": 0.3046328127384186,
      "learning_rate": 9.655265144163164e-06,
      "loss": 0.4753,
      "step": 316
    },
    {
      "epoch": 0.6231979030144168,
      "grad_norm": 0.3317876160144806,
      "learning_rate": 9.651072340329087e-06,
      "loss": 0.4943,
      "step": 317
    },
    {
      "epoch": 0.6251638269986893,
      "grad_norm": 0.3519972860813141,
      "learning_rate": 9.646855114691081e-06,
      "loss": 0.4808,
      "step": 318
    },
    {
      "epoch": 0.627129750982962,
      "grad_norm": 0.34271880984306335,
      "learning_rate": 9.642613489392916e-06,
      "loss": 0.4735,
      "step": 319
    },
    {
      "epoch": 0.6290956749672346,
      "grad_norm": 0.31241729855537415,
      "learning_rate": 9.638347486706474e-06,
      "loss": 0.4605,
      "step": 320
    },
    {
      "epoch": 0.6310615989515073,
      "grad_norm": 0.33519551157951355,
      "learning_rate": 9.63405712903164e-06,
      "loss": 0.4917,
      "step": 321
    },
    {
      "epoch": 0.6330275229357798,
      "grad_norm": 0.33583033084869385,
      "learning_rate": 9.62974243889618e-06,
      "loss": 0.4685,
      "step": 322
    },
    {
      "epoch": 0.6349934469200524,
      "grad_norm": 0.33649152517318726,
      "learning_rate": 9.625403438955625e-06,
      "loss": 0.47,
      "step": 323
    },
    {
      "epoch": 0.6369593709043251,
      "grad_norm": 0.36066529154777527,
      "learning_rate": 9.621040151993153e-06,
      "loss": 0.4822,
      "step": 324
    },
    {
      "epoch": 0.6389252948885976,
      "grad_norm": 0.32451003789901733,
      "learning_rate": 9.616652600919468e-06,
      "loss": 0.4931,
      "step": 325
    },
    {
      "epoch": 0.6408912188728703,
      "grad_norm": 0.34756049513816833,
      "learning_rate": 9.61224080877268e-06,
      "loss": 0.4645,
      "step": 326
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 0.3787544071674347,
      "learning_rate": 9.607804798718182e-06,
      "loss": 0.463,
      "step": 327
    },
    {
      "epoch": 0.6448230668414154,
      "grad_norm": 0.3527866005897522,
      "learning_rate": 9.603344594048526e-06,
      "loss": 0.4617,
      "step": 328
    },
    {
      "epoch": 0.6467889908256881,
      "grad_norm": 0.353280633687973,
      "learning_rate": 9.598860218183318e-06,
      "loss": 0.4791,
      "step": 329
    },
    {
      "epoch": 0.6487549148099607,
      "grad_norm": 0.3094800114631653,
      "learning_rate": 9.59435169466907e-06,
      "loss": 0.4835,
      "step": 330
    },
    {
      "epoch": 0.6507208387942333,
      "grad_norm": 0.30767667293548584,
      "learning_rate": 9.589819047179094e-06,
      "loss": 0.4698,
      "step": 331
    },
    {
      "epoch": 0.6526867627785059,
      "grad_norm": 0.3043842315673828,
      "learning_rate": 9.585262299513366e-06,
      "loss": 0.4899,
      "step": 332
    },
    {
      "epoch": 0.6546526867627785,
      "grad_norm": 0.3159531056880951,
      "learning_rate": 9.580681475598413e-06,
      "loss": 0.469,
      "step": 333
    },
    {
      "epoch": 0.6566186107470511,
      "grad_norm": 0.3240654170513153,
      "learning_rate": 9.576076599487181e-06,
      "loss": 0.4888,
      "step": 334
    },
    {
      "epoch": 0.6585845347313237,
      "grad_norm": 0.32407113909721375,
      "learning_rate": 9.571447695358903e-06,
      "loss": 0.4726,
      "step": 335
    },
    {
      "epoch": 0.6605504587155964,
      "grad_norm": 0.33088815212249756,
      "learning_rate": 9.566794787518986e-06,
      "loss": 0.4531,
      "step": 336
    },
    {
      "epoch": 0.6625163826998689,
      "grad_norm": 0.3312188684940338,
      "learning_rate": 9.56211790039887e-06,
      "loss": 0.4933,
      "step": 337
    },
    {
      "epoch": 0.6644823066841415,
      "grad_norm": 0.34961023926734924,
      "learning_rate": 9.557417058555909e-06,
      "loss": 0.4705,
      "step": 338
    },
    {
      "epoch": 0.6664482306684142,
      "grad_norm": 0.3974532186985016,
      "learning_rate": 9.552692286673231e-06,
      "loss": 0.4846,
      "step": 339
    },
    {
      "epoch": 0.6684141546526867,
      "grad_norm": 0.3288216292858124,
      "learning_rate": 9.547943609559629e-06,
      "loss": 0.4617,
      "step": 340
    },
    {
      "epoch": 0.6703800786369594,
      "grad_norm": 0.3328869938850403,
      "learning_rate": 9.5431710521494e-06,
      "loss": 0.4801,
      "step": 341
    },
    {
      "epoch": 0.672346002621232,
      "grad_norm": 0.34561023116111755,
      "learning_rate": 9.538374639502247e-06,
      "loss": 0.4926,
      "step": 342
    },
    {
      "epoch": 0.6743119266055045,
      "grad_norm": 0.30892154574394226,
      "learning_rate": 9.533554396803123e-06,
      "loss": 0.4694,
      "step": 343
    },
    {
      "epoch": 0.6762778505897772,
      "grad_norm": 0.33325186371803284,
      "learning_rate": 9.528710349362112e-06,
      "loss": 0.4772,
      "step": 344
    },
    {
      "epoch": 0.6782437745740498,
      "grad_norm": 0.31278443336486816,
      "learning_rate": 9.523842522614285e-06,
      "loss": 0.4896,
      "step": 345
    },
    {
      "epoch": 0.6802096985583224,
      "grad_norm": 0.3059135377407074,
      "learning_rate": 9.518950942119582e-06,
      "loss": 0.5063,
      "step": 346
    },
    {
      "epoch": 0.682175622542595,
      "grad_norm": 0.400730162858963,
      "learning_rate": 9.514035633562662e-06,
      "loss": 0.4876,
      "step": 347
    },
    {
      "epoch": 0.6841415465268676,
      "grad_norm": 0.3564932942390442,
      "learning_rate": 9.509096622752781e-06,
      "loss": 0.4819,
      "step": 348
    },
    {
      "epoch": 0.6861074705111402,
      "grad_norm": 0.30216097831726074,
      "learning_rate": 9.504133935623643e-06,
      "loss": 0.4565,
      "step": 349
    },
    {
      "epoch": 0.6880733944954128,
      "grad_norm": 0.3681231737136841,
      "learning_rate": 9.49914759823328e-06,
      "loss": 0.4619,
      "step": 350
    },
    {
      "epoch": 0.6900393184796855,
      "grad_norm": 0.2910490334033966,
      "learning_rate": 9.4941376367639e-06,
      "loss": 0.4784,
      "step": 351
    },
    {
      "epoch": 0.6920052424639581,
      "grad_norm": 0.43464285135269165,
      "learning_rate": 9.489104077521761e-06,
      "loss": 0.4861,
      "step": 352
    },
    {
      "epoch": 0.6939711664482306,
      "grad_norm": 0.31047558784484863,
      "learning_rate": 9.484046946937021e-06,
      "loss": 0.4656,
      "step": 353
    },
    {
      "epoch": 0.6959370904325033,
      "grad_norm": 0.352581650018692,
      "learning_rate": 9.478966271563614e-06,
      "loss": 0.4823,
      "step": 354
    },
    {
      "epoch": 0.6979030144167759,
      "grad_norm": 0.3447401523590088,
      "learning_rate": 9.473862078079098e-06,
      "loss": 0.4749,
      "step": 355
    },
    {
      "epoch": 0.6998689384010485,
      "grad_norm": 0.3738485276699066,
      "learning_rate": 9.468734393284517e-06,
      "loss": 0.4895,
      "step": 356
    },
    {
      "epoch": 0.7018348623853211,
      "grad_norm": 0.32498571276664734,
      "learning_rate": 9.463583244104274e-06,
      "loss": 0.4939,
      "step": 357
    },
    {
      "epoch": 0.7038007863695938,
      "grad_norm": 0.31145179271698,
      "learning_rate": 9.458408657585968e-06,
      "loss": 0.466,
      "step": 358
    },
    {
      "epoch": 0.7057667103538663,
      "grad_norm": 0.323177307844162,
      "learning_rate": 9.453210660900264e-06,
      "loss": 0.5062,
      "step": 359
    },
    {
      "epoch": 0.7077326343381389,
      "grad_norm": 0.2810510993003845,
      "learning_rate": 9.447989281340753e-06,
      "loss": 0.4782,
      "step": 360
    },
    {
      "epoch": 0.7096985583224116,
      "grad_norm": 0.36067265272140503,
      "learning_rate": 9.442744546323804e-06,
      "loss": 0.4846,
      "step": 361
    },
    {
      "epoch": 0.7116644823066841,
      "grad_norm": 0.2921036183834076,
      "learning_rate": 9.437476483388415e-06,
      "loss": 0.4735,
      "step": 362
    },
    {
      "epoch": 0.7136304062909568,
      "grad_norm": 0.30914342403411865,
      "learning_rate": 9.43218512019608e-06,
      "loss": 0.4662,
      "step": 363
    },
    {
      "epoch": 0.7155963302752294,
      "grad_norm": 0.311714768409729,
      "learning_rate": 9.426870484530638e-06,
      "loss": 0.4682,
      "step": 364
    },
    {
      "epoch": 0.7175622542595019,
      "grad_norm": 0.3075977861881256,
      "learning_rate": 9.421532604298118e-06,
      "loss": 0.4855,
      "step": 365
    },
    {
      "epoch": 0.7195281782437746,
      "grad_norm": 0.318114310503006,
      "learning_rate": 9.416171507526615e-06,
      "loss": 0.4973,
      "step": 366
    },
    {
      "epoch": 0.7214941022280472,
      "grad_norm": 0.3103352189064026,
      "learning_rate": 9.410787222366117e-06,
      "loss": 0.4784,
      "step": 367
    },
    {
      "epoch": 0.7234600262123198,
      "grad_norm": 0.31439492106437683,
      "learning_rate": 9.405379777088379e-06,
      "loss": 0.4677,
      "step": 368
    },
    {
      "epoch": 0.7254259501965924,
      "grad_norm": 0.28842946887016296,
      "learning_rate": 9.399949200086757e-06,
      "loss": 0.482,
      "step": 369
    },
    {
      "epoch": 0.727391874180865,
      "grad_norm": 0.3343795835971832,
      "learning_rate": 9.394495519876071e-06,
      "loss": 0.4704,
      "step": 370
    },
    {
      "epoch": 0.7293577981651376,
      "grad_norm": 0.3141765892505646,
      "learning_rate": 9.38901876509245e-06,
      "loss": 0.455,
      "step": 371
    },
    {
      "epoch": 0.7313237221494102,
      "grad_norm": 0.28839895129203796,
      "learning_rate": 9.383518964493183e-06,
      "loss": 0.484,
      "step": 372
    },
    {
      "epoch": 0.7332896461336829,
      "grad_norm": 0.3266107141971588,
      "learning_rate": 9.377996146956569e-06,
      "loss": 0.4964,
      "step": 373
    },
    {
      "epoch": 0.7352555701179554,
      "grad_norm": 0.3106229901313782,
      "learning_rate": 9.372450341481762e-06,
      "loss": 0.4823,
      "step": 374
    },
    {
      "epoch": 0.737221494102228,
      "grad_norm": 0.30188578367233276,
      "learning_rate": 9.36688157718862e-06,
      "loss": 0.4625,
      "step": 375
    },
    {
      "epoch": 0.7391874180865007,
      "grad_norm": 0.35195842385292053,
      "learning_rate": 9.361289883317558e-06,
      "loss": 0.4947,
      "step": 376
    },
    {
      "epoch": 0.7411533420707732,
      "grad_norm": 0.31608274579048157,
      "learning_rate": 9.355675289229384e-06,
      "loss": 0.4843,
      "step": 377
    },
    {
      "epoch": 0.7431192660550459,
      "grad_norm": 0.3384999930858612,
      "learning_rate": 9.350037824405151e-06,
      "loss": 0.4908,
      "step": 378
    },
    {
      "epoch": 0.7450851900393185,
      "grad_norm": 0.3566296696662903,
      "learning_rate": 9.344377518446006e-06,
      "loss": 0.4744,
      "step": 379
    },
    {
      "epoch": 0.747051114023591,
      "grad_norm": 0.34384074807167053,
      "learning_rate": 9.338694401073026e-06,
      "loss": 0.4744,
      "step": 380
    },
    {
      "epoch": 0.7490170380078637,
      "grad_norm": 0.3111206889152527,
      "learning_rate": 9.332988502127063e-06,
      "loss": 0.4749,
      "step": 381
    },
    {
      "epoch": 0.7509829619921363,
      "grad_norm": 0.29371377825737,
      "learning_rate": 9.327259851568596e-06,
      "loss": 0.4849,
      "step": 382
    },
    {
      "epoch": 0.752948885976409,
      "grad_norm": 0.2952675521373749,
      "learning_rate": 9.321508479477568e-06,
      "loss": 0.464,
      "step": 383
    },
    {
      "epoch": 0.7549148099606815,
      "grad_norm": 0.2904905676841736,
      "learning_rate": 9.315734416053223e-06,
      "loss": 0.47,
      "step": 384
    },
    {
      "epoch": 0.7568807339449541,
      "grad_norm": 0.30284953117370605,
      "learning_rate": 9.309937691613957e-06,
      "loss": 0.4658,
      "step": 385
    },
    {
      "epoch": 0.7588466579292268,
      "grad_norm": 0.28848186135292053,
      "learning_rate": 9.30411833659715e-06,
      "loss": 0.4643,
      "step": 386
    },
    {
      "epoch": 0.7608125819134993,
      "grad_norm": 0.32645782828330994,
      "learning_rate": 9.298276381559015e-06,
      "loss": 0.489,
      "step": 387
    },
    {
      "epoch": 0.762778505897772,
      "grad_norm": 0.3258381187915802,
      "learning_rate": 9.29241185717443e-06,
      "loss": 0.4827,
      "step": 388
    },
    {
      "epoch": 0.7647444298820446,
      "grad_norm": 0.41076749563217163,
      "learning_rate": 9.286524794236783e-06,
      "loss": 0.4868,
      "step": 389
    },
    {
      "epoch": 0.7667103538663171,
      "grad_norm": 0.299274742603302,
      "learning_rate": 9.280615223657801e-06,
      "loss": 0.48,
      "step": 390
    },
    {
      "epoch": 0.7686762778505898,
      "grad_norm": 0.4232471287250519,
      "learning_rate": 9.2746831764674e-06,
      "loss": 0.4899,
      "step": 391
    },
    {
      "epoch": 0.7706422018348624,
      "grad_norm": 0.3225405514240265,
      "learning_rate": 9.268728683813517e-06,
      "loss": 0.4911,
      "step": 392
    },
    {
      "epoch": 0.772608125819135,
      "grad_norm": 0.4578843116760254,
      "learning_rate": 9.262751776961936e-06,
      "loss": 0.5114,
      "step": 393
    },
    {
      "epoch": 0.7745740498034076,
      "grad_norm": 0.32411909103393555,
      "learning_rate": 9.256752487296142e-06,
      "loss": 0.4885,
      "step": 394
    },
    {
      "epoch": 0.7765399737876802,
      "grad_norm": 0.3600336015224457,
      "learning_rate": 9.250730846317145e-06,
      "loss": 0.4759,
      "step": 395
    },
    {
      "epoch": 0.7785058977719528,
      "grad_norm": 0.2981262505054474,
      "learning_rate": 9.24468688564332e-06,
      "loss": 0.4608,
      "step": 396
    },
    {
      "epoch": 0.7804718217562254,
      "grad_norm": 0.347411185503006,
      "learning_rate": 9.23862063701023e-06,
      "loss": 0.4833,
      "step": 397
    },
    {
      "epoch": 0.7824377457404981,
      "grad_norm": 0.34579214453697205,
      "learning_rate": 9.232532132270472e-06,
      "loss": 0.4734,
      "step": 398
    },
    {
      "epoch": 0.7844036697247706,
      "grad_norm": 0.3555125594139099,
      "learning_rate": 9.226421403393513e-06,
      "loss": 0.4809,
      "step": 399
    },
    {
      "epoch": 0.7863695937090432,
      "grad_norm": 0.3262729346752167,
      "learning_rate": 9.220288482465499e-06,
      "loss": 0.4746,
      "step": 400
    },
    {
      "epoch": 0.7883355176933159,
      "grad_norm": 0.39413920044898987,
      "learning_rate": 9.214133401689113e-06,
      "loss": 0.48,
      "step": 401
    },
    {
      "epoch": 0.7903014416775884,
      "grad_norm": 0.33590251207351685,
      "learning_rate": 9.207956193383392e-06,
      "loss": 0.4907,
      "step": 402
    },
    {
      "epoch": 0.7922673656618611,
      "grad_norm": 0.3144283592700958,
      "learning_rate": 9.201756889983557e-06,
      "loss": 0.4786,
      "step": 403
    },
    {
      "epoch": 0.7942332896461337,
      "grad_norm": 0.4004082977771759,
      "learning_rate": 9.19553552404085e-06,
      "loss": 0.4879,
      "step": 404
    },
    {
      "epoch": 0.7961992136304062,
      "grad_norm": 0.31236234307289124,
      "learning_rate": 9.189292128222355e-06,
      "loss": 0.4764,
      "step": 405
    },
    {
      "epoch": 0.7981651376146789,
      "grad_norm": 0.4149709939956665,
      "learning_rate": 9.183026735310834e-06,
      "loss": 0.4581,
      "step": 406
    },
    {
      "epoch": 0.8001310615989515,
      "grad_norm": 0.3310157358646393,
      "learning_rate": 9.17673937820455e-06,
      "loss": 0.4735,
      "step": 407
    },
    {
      "epoch": 0.8020969855832241,
      "grad_norm": 0.37418389320373535,
      "learning_rate": 9.170430089917089e-06,
      "loss": 0.4939,
      "step": 408
    },
    {
      "epoch": 0.8040629095674967,
      "grad_norm": 0.3040340840816498,
      "learning_rate": 9.164098903577203e-06,
      "loss": 0.4569,
      "step": 409
    },
    {
      "epoch": 0.8060288335517694,
      "grad_norm": 0.37164947390556335,
      "learning_rate": 9.157745852428624e-06,
      "loss": 0.4764,
      "step": 410
    },
    {
      "epoch": 0.8079947575360419,
      "grad_norm": 0.33743494749069214,
      "learning_rate": 9.151370969829883e-06,
      "loss": 0.4776,
      "step": 411
    },
    {
      "epoch": 0.8099606815203145,
      "grad_norm": 0.39068764448165894,
      "learning_rate": 9.144974289254154e-06,
      "loss": 0.4812,
      "step": 412
    },
    {
      "epoch": 0.8119266055045872,
      "grad_norm": 0.3487973213195801,
      "learning_rate": 9.138555844289061e-06,
      "loss": 0.4847,
      "step": 413
    },
    {
      "epoch": 0.8138925294888598,
      "grad_norm": 0.42854028940200806,
      "learning_rate": 9.132115668636512e-06,
      "loss": 0.4738,
      "step": 414
    },
    {
      "epoch": 0.8158584534731324,
      "grad_norm": 0.2957235276699066,
      "learning_rate": 9.125653796112514e-06,
      "loss": 0.4461,
      "step": 415
    },
    {
      "epoch": 0.817824377457405,
      "grad_norm": 0.44258782267570496,
      "learning_rate": 9.119170260647005e-06,
      "loss": 0.5026,
      "step": 416
    },
    {
      "epoch": 0.8197903014416776,
      "grad_norm": 0.334903359413147,
      "learning_rate": 9.112665096283668e-06,
      "loss": 0.4774,
      "step": 417
    },
    {
      "epoch": 0.8217562254259502,
      "grad_norm": 0.3543790876865387,
      "learning_rate": 9.106138337179754e-06,
      "loss": 0.4854,
      "step": 418
    },
    {
      "epoch": 0.8237221494102228,
      "grad_norm": 0.3583733141422272,
      "learning_rate": 9.099590017605903e-06,
      "loss": 0.48,
      "step": 419
    },
    {
      "epoch": 0.8256880733944955,
      "grad_norm": 0.3436172902584076,
      "learning_rate": 9.093020171945966e-06,
      "loss": 0.45,
      "step": 420
    },
    {
      "epoch": 0.827653997378768,
      "grad_norm": 0.3209732472896576,
      "learning_rate": 9.086428834696824e-06,
      "loss": 0.4821,
      "step": 421
    },
    {
      "epoch": 0.8296199213630406,
      "grad_norm": 0.37522175908088684,
      "learning_rate": 9.079816040468201e-06,
      "loss": 0.4611,
      "step": 422
    },
    {
      "epoch": 0.8315858453473133,
      "grad_norm": 0.30042916536331177,
      "learning_rate": 9.073181823982495e-06,
      "loss": 0.5015,
      "step": 423
    },
    {
      "epoch": 0.8335517693315858,
      "grad_norm": 0.3324982523918152,
      "learning_rate": 9.066526220074582e-06,
      "loss": 0.4796,
      "step": 424
    },
    {
      "epoch": 0.8355176933158585,
      "grad_norm": 0.352521151304245,
      "learning_rate": 9.059849263691638e-06,
      "loss": 0.4543,
      "step": 425
    },
    {
      "epoch": 0.8374836173001311,
      "grad_norm": 0.2786426842212677,
      "learning_rate": 9.05315098989296e-06,
      "loss": 0.4571,
      "step": 426
    },
    {
      "epoch": 0.8394495412844036,
      "grad_norm": 0.3154030740261078,
      "learning_rate": 9.046431433849778e-06,
      "loss": 0.478,
      "step": 427
    },
    {
      "epoch": 0.8414154652686763,
      "grad_norm": 0.29987379908561707,
      "learning_rate": 9.039690630845067e-06,
      "loss": 0.4688,
      "step": 428
    },
    {
      "epoch": 0.8433813892529489,
      "grad_norm": 0.3286818861961365,
      "learning_rate": 9.032928616273369e-06,
      "loss": 0.479,
      "step": 429
    },
    {
      "epoch": 0.8453473132372215,
      "grad_norm": 0.3046414852142334,
      "learning_rate": 9.026145425640601e-06,
      "loss": 0.4758,
      "step": 430
    },
    {
      "epoch": 0.8473132372214941,
      "grad_norm": 0.32858696579933167,
      "learning_rate": 9.019341094563875e-06,
      "loss": 0.4703,
      "step": 431
    },
    {
      "epoch": 0.8492791612057667,
      "grad_norm": 0.29131099581718445,
      "learning_rate": 9.012515658771301e-06,
      "loss": 0.4745,
      "step": 432
    },
    {
      "epoch": 0.8512450851900393,
      "grad_norm": 0.3147415220737457,
      "learning_rate": 9.005669154101811e-06,
      "loss": 0.4996,
      "step": 433
    },
    {
      "epoch": 0.8532110091743119,
      "grad_norm": 0.32791829109191895,
      "learning_rate": 8.998801616504962e-06,
      "loss": 0.4642,
      "step": 434
    },
    {
      "epoch": 0.8551769331585846,
      "grad_norm": 0.3501948416233063,
      "learning_rate": 8.991913082040752e-06,
      "loss": 0.4884,
      "step": 435
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.3144097328186035,
      "learning_rate": 8.985003586879428e-06,
      "loss": 0.4627,
      "step": 436
    },
    {
      "epoch": 0.8591087811271297,
      "grad_norm": 0.32123497128486633,
      "learning_rate": 8.978073167301298e-06,
      "loss": 0.4683,
      "step": 437
    },
    {
      "epoch": 0.8610747051114024,
      "grad_norm": 0.3028917610645294,
      "learning_rate": 8.971121859696539e-06,
      "loss": 0.4688,
      "step": 438
    },
    {
      "epoch": 0.8630406290956749,
      "grad_norm": 0.29881271719932556,
      "learning_rate": 8.964149700565006e-06,
      "loss": 0.466,
      "step": 439
    },
    {
      "epoch": 0.8650065530799476,
      "grad_norm": 0.33461838960647583,
      "learning_rate": 8.957156726516047e-06,
      "loss": 0.4834,
      "step": 440
    },
    {
      "epoch": 0.8669724770642202,
      "grad_norm": 0.29429683089256287,
      "learning_rate": 8.950142974268295e-06,
      "loss": 0.4531,
      "step": 441
    },
    {
      "epoch": 0.8689384010484927,
      "grad_norm": 0.3107634484767914,
      "learning_rate": 8.943108480649492e-06,
      "loss": 0.4488,
      "step": 442
    },
    {
      "epoch": 0.8709043250327654,
      "grad_norm": 0.3410981595516205,
      "learning_rate": 8.936053282596284e-06,
      "loss": 0.4803,
      "step": 443
    },
    {
      "epoch": 0.872870249017038,
      "grad_norm": 0.2996877133846283,
      "learning_rate": 8.928977417154037e-06,
      "loss": 0.469,
      "step": 444
    },
    {
      "epoch": 0.8748361730013107,
      "grad_norm": 0.3480963408946991,
      "learning_rate": 8.921880921476634e-06,
      "loss": 0.4671,
      "step": 445
    },
    {
      "epoch": 0.8768020969855832,
      "grad_norm": 0.2955884635448456,
      "learning_rate": 8.914763832826282e-06,
      "loss": 0.4607,
      "step": 446
    },
    {
      "epoch": 0.8787680209698558,
      "grad_norm": 0.34652408957481384,
      "learning_rate": 8.907626188573319e-06,
      "loss": 0.4946,
      "step": 447
    },
    {
      "epoch": 0.8807339449541285,
      "grad_norm": 0.3643608093261719,
      "learning_rate": 8.900468026196017e-06,
      "loss": 0.4721,
      "step": 448
    },
    {
      "epoch": 0.882699868938401,
      "grad_norm": 0.26901280879974365,
      "learning_rate": 8.893289383280379e-06,
      "loss": 0.4799,
      "step": 449
    },
    {
      "epoch": 0.8846657929226737,
      "grad_norm": 0.36749908328056335,
      "learning_rate": 8.886090297519956e-06,
      "loss": 0.4523,
      "step": 450
    },
    {
      "epoch": 0.8866317169069463,
      "grad_norm": 0.3095279932022095,
      "learning_rate": 8.878870806715631e-06,
      "loss": 0.4628,
      "step": 451
    },
    {
      "epoch": 0.8885976408912188,
      "grad_norm": 0.3318431079387665,
      "learning_rate": 8.871630948775438e-06,
      "loss": 0.4787,
      "step": 452
    },
    {
      "epoch": 0.8905635648754915,
      "grad_norm": 0.3512536585330963,
      "learning_rate": 8.864370761714348e-06,
      "loss": 0.4677,
      "step": 453
    },
    {
      "epoch": 0.8925294888597641,
      "grad_norm": 0.2928638756275177,
      "learning_rate": 8.857090283654078e-06,
      "loss": 0.4718,
      "step": 454
    },
    {
      "epoch": 0.8944954128440367,
      "grad_norm": 0.2978120744228363,
      "learning_rate": 8.84978955282289e-06,
      "loss": 0.4771,
      "step": 455
    },
    {
      "epoch": 0.8964613368283093,
      "grad_norm": 0.2687295079231262,
      "learning_rate": 8.842468607555389e-06,
      "loss": 0.4753,
      "step": 456
    },
    {
      "epoch": 0.898427260812582,
      "grad_norm": 0.2807966470718384,
      "learning_rate": 8.835127486292318e-06,
      "loss": 0.456,
      "step": 457
    },
    {
      "epoch": 0.9003931847968545,
      "grad_norm": 0.25572317838668823,
      "learning_rate": 8.827766227580366e-06,
      "loss": 0.4701,
      "step": 458
    },
    {
      "epoch": 0.9023591087811271,
      "grad_norm": 0.30143314599990845,
      "learning_rate": 8.820384870071951e-06,
      "loss": 0.4743,
      "step": 459
    },
    {
      "epoch": 0.9043250327653998,
      "grad_norm": 0.2788403034210205,
      "learning_rate": 8.812983452525038e-06,
      "loss": 0.4877,
      "step": 460
    },
    {
      "epoch": 0.9062909567496723,
      "grad_norm": 0.3468334674835205,
      "learning_rate": 8.805562013802914e-06,
      "loss": 0.4678,
      "step": 461
    },
    {
      "epoch": 0.908256880733945,
      "grad_norm": 0.2584637403488159,
      "learning_rate": 8.79812059287399e-06,
      "loss": 0.4737,
      "step": 462
    },
    {
      "epoch": 0.9102228047182176,
      "grad_norm": 0.39624977111816406,
      "learning_rate": 8.790659228811609e-06,
      "loss": 0.4647,
      "step": 463
    },
    {
      "epoch": 0.9121887287024901,
      "grad_norm": 0.3191069960594177,
      "learning_rate": 8.783177960793825e-06,
      "loss": 0.4876,
      "step": 464
    },
    {
      "epoch": 0.9141546526867628,
      "grad_norm": 0.3358793258666992,
      "learning_rate": 8.775676828103205e-06,
      "loss": 0.4643,
      "step": 465
    },
    {
      "epoch": 0.9161205766710354,
      "grad_norm": 0.3253157436847687,
      "learning_rate": 8.768155870126622e-06,
      "loss": 0.4637,
      "step": 466
    },
    {
      "epoch": 0.918086500655308,
      "grad_norm": 0.33859390020370483,
      "learning_rate": 8.760615126355045e-06,
      "loss": 0.482,
      "step": 467
    },
    {
      "epoch": 0.9200524246395806,
      "grad_norm": 0.37952473759651184,
      "learning_rate": 8.753054636383336e-06,
      "loss": 0.4976,
      "step": 468
    },
    {
      "epoch": 0.9220183486238532,
      "grad_norm": 0.3037833273410797,
      "learning_rate": 8.745474439910043e-06,
      "loss": 0.4541,
      "step": 469
    },
    {
      "epoch": 0.9239842726081258,
      "grad_norm": 0.32147467136383057,
      "learning_rate": 8.73787457673718e-06,
      "loss": 0.4663,
      "step": 470
    },
    {
      "epoch": 0.9259501965923984,
      "grad_norm": 0.30466553568840027,
      "learning_rate": 8.730255086770037e-06,
      "loss": 0.4746,
      "step": 471
    },
    {
      "epoch": 0.927916120576671,
      "grad_norm": 0.2942729592323303,
      "learning_rate": 8.722616010016953e-06,
      "loss": 0.4775,
      "step": 472
    },
    {
      "epoch": 0.9298820445609436,
      "grad_norm": 0.30535486340522766,
      "learning_rate": 8.714957386589115e-06,
      "loss": 0.4621,
      "step": 473
    },
    {
      "epoch": 0.9318479685452162,
      "grad_norm": 0.30644193291664124,
      "learning_rate": 8.707279256700348e-06,
      "loss": 0.4674,
      "step": 474
    },
    {
      "epoch": 0.9338138925294889,
      "grad_norm": 0.314421147108078,
      "learning_rate": 8.699581660666897e-06,
      "loss": 0.4773,
      "step": 475
    },
    {
      "epoch": 0.9357798165137615,
      "grad_norm": 0.31318455934524536,
      "learning_rate": 8.691864638907224e-06,
      "loss": 0.4622,
      "step": 476
    },
    {
      "epoch": 0.937745740498034,
      "grad_norm": 0.3173219561576843,
      "learning_rate": 8.684128231941789e-06,
      "loss": 0.4775,
      "step": 477
    },
    {
      "epoch": 0.9397116644823067,
      "grad_norm": 0.31080108880996704,
      "learning_rate": 8.67637248039284e-06,
      "loss": 0.4701,
      "step": 478
    },
    {
      "epoch": 0.9416775884665793,
      "grad_norm": 0.2851126790046692,
      "learning_rate": 8.668597424984196e-06,
      "loss": 0.4605,
      "step": 479
    },
    {
      "epoch": 0.9436435124508519,
      "grad_norm": 0.2801235318183899,
      "learning_rate": 8.660803106541044e-06,
      "loss": 0.456,
      "step": 480
    },
    {
      "epoch": 0.9456094364351245,
      "grad_norm": 0.31265968084335327,
      "learning_rate": 8.65298956598971e-06,
      "loss": 0.4688,
      "step": 481
    },
    {
      "epoch": 0.9475753604193972,
      "grad_norm": 0.29262563586235046,
      "learning_rate": 8.64515684435746e-06,
      "loss": 0.4576,
      "step": 482
    },
    {
      "epoch": 0.9495412844036697,
      "grad_norm": 0.3408479690551758,
      "learning_rate": 8.637304982772263e-06,
      "loss": 0.4826,
      "step": 483
    },
    {
      "epoch": 0.9515072083879423,
      "grad_norm": 0.3060987889766693,
      "learning_rate": 8.629434022462598e-06,
      "loss": 0.4606,
      "step": 484
    },
    {
      "epoch": 0.953473132372215,
      "grad_norm": 0.29849743843078613,
      "learning_rate": 8.621544004757226e-06,
      "loss": 0.476,
      "step": 485
    },
    {
      "epoch": 0.9554390563564875,
      "grad_norm": 0.31482797861099243,
      "learning_rate": 8.613634971084967e-06,
      "loss": 0.4533,
      "step": 486
    },
    {
      "epoch": 0.9574049803407602,
      "grad_norm": 0.32803142070770264,
      "learning_rate": 8.605706962974503e-06,
      "loss": 0.4972,
      "step": 487
    },
    {
      "epoch": 0.9593709043250328,
      "grad_norm": 0.30527326464653015,
      "learning_rate": 8.597760022054134e-06,
      "loss": 0.4785,
      "step": 488
    },
    {
      "epoch": 0.9613368283093053,
      "grad_norm": 0.2986467182636261,
      "learning_rate": 8.589794190051582e-06,
      "loss": 0.4817,
      "step": 489
    },
    {
      "epoch": 0.963302752293578,
      "grad_norm": 0.2935220003128052,
      "learning_rate": 8.581809508793753e-06,
      "loss": 0.4977,
      "step": 490
    },
    {
      "epoch": 0.9652686762778506,
      "grad_norm": 0.3608831465244293,
      "learning_rate": 8.573806020206534e-06,
      "loss": 0.4872,
      "step": 491
    },
    {
      "epoch": 0.9672346002621232,
      "grad_norm": 0.2806779146194458,
      "learning_rate": 8.56578376631456e-06,
      "loss": 0.4599,
      "step": 492
    },
    {
      "epoch": 0.9692005242463958,
      "grad_norm": 0.3233669400215149,
      "learning_rate": 8.557742789241003e-06,
      "loss": 0.4578,
      "step": 493
    },
    {
      "epoch": 0.9711664482306684,
      "grad_norm": 0.29287710785865784,
      "learning_rate": 8.549683131207347e-06,
      "loss": 0.4728,
      "step": 494
    },
    {
      "epoch": 0.973132372214941,
      "grad_norm": 0.31075453758239746,
      "learning_rate": 8.541604834533159e-06,
      "loss": 0.4808,
      "step": 495
    },
    {
      "epoch": 0.9750982961992136,
      "grad_norm": 0.314624547958374,
      "learning_rate": 8.533507941635882e-06,
      "loss": 0.4746,
      "step": 496
    },
    {
      "epoch": 0.9770642201834863,
      "grad_norm": 0.2979452311992645,
      "learning_rate": 8.525392495030596e-06,
      "loss": 0.4732,
      "step": 497
    },
    {
      "epoch": 0.9790301441677588,
      "grad_norm": 0.2798508405685425,
      "learning_rate": 8.51725853732981e-06,
      "loss": 0.4902,
      "step": 498
    },
    {
      "epoch": 0.9809960681520314,
      "grad_norm": 0.32786044478416443,
      "learning_rate": 8.509106111243223e-06,
      "loss": 0.4829,
      "step": 499
    },
    {
      "epoch": 0.9829619921363041,
      "grad_norm": 0.32859888672828674,
      "learning_rate": 8.500935259577517e-06,
      "loss": 0.4585,
      "step": 500
    },
    {
      "epoch": 0.9849279161205766,
      "grad_norm": 0.2928418815135956,
      "learning_rate": 8.492746025236113e-06,
      "loss": 0.4396,
      "step": 501
    },
    {
      "epoch": 0.9868938401048493,
      "grad_norm": 0.32088446617126465,
      "learning_rate": 8.484538451218966e-06,
      "loss": 0.4814,
      "step": 502
    },
    {
      "epoch": 0.9888597640891219,
      "grad_norm": 0.33992063999176025,
      "learning_rate": 8.476312580622318e-06,
      "loss": 0.4865,
      "step": 503
    },
    {
      "epoch": 0.9908256880733946,
      "grad_norm": 0.39013153314590454,
      "learning_rate": 8.468068456638491e-06,
      "loss": 0.4599,
      "step": 504
    },
    {
      "epoch": 0.9927916120576671,
      "grad_norm": 0.33563679456710815,
      "learning_rate": 8.459806122555648e-06,
      "loss": 0.4696,
      "step": 505
    },
    {
      "epoch": 0.9947575360419397,
      "grad_norm": 0.4276735484600067,
      "learning_rate": 8.451525621757571e-06,
      "loss": 0.4824,
      "step": 506
    },
    {
      "epoch": 0.9967234600262124,
      "grad_norm": 0.2916342616081238,
      "learning_rate": 8.443226997723426e-06,
      "loss": 0.463,
      "step": 507
    },
    {
      "epoch": 0.9986893840104849,
      "grad_norm": 0.4475458860397339,
      "learning_rate": 8.434910294027547e-06,
      "loss": 0.4787,
      "step": 508
    },
    {
      "epoch": 1.0006553079947575,
      "grad_norm": 0.42703890800476074,
      "learning_rate": 8.4265755543392e-06,
      "loss": 0.6154,
      "step": 509
    },
    {
      "epoch": 1.0026212319790302,
      "grad_norm": 0.41458940505981445,
      "learning_rate": 8.418222822422348e-06,
      "loss": 0.4612,
      "step": 510
    },
    {
      "epoch": 1.0045871559633028,
      "grad_norm": 0.34233203530311584,
      "learning_rate": 8.409852142135438e-06,
      "loss": 0.4433,
      "step": 511
    },
    {
      "epoch": 1.0065530799475753,
      "grad_norm": 0.32114264369010925,
      "learning_rate": 8.40146355743115e-06,
      "loss": 0.4444,
      "step": 512
    },
    {
      "epoch": 1.008519003931848,
      "grad_norm": 0.30860528349876404,
      "learning_rate": 8.393057112356181e-06,
      "loss": 0.435,
      "step": 513
    },
    {
      "epoch": 1.0104849279161205,
      "grad_norm": 0.31313031911849976,
      "learning_rate": 8.38463285105101e-06,
      "loss": 0.4331,
      "step": 514
    },
    {
      "epoch": 1.0124508519003932,
      "grad_norm": 0.32860052585601807,
      "learning_rate": 8.376190817749658e-06,
      "loss": 0.4323,
      "step": 515
    },
    {
      "epoch": 1.0144167758846658,
      "grad_norm": 0.3043825924396515,
      "learning_rate": 8.367731056779476e-06,
      "loss": 0.4568,
      "step": 516
    },
    {
      "epoch": 1.0163826998689385,
      "grad_norm": 0.295673131942749,
      "learning_rate": 8.35925361256089e-06,
      "loss": 0.4379,
      "step": 517
    },
    {
      "epoch": 1.018348623853211,
      "grad_norm": 0.314635306596756,
      "learning_rate": 8.350758529607174e-06,
      "loss": 0.4407,
      "step": 518
    },
    {
      "epoch": 1.0203145478374835,
      "grad_norm": 0.3012637794017792,
      "learning_rate": 8.342245852524229e-06,
      "loss": 0.4449,
      "step": 519
    },
    {
      "epoch": 1.0222804718217562,
      "grad_norm": 0.3112142086029053,
      "learning_rate": 8.333715626010334e-06,
      "loss": 0.4563,
      "step": 520
    },
    {
      "epoch": 1.0242463958060288,
      "grad_norm": 0.33864012360572815,
      "learning_rate": 8.325167894855917e-06,
      "loss": 0.4624,
      "step": 521
    },
    {
      "epoch": 1.0262123197903015,
      "grad_norm": 0.31816527247428894,
      "learning_rate": 8.316602703943315e-06,
      "loss": 0.4403,
      "step": 522
    },
    {
      "epoch": 1.0281782437745741,
      "grad_norm": 0.32534271478652954,
      "learning_rate": 8.308020098246552e-06,
      "loss": 0.4631,
      "step": 523
    },
    {
      "epoch": 1.0301441677588468,
      "grad_norm": 0.3206179141998291,
      "learning_rate": 8.299420122831084e-06,
      "loss": 0.4498,
      "step": 524
    },
    {
      "epoch": 1.0321100917431192,
      "grad_norm": 0.33189356327056885,
      "learning_rate": 8.290802822853576e-06,
      "loss": 0.4574,
      "step": 525
    },
    {
      "epoch": 1.0340760157273918,
      "grad_norm": 0.3216302990913391,
      "learning_rate": 8.28216824356166e-06,
      "loss": 0.4702,
      "step": 526
    },
    {
      "epoch": 1.0360419397116645,
      "grad_norm": 0.2860296666622162,
      "learning_rate": 8.2735164302937e-06,
      "loss": 0.4557,
      "step": 527
    },
    {
      "epoch": 1.0380078636959371,
      "grad_norm": 0.46882984042167664,
      "learning_rate": 8.26484742847855e-06,
      "loss": 0.4488,
      "step": 528
    },
    {
      "epoch": 1.0399737876802098,
      "grad_norm": 0.2686419188976288,
      "learning_rate": 8.256161283635315e-06,
      "loss": 0.4468,
      "step": 529
    },
    {
      "epoch": 1.0419397116644824,
      "grad_norm": 0.39462658762931824,
      "learning_rate": 8.24745804137312e-06,
      "loss": 0.4587,
      "step": 530
    },
    {
      "epoch": 1.0439056356487548,
      "grad_norm": 0.3102317750453949,
      "learning_rate": 8.238737747390859e-06,
      "loss": 0.4521,
      "step": 531
    },
    {
      "epoch": 1.0458715596330275,
      "grad_norm": 0.42338991165161133,
      "learning_rate": 8.230000447476968e-06,
      "loss": 0.4457,
      "step": 532
    },
    {
      "epoch": 1.0478374836173001,
      "grad_norm": 0.2850761413574219,
      "learning_rate": 8.22124618750917e-06,
      "loss": 0.4742,
      "step": 533
    },
    {
      "epoch": 1.0498034076015728,
      "grad_norm": 0.397907555103302,
      "learning_rate": 8.212475013454249e-06,
      "loss": 0.459,
      "step": 534
    },
    {
      "epoch": 1.0517693315858454,
      "grad_norm": 0.38815879821777344,
      "learning_rate": 8.203686971367795e-06,
      "loss": 0.4502,
      "step": 535
    },
    {
      "epoch": 1.053735255570118,
      "grad_norm": 0.33456432819366455,
      "learning_rate": 8.19488210739397e-06,
      "loss": 0.4383,
      "step": 536
    },
    {
      "epoch": 1.0557011795543905,
      "grad_norm": 0.2970755696296692,
      "learning_rate": 8.186060467765268e-06,
      "loss": 0.459,
      "step": 537
    },
    {
      "epoch": 1.0576671035386631,
      "grad_norm": 0.34444984793663025,
      "learning_rate": 8.177222098802264e-06,
      "loss": 0.4477,
      "step": 538
    },
    {
      "epoch": 1.0596330275229358,
      "grad_norm": 0.30864620208740234,
      "learning_rate": 8.16836704691338e-06,
      "loss": 0.4542,
      "step": 539
    },
    {
      "epoch": 1.0615989515072084,
      "grad_norm": 0.3233746588230133,
      "learning_rate": 8.159495358594627e-06,
      "loss": 0.4541,
      "step": 540
    },
    {
      "epoch": 1.063564875491481,
      "grad_norm": 0.35800066590309143,
      "learning_rate": 8.150607080429385e-06,
      "loss": 0.4482,
      "step": 541
    },
    {
      "epoch": 1.0655307994757537,
      "grad_norm": 0.33229756355285645,
      "learning_rate": 8.14170225908813e-06,
      "loss": 0.4291,
      "step": 542
    },
    {
      "epoch": 1.067496723460026,
      "grad_norm": 0.3414958417415619,
      "learning_rate": 8.13278094132821e-06,
      "loss": 0.4362,
      "step": 543
    },
    {
      "epoch": 1.0694626474442988,
      "grad_norm": 0.3265877962112427,
      "learning_rate": 8.123843173993592e-06,
      "loss": 0.4422,
      "step": 544
    },
    {
      "epoch": 1.0714285714285714,
      "grad_norm": 0.3254944980144501,
      "learning_rate": 8.114889004014612e-06,
      "loss": 0.4601,
      "step": 545
    },
    {
      "epoch": 1.073394495412844,
      "grad_norm": 0.3091132640838623,
      "learning_rate": 8.10591847840774e-06,
      "loss": 0.446,
      "step": 546
    },
    {
      "epoch": 1.0753604193971167,
      "grad_norm": 0.27796778082847595,
      "learning_rate": 8.096931644275318e-06,
      "loss": 0.435,
      "step": 547
    },
    {
      "epoch": 1.0773263433813893,
      "grad_norm": 0.3258281648159027,
      "learning_rate": 8.087928548805326e-06,
      "loss": 0.4569,
      "step": 548
    },
    {
      "epoch": 1.0792922673656618,
      "grad_norm": 0.29360657930374146,
      "learning_rate": 8.078909239271127e-06,
      "loss": 0.4372,
      "step": 549
    },
    {
      "epoch": 1.0812581913499344,
      "grad_norm": 0.34414565563201904,
      "learning_rate": 8.069873763031221e-06,
      "loss": 0.438,
      "step": 550
    },
    {
      "epoch": 1.083224115334207,
      "grad_norm": 0.2912955582141876,
      "learning_rate": 8.060822167528996e-06,
      "loss": 0.4519,
      "step": 551
    },
    {
      "epoch": 1.0851900393184797,
      "grad_norm": 0.27693864703178406,
      "learning_rate": 8.051754500292479e-06,
      "loss": 0.4423,
      "step": 552
    },
    {
      "epoch": 1.0871559633027523,
      "grad_norm": 0.29716360569000244,
      "learning_rate": 8.042670808934086e-06,
      "loss": 0.4736,
      "step": 553
    },
    {
      "epoch": 1.089121887287025,
      "grad_norm": 0.25404056906700134,
      "learning_rate": 8.033571141150371e-06,
      "loss": 0.4185,
      "step": 554
    },
    {
      "epoch": 1.0910878112712976,
      "grad_norm": 0.28415998816490173,
      "learning_rate": 8.024455544721778e-06,
      "loss": 0.459,
      "step": 555
    },
    {
      "epoch": 1.09305373525557,
      "grad_norm": 0.25493794679641724,
      "learning_rate": 8.015324067512393e-06,
      "loss": 0.4409,
      "step": 556
    },
    {
      "epoch": 1.0950196592398427,
      "grad_norm": 0.3459571599960327,
      "learning_rate": 8.006176757469681e-06,
      "loss": 0.4351,
      "step": 557
    },
    {
      "epoch": 1.0969855832241153,
      "grad_norm": 0.2602083683013916,
      "learning_rate": 7.997013662624246e-06,
      "loss": 0.4313,
      "step": 558
    },
    {
      "epoch": 1.098951507208388,
      "grad_norm": 0.3519684374332428,
      "learning_rate": 7.987834831089576e-06,
      "loss": 0.4401,
      "step": 559
    },
    {
      "epoch": 1.1009174311926606,
      "grad_norm": 0.2838681936264038,
      "learning_rate": 7.978640311061787e-06,
      "loss": 0.4671,
      "step": 560
    },
    {
      "epoch": 1.1028833551769333,
      "grad_norm": 0.31308382749557495,
      "learning_rate": 7.969430150819372e-06,
      "loss": 0.4381,
      "step": 561
    },
    {
      "epoch": 1.1048492791612057,
      "grad_norm": 0.2971574664115906,
      "learning_rate": 7.960204398722947e-06,
      "loss": 0.4509,
      "step": 562
    },
    {
      "epoch": 1.1068152031454783,
      "grad_norm": 0.2653731107711792,
      "learning_rate": 7.950963103215e-06,
      "loss": 0.4312,
      "step": 563
    },
    {
      "epoch": 1.108781127129751,
      "grad_norm": 0.2692069113254547,
      "learning_rate": 7.941706312819632e-06,
      "loss": 0.4427,
      "step": 564
    },
    {
      "epoch": 1.1107470511140236,
      "grad_norm": 0.29672127962112427,
      "learning_rate": 7.932434076142304e-06,
      "loss": 0.4527,
      "step": 565
    },
    {
      "epoch": 1.1127129750982963,
      "grad_norm": 0.2934640944004059,
      "learning_rate": 7.923146441869585e-06,
      "loss": 0.4736,
      "step": 566
    },
    {
      "epoch": 1.114678899082569,
      "grad_norm": 0.2855452597141266,
      "learning_rate": 7.913843458768892e-06,
      "loss": 0.4476,
      "step": 567
    },
    {
      "epoch": 1.1166448230668413,
      "grad_norm": 0.2716505825519562,
      "learning_rate": 7.904525175688234e-06,
      "loss": 0.4414,
      "step": 568
    },
    {
      "epoch": 1.118610747051114,
      "grad_norm": 0.2886597812175751,
      "learning_rate": 7.895191641555957e-06,
      "loss": 0.4302,
      "step": 569
    },
    {
      "epoch": 1.1205766710353866,
      "grad_norm": 0.31945592164993286,
      "learning_rate": 7.88584290538049e-06,
      "loss": 0.4672,
      "step": 570
    },
    {
      "epoch": 1.1225425950196593,
      "grad_norm": 0.2809208929538727,
      "learning_rate": 7.876479016250082e-06,
      "loss": 0.4548,
      "step": 571
    },
    {
      "epoch": 1.124508519003932,
      "grad_norm": 0.3163772523403168,
      "learning_rate": 7.86710002333255e-06,
      "loss": 0.4428,
      "step": 572
    },
    {
      "epoch": 1.1264744429882045,
      "grad_norm": 0.3288361728191376,
      "learning_rate": 7.857705975875015e-06,
      "loss": 0.465,
      "step": 573
    },
    {
      "epoch": 1.1284403669724772,
      "grad_norm": 0.28767189383506775,
      "learning_rate": 7.848296923203645e-06,
      "loss": 0.441,
      "step": 574
    },
    {
      "epoch": 1.1304062909567496,
      "grad_norm": 0.34927651286125183,
      "learning_rate": 7.838872914723403e-06,
      "loss": 0.4482,
      "step": 575
    },
    {
      "epoch": 1.1323722149410222,
      "grad_norm": 0.2668762505054474,
      "learning_rate": 7.829433999917773e-06,
      "loss": 0.4424,
      "step": 576
    },
    {
      "epoch": 1.134338138925295,
      "grad_norm": 0.3305877447128296,
      "learning_rate": 7.819980228348519e-06,
      "loss": 0.4607,
      "step": 577
    },
    {
      "epoch": 1.1363040629095675,
      "grad_norm": 0.2858606278896332,
      "learning_rate": 7.810511649655406e-06,
      "loss": 0.4625,
      "step": 578
    },
    {
      "epoch": 1.1382699868938402,
      "grad_norm": 0.2797732353210449,
      "learning_rate": 7.801028313555954e-06,
      "loss": 0.4558,
      "step": 579
    },
    {
      "epoch": 1.1402359108781126,
      "grad_norm": 0.33720502257347107,
      "learning_rate": 7.79153026984517e-06,
      "loss": 0.4454,
      "step": 580
    },
    {
      "epoch": 1.1422018348623852,
      "grad_norm": 0.2824835777282715,
      "learning_rate": 7.782017568395284e-06,
      "loss": 0.4509,
      "step": 581
    },
    {
      "epoch": 1.144167758846658,
      "grad_norm": 0.2600245475769043,
      "learning_rate": 7.772490259155493e-06,
      "loss": 0.4316,
      "step": 582
    },
    {
      "epoch": 1.1461336828309305,
      "grad_norm": 0.3566450774669647,
      "learning_rate": 7.762948392151699e-06,
      "loss": 0.4615,
      "step": 583
    },
    {
      "epoch": 1.1480996068152032,
      "grad_norm": 0.2671070992946625,
      "learning_rate": 7.75339201748624e-06,
      "loss": 0.445,
      "step": 584
    },
    {
      "epoch": 1.1500655307994758,
      "grad_norm": 0.29686596989631653,
      "learning_rate": 7.743821185337634e-06,
      "loss": 0.4493,
      "step": 585
    },
    {
      "epoch": 1.1520314547837485,
      "grad_norm": 0.2697345018386841,
      "learning_rate": 7.734235945960307e-06,
      "loss": 0.4495,
      "step": 586
    },
    {
      "epoch": 1.1539973787680209,
      "grad_norm": 0.3056127727031708,
      "learning_rate": 7.724636349684341e-06,
      "loss": 0.4455,
      "step": 587
    },
    {
      "epoch": 1.1559633027522935,
      "grad_norm": 0.3014778792858124,
      "learning_rate": 7.715022446915195e-06,
      "loss": 0.4435,
      "step": 588
    },
    {
      "epoch": 1.1579292267365662,
      "grad_norm": 0.2972736060619354,
      "learning_rate": 7.705394288133459e-06,
      "loss": 0.4624,
      "step": 589
    },
    {
      "epoch": 1.1598951507208388,
      "grad_norm": 0.326951265335083,
      "learning_rate": 7.695751923894568e-06,
      "loss": 0.4377,
      "step": 590
    },
    {
      "epoch": 1.1618610747051115,
      "grad_norm": 0.3335776925086975,
      "learning_rate": 7.686095404828552e-06,
      "loss": 0.4514,
      "step": 591
    },
    {
      "epoch": 1.163826998689384,
      "grad_norm": 0.3000209331512451,
      "learning_rate": 7.676424781639768e-06,
      "loss": 0.4668,
      "step": 592
    },
    {
      "epoch": 1.1657929226736565,
      "grad_norm": 0.3267945945262909,
      "learning_rate": 7.666740105106623e-06,
      "loss": 0.4515,
      "step": 593
    },
    {
      "epoch": 1.1677588466579292,
      "grad_norm": 0.2960260212421417,
      "learning_rate": 7.65704142608132e-06,
      "loss": 0.4344,
      "step": 594
    },
    {
      "epoch": 1.1697247706422018,
      "grad_norm": 0.27537232637405396,
      "learning_rate": 7.647328795489585e-06,
      "loss": 0.4644,
      "step": 595
    },
    {
      "epoch": 1.1716906946264745,
      "grad_norm": 0.3565449118614197,
      "learning_rate": 7.637602264330406e-06,
      "loss": 0.4714,
      "step": 596
    },
    {
      "epoch": 1.173656618610747,
      "grad_norm": 0.2900427281856537,
      "learning_rate": 7.627861883675748e-06,
      "loss": 0.4498,
      "step": 597
    },
    {
      "epoch": 1.1756225425950197,
      "grad_norm": 0.31862181425094604,
      "learning_rate": 7.618107704670308e-06,
      "loss": 0.4442,
      "step": 598
    },
    {
      "epoch": 1.1775884665792922,
      "grad_norm": 0.28148674964904785,
      "learning_rate": 7.60833977853123e-06,
      "loss": 0.441,
      "step": 599
    },
    {
      "epoch": 1.1795543905635648,
      "grad_norm": 0.35680100321769714,
      "learning_rate": 7.598558156547842e-06,
      "loss": 0.4632,
      "step": 600
    },
    {
      "epoch": 1.1815203145478375,
      "grad_norm": 0.2732924222946167,
      "learning_rate": 7.588762890081387e-06,
      "loss": 0.4565,
      "step": 601
    },
    {
      "epoch": 1.18348623853211,
      "grad_norm": 0.3304100036621094,
      "learning_rate": 7.578954030564754e-06,
      "loss": 0.4511,
      "step": 602
    },
    {
      "epoch": 1.1854521625163827,
      "grad_norm": 0.2812047004699707,
      "learning_rate": 7.569131629502201e-06,
      "loss": 0.4534,
      "step": 603
    },
    {
      "epoch": 1.1874180865006554,
      "grad_norm": 0.2930050194263458,
      "learning_rate": 7.559295738469094e-06,
      "loss": 0.4394,
      "step": 604
    },
    {
      "epoch": 1.189384010484928,
      "grad_norm": 0.26927125453948975,
      "learning_rate": 7.549446409111629e-06,
      "loss": 0.4413,
      "step": 605
    },
    {
      "epoch": 1.1913499344692005,
      "grad_norm": 0.2692791521549225,
      "learning_rate": 7.53958369314657e-06,
      "loss": 0.4618,
      "step": 606
    },
    {
      "epoch": 1.193315858453473,
      "grad_norm": 0.26487746834754944,
      "learning_rate": 7.529707642360963e-06,
      "loss": 0.462,
      "step": 607
    },
    {
      "epoch": 1.1952817824377457,
      "grad_norm": 0.2600611448287964,
      "learning_rate": 7.519818308611877e-06,
      "loss": 0.4454,
      "step": 608
    },
    {
      "epoch": 1.1972477064220184,
      "grad_norm": 0.3271210789680481,
      "learning_rate": 7.509915743826128e-06,
      "loss": 0.4537,
      "step": 609
    },
    {
      "epoch": 1.199213630406291,
      "grad_norm": 0.27632102370262146,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.4431,
      "step": 610
    },
    {
      "epoch": 1.2011795543905635,
      "grad_norm": 0.2809217870235443,
      "learning_rate": 7.490071129198986e-06,
      "loss": 0.4661,
      "step": 611
    },
    {
      "epoch": 1.203145478374836,
      "grad_norm": 0.28529879450798035,
      "learning_rate": 7.480129183557499e-06,
      "loss": 0.4543,
      "step": 612
    },
    {
      "epoch": 1.2051114023591087,
      "grad_norm": 0.2784067690372467,
      "learning_rate": 7.470174215278605e-06,
      "loss": 0.4557,
      "step": 613
    },
    {
      "epoch": 1.2070773263433814,
      "grad_norm": 0.29982420802116394,
      "learning_rate": 7.460206276633754e-06,
      "loss": 0.4604,
      "step": 614
    },
    {
      "epoch": 1.209043250327654,
      "grad_norm": 0.30934128165245056,
      "learning_rate": 7.450225419962498e-06,
      "loss": 0.4414,
      "step": 615
    },
    {
      "epoch": 1.2110091743119267,
      "grad_norm": 0.31483328342437744,
      "learning_rate": 7.440231697672218e-06,
      "loss": 0.4498,
      "step": 616
    },
    {
      "epoch": 1.2129750982961993,
      "grad_norm": 0.33800315856933594,
      "learning_rate": 7.430225162237852e-06,
      "loss": 0.4544,
      "step": 617
    },
    {
      "epoch": 1.2149410222804717,
      "grad_norm": 0.2889743745326996,
      "learning_rate": 7.4202058662016155e-06,
      "loss": 0.4515,
      "step": 618
    },
    {
      "epoch": 1.2169069462647444,
      "grad_norm": 0.300650954246521,
      "learning_rate": 7.4101738621727245e-06,
      "loss": 0.456,
      "step": 619
    },
    {
      "epoch": 1.218872870249017,
      "grad_norm": 0.2623285949230194,
      "learning_rate": 7.400129202827129e-06,
      "loss": 0.4351,
      "step": 620
    },
    {
      "epoch": 1.2208387942332897,
      "grad_norm": 0.29180240631103516,
      "learning_rate": 7.390071940907222e-06,
      "loss": 0.4321,
      "step": 621
    },
    {
      "epoch": 1.2228047182175623,
      "grad_norm": 0.2687191367149353,
      "learning_rate": 7.380002129221576e-06,
      "loss": 0.4416,
      "step": 622
    },
    {
      "epoch": 1.224770642201835,
      "grad_norm": 0.31968826055526733,
      "learning_rate": 7.369919820644656e-06,
      "loss": 0.4496,
      "step": 623
    },
    {
      "epoch": 1.2267365661861074,
      "grad_norm": 0.2760871946811676,
      "learning_rate": 7.3598250681165485e-06,
      "loss": 0.4561,
      "step": 624
    },
    {
      "epoch": 1.22870249017038,
      "grad_norm": 0.3169286549091339,
      "learning_rate": 7.3497179246426764e-06,
      "loss": 0.449,
      "step": 625
    },
    {
      "epoch": 1.2306684141546527,
      "grad_norm": 0.2949203848838806,
      "learning_rate": 7.339598443293526e-06,
      "loss": 0.4466,
      "step": 626
    },
    {
      "epoch": 1.2326343381389253,
      "grad_norm": 0.36728623509407043,
      "learning_rate": 7.329466677204371e-06,
      "loss": 0.4604,
      "step": 627
    },
    {
      "epoch": 1.234600262123198,
      "grad_norm": 0.2636052668094635,
      "learning_rate": 7.319322679574985e-06,
      "loss": 0.4538,
      "step": 628
    },
    {
      "epoch": 1.2365661861074706,
      "grad_norm": 0.3246074616909027,
      "learning_rate": 7.3091665036693716e-06,
      "loss": 0.4745,
      "step": 629
    },
    {
      "epoch": 1.238532110091743,
      "grad_norm": 0.29816025495529175,
      "learning_rate": 7.298998202815474e-06,
      "loss": 0.4579,
      "step": 630
    },
    {
      "epoch": 1.2404980340760157,
      "grad_norm": 0.30991989374160767,
      "learning_rate": 7.288817830404906e-06,
      "loss": 0.4436,
      "step": 631
    },
    {
      "epoch": 1.2424639580602883,
      "grad_norm": 0.31033027172088623,
      "learning_rate": 7.278625439892665e-06,
      "loss": 0.4639,
      "step": 632
    },
    {
      "epoch": 1.244429882044561,
      "grad_norm": 0.28393280506134033,
      "learning_rate": 7.268421084796852e-06,
      "loss": 0.449,
      "step": 633
    },
    {
      "epoch": 1.2463958060288336,
      "grad_norm": 0.305463045835495,
      "learning_rate": 7.258204818698394e-06,
      "loss": 0.4489,
      "step": 634
    },
    {
      "epoch": 1.2483617300131062,
      "grad_norm": 0.29833632707595825,
      "learning_rate": 7.247976695240754e-06,
      "loss": 0.4381,
      "step": 635
    },
    {
      "epoch": 1.2503276539973789,
      "grad_norm": 0.2623004913330078,
      "learning_rate": 7.237736768129663e-06,
      "loss": 0.4384,
      "step": 636
    },
    {
      "epoch": 1.2522935779816513,
      "grad_norm": 0.3139297664165497,
      "learning_rate": 7.2274850911328275e-06,
      "loss": 0.451,
      "step": 637
    },
    {
      "epoch": 1.254259501965924,
      "grad_norm": 0.26940205693244934,
      "learning_rate": 7.21722171807965e-06,
      "loss": 0.4595,
      "step": 638
    },
    {
      "epoch": 1.2562254259501966,
      "grad_norm": 0.2619832754135132,
      "learning_rate": 7.206946702860948e-06,
      "loss": 0.4436,
      "step": 639
    },
    {
      "epoch": 1.2581913499344692,
      "grad_norm": 0.25589093565940857,
      "learning_rate": 7.196660099428665e-06,
      "loss": 0.4587,
      "step": 640
    },
    {
      "epoch": 1.2601572739187419,
      "grad_norm": 0.27356746792793274,
      "learning_rate": 7.186361961795596e-06,
      "loss": 0.4408,
      "step": 641
    },
    {
      "epoch": 1.2621231979030143,
      "grad_norm": 0.26509717106819153,
      "learning_rate": 7.176052344035101e-06,
      "loss": 0.4462,
      "step": 642
    },
    {
      "epoch": 1.264089121887287,
      "grad_norm": 0.32008934020996094,
      "learning_rate": 7.165731300280814e-06,
      "loss": 0.4422,
      "step": 643
    },
    {
      "epoch": 1.2660550458715596,
      "grad_norm": 0.2853245437145233,
      "learning_rate": 7.15539888472637e-06,
      "loss": 0.4586,
      "step": 644
    },
    {
      "epoch": 1.2680209698558322,
      "grad_norm": 0.32988548278808594,
      "learning_rate": 7.145055151625113e-06,
      "loss": 0.4414,
      "step": 645
    },
    {
      "epoch": 1.2699868938401049,
      "grad_norm": 0.3138856589794159,
      "learning_rate": 7.134700155289812e-06,
      "loss": 0.4303,
      "step": 646
    },
    {
      "epoch": 1.2719528178243775,
      "grad_norm": 0.2495591640472412,
      "learning_rate": 7.124333950092377e-06,
      "loss": 0.4444,
      "step": 647
    },
    {
      "epoch": 1.2739187418086502,
      "grad_norm": 0.4393638074398041,
      "learning_rate": 7.1139565904635755e-06,
      "loss": 0.4488,
      "step": 648
    },
    {
      "epoch": 1.2758846657929226,
      "grad_norm": 0.2642851769924164,
      "learning_rate": 7.103568130892742e-06,
      "loss": 0.4303,
      "step": 649
    },
    {
      "epoch": 1.2778505897771952,
      "grad_norm": 0.29225823283195496,
      "learning_rate": 7.093168625927497e-06,
      "loss": 0.4478,
      "step": 650
    },
    {
      "epoch": 1.2798165137614679,
      "grad_norm": 0.3072997033596039,
      "learning_rate": 7.082758130173456e-06,
      "loss": 0.4404,
      "step": 651
    },
    {
      "epoch": 1.2817824377457405,
      "grad_norm": 0.28156915307044983,
      "learning_rate": 7.072336698293946e-06,
      "loss": 0.4576,
      "step": 652
    },
    {
      "epoch": 1.2837483617300132,
      "grad_norm": 0.34902632236480713,
      "learning_rate": 7.0619043850097165e-06,
      "loss": 0.4664,
      "step": 653
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 0.2630963623523712,
      "learning_rate": 7.051461245098654e-06,
      "loss": 0.4455,
      "step": 654
    },
    {
      "epoch": 1.2876802096985585,
      "grad_norm": 0.2813081741333008,
      "learning_rate": 7.041007333395491e-06,
      "loss": 0.4423,
      "step": 655
    },
    {
      "epoch": 1.2896461336828309,
      "grad_norm": 0.27456530928611755,
      "learning_rate": 7.030542704791522e-06,
      "loss": 0.4482,
      "step": 656
    },
    {
      "epoch": 1.2916120576671035,
      "grad_norm": 0.2905554473400116,
      "learning_rate": 7.020067414234315e-06,
      "loss": 0.4399,
      "step": 657
    },
    {
      "epoch": 1.2935779816513762,
      "grad_norm": 0.27594971656799316,
      "learning_rate": 7.009581516727422e-06,
      "loss": 0.4563,
      "step": 658
    },
    {
      "epoch": 1.2955439056356488,
      "grad_norm": 0.2840750217437744,
      "learning_rate": 6.999085067330085e-06,
      "loss": 0.4711,
      "step": 659
    },
    {
      "epoch": 1.2975098296199215,
      "grad_norm": 0.3185199499130249,
      "learning_rate": 6.988578121156956e-06,
      "loss": 0.4557,
      "step": 660
    },
    {
      "epoch": 1.2994757536041939,
      "grad_norm": 0.2815924286842346,
      "learning_rate": 6.978060733377805e-06,
      "loss": 0.4383,
      "step": 661
    },
    {
      "epoch": 1.3014416775884665,
      "grad_norm": 0.28519296646118164,
      "learning_rate": 6.967532959217223e-06,
      "loss": 0.4462,
      "step": 662
    },
    {
      "epoch": 1.3034076015727392,
      "grad_norm": 0.27675876021385193,
      "learning_rate": 6.956994853954342e-06,
      "loss": 0.4386,
      "step": 663
    },
    {
      "epoch": 1.3053735255570118,
      "grad_norm": 0.2669420540332794,
      "learning_rate": 6.946446472922539e-06,
      "loss": 0.4316,
      "step": 664
    },
    {
      "epoch": 1.3073394495412844,
      "grad_norm": 0.2995237708091736,
      "learning_rate": 6.935887871509147e-06,
      "loss": 0.4512,
      "step": 665
    },
    {
      "epoch": 1.309305373525557,
      "grad_norm": 0.28722241520881653,
      "learning_rate": 6.925319105155165e-06,
      "loss": 0.4626,
      "step": 666
    },
    {
      "epoch": 1.3112712975098297,
      "grad_norm": 0.27150627970695496,
      "learning_rate": 6.914740229354964e-06,
      "loss": 0.4338,
      "step": 667
    },
    {
      "epoch": 1.3132372214941022,
      "grad_norm": 0.29424235224723816,
      "learning_rate": 6.904151299656e-06,
      "loss": 0.4516,
      "step": 668
    },
    {
      "epoch": 1.3152031454783748,
      "grad_norm": 2.394724130630493,
      "learning_rate": 6.8935523716585195e-06,
      "loss": 0.4346,
      "step": 669
    },
    {
      "epoch": 1.3171690694626474,
      "grad_norm": 0.3479631841182709,
      "learning_rate": 6.882943501015264e-06,
      "loss": 0.4519,
      "step": 670
    },
    {
      "epoch": 1.31913499344692,
      "grad_norm": 0.2685585916042328,
      "learning_rate": 6.872324743431189e-06,
      "loss": 0.4346,
      "step": 671
    },
    {
      "epoch": 1.3211009174311927,
      "grad_norm": 0.3368861675262451,
      "learning_rate": 6.8616961546631575e-06,
      "loss": 0.4434,
      "step": 672
    },
    {
      "epoch": 1.3230668414154652,
      "grad_norm": 0.2756919264793396,
      "learning_rate": 6.851057790519656e-06,
      "loss": 0.4499,
      "step": 673
    },
    {
      "epoch": 1.3250327653997378,
      "grad_norm": 0.4107244312763214,
      "learning_rate": 6.840409706860502e-06,
      "loss": 0.4401,
      "step": 674
    },
    {
      "epoch": 1.3269986893840104,
      "grad_norm": 0.2796277701854706,
      "learning_rate": 6.829751959596544e-06,
      "loss": 0.4295,
      "step": 675
    },
    {
      "epoch": 1.328964613368283,
      "grad_norm": 0.37067267298698425,
      "learning_rate": 6.819084604689379e-06,
      "loss": 0.4526,
      "step": 676
    },
    {
      "epoch": 1.3309305373525557,
      "grad_norm": 0.30490928888320923,
      "learning_rate": 6.808407698151041e-06,
      "loss": 0.4392,
      "step": 677
    },
    {
      "epoch": 1.3328964613368284,
      "grad_norm": 0.30019262433052063,
      "learning_rate": 6.797721296043727e-06,
      "loss": 0.441,
      "step": 678
    },
    {
      "epoch": 1.334862385321101,
      "grad_norm": 0.26304328441619873,
      "learning_rate": 6.787025454479489e-06,
      "loss": 0.4261,
      "step": 679
    },
    {
      "epoch": 1.3368283093053734,
      "grad_norm": 0.3093208372592926,
      "learning_rate": 6.776320229619944e-06,
      "loss": 0.4444,
      "step": 680
    },
    {
      "epoch": 1.338794233289646,
      "grad_norm": 0.2719149887561798,
      "learning_rate": 6.765605677675982e-06,
      "loss": 0.4339,
      "step": 681
    },
    {
      "epoch": 1.3407601572739187,
      "grad_norm": 0.2760200798511505,
      "learning_rate": 6.754881854907461e-06,
      "loss": 0.4657,
      "step": 682
    },
    {
      "epoch": 1.3427260812581914,
      "grad_norm": 0.29559653997421265,
      "learning_rate": 6.744148817622924e-06,
      "loss": 0.4407,
      "step": 683
    },
    {
      "epoch": 1.344692005242464,
      "grad_norm": 0.280727356672287,
      "learning_rate": 6.733406622179295e-06,
      "loss": 0.4341,
      "step": 684
    },
    {
      "epoch": 1.3466579292267364,
      "grad_norm": 0.3014304041862488,
      "learning_rate": 6.722655324981584e-06,
      "loss": 0.4497,
      "step": 685
    },
    {
      "epoch": 1.3486238532110093,
      "grad_norm": 0.27196088433265686,
      "learning_rate": 6.711894982482598e-06,
      "loss": 0.4312,
      "step": 686
    },
    {
      "epoch": 1.3505897771952817,
      "grad_norm": 0.33416539430618286,
      "learning_rate": 6.701125651182631e-06,
      "loss": 0.4446,
      "step": 687
    },
    {
      "epoch": 1.3525557011795544,
      "grad_norm": 0.29358693957328796,
      "learning_rate": 6.690347387629184e-06,
      "loss": 0.4526,
      "step": 688
    },
    {
      "epoch": 1.354521625163827,
      "grad_norm": 0.28325355052948,
      "learning_rate": 6.679560248416652e-06,
      "loss": 0.4241,
      "step": 689
    },
    {
      "epoch": 1.3564875491480997,
      "grad_norm": 0.29557231068611145,
      "learning_rate": 6.668764290186039e-06,
      "loss": 0.4447,
      "step": 690
    },
    {
      "epoch": 1.3584534731323723,
      "grad_norm": 0.2679084539413452,
      "learning_rate": 6.6579595696246536e-06,
      "loss": 0.4284,
      "step": 691
    },
    {
      "epoch": 1.3604193971166447,
      "grad_norm": 0.29322460293769836,
      "learning_rate": 6.6471461434658135e-06,
      "loss": 0.4394,
      "step": 692
    },
    {
      "epoch": 1.3623853211009174,
      "grad_norm": 0.2770942449569702,
      "learning_rate": 6.6363240684885465e-06,
      "loss": 0.4435,
      "step": 693
    },
    {
      "epoch": 1.36435124508519,
      "grad_norm": 0.3174556493759155,
      "learning_rate": 6.625493401517299e-06,
      "loss": 0.4484,
      "step": 694
    },
    {
      "epoch": 1.3663171690694627,
      "grad_norm": 0.26972174644470215,
      "learning_rate": 6.614654199421625e-06,
      "loss": 0.4628,
      "step": 695
    },
    {
      "epoch": 1.3682830930537353,
      "grad_norm": 0.36254197359085083,
      "learning_rate": 6.603806519115899e-06,
      "loss": 0.4552,
      "step": 696
    },
    {
      "epoch": 1.370249017038008,
      "grad_norm": 0.30667033791542053,
      "learning_rate": 6.592950417559013e-06,
      "loss": 0.4467,
      "step": 697
    },
    {
      "epoch": 1.3722149410222806,
      "grad_norm": 0.29005348682403564,
      "learning_rate": 6.582085951754076e-06,
      "loss": 0.4488,
      "step": 698
    },
    {
      "epoch": 1.374180865006553,
      "grad_norm": 0.32942113280296326,
      "learning_rate": 6.571213178748112e-06,
      "loss": 0.4491,
      "step": 699
    },
    {
      "epoch": 1.3761467889908257,
      "grad_norm": 0.26501721143722534,
      "learning_rate": 6.560332155631774e-06,
      "loss": 0.4455,
      "step": 700
    },
    {
      "epoch": 1.3781127129750983,
      "grad_norm": 0.31284385919570923,
      "learning_rate": 6.549442939539026e-06,
      "loss": 0.4534,
      "step": 701
    },
    {
      "epoch": 1.380078636959371,
      "grad_norm": 0.2804517149925232,
      "learning_rate": 6.538545587646854e-06,
      "loss": 0.446,
      "step": 702
    },
    {
      "epoch": 1.3820445609436436,
      "grad_norm": 0.27337881922721863,
      "learning_rate": 6.527640157174964e-06,
      "loss": 0.4472,
      "step": 703
    },
    {
      "epoch": 1.384010484927916,
      "grad_norm": 0.30470162630081177,
      "learning_rate": 6.516726705385482e-06,
      "loss": 0.4462,
      "step": 704
    },
    {
      "epoch": 1.3859764089121887,
      "grad_norm": 0.2848241627216339,
      "learning_rate": 6.50580528958265e-06,
      "loss": 0.442,
      "step": 705
    },
    {
      "epoch": 1.3879423328964613,
      "grad_norm": 0.27773287892341614,
      "learning_rate": 6.494875967112529e-06,
      "loss": 0.4476,
      "step": 706
    },
    {
      "epoch": 1.389908256880734,
      "grad_norm": 0.28546908497810364,
      "learning_rate": 6.483938795362695e-06,
      "loss": 0.4259,
      "step": 707
    },
    {
      "epoch": 1.3918741808650066,
      "grad_norm": 0.25834253430366516,
      "learning_rate": 6.47299383176194e-06,
      "loss": 0.4365,
      "step": 708
    },
    {
      "epoch": 1.3938401048492792,
      "grad_norm": 0.26199427247047424,
      "learning_rate": 6.462041133779969e-06,
      "loss": 0.4295,
      "step": 709
    },
    {
      "epoch": 1.3958060288335519,
      "grad_norm": 0.26836487650871277,
      "learning_rate": 6.451080758927099e-06,
      "loss": 0.4316,
      "step": 710
    },
    {
      "epoch": 1.3977719528178243,
      "grad_norm": 0.2869493067264557,
      "learning_rate": 6.440112764753956e-06,
      "loss": 0.425,
      "step": 711
    },
    {
      "epoch": 1.399737876802097,
      "grad_norm": 0.27614301443099976,
      "learning_rate": 6.429137208851172e-06,
      "loss": 0.4602,
      "step": 712
    },
    {
      "epoch": 1.4017038007863696,
      "grad_norm": 0.27254679799079895,
      "learning_rate": 6.418154148849087e-06,
      "loss": 0.4415,
      "step": 713
    },
    {
      "epoch": 1.4036697247706422,
      "grad_norm": 0.26700153946876526,
      "learning_rate": 6.4071636424174435e-06,
      "loss": 0.4358,
      "step": 714
    },
    {
      "epoch": 1.4056356487549149,
      "grad_norm": 0.23848773539066315,
      "learning_rate": 6.396165747265079e-06,
      "loss": 0.4416,
      "step": 715
    },
    {
      "epoch": 1.4076015727391873,
      "grad_norm": 0.24086840450763702,
      "learning_rate": 6.385160521139633e-06,
      "loss": 0.427,
      "step": 716
    },
    {
      "epoch": 1.4095674967234602,
      "grad_norm": 0.23305992782115936,
      "learning_rate": 6.374148021827237e-06,
      "loss": 0.4443,
      "step": 717
    },
    {
      "epoch": 1.4115334207077326,
      "grad_norm": 0.2489941418170929,
      "learning_rate": 6.36312830715221e-06,
      "loss": 0.4353,
      "step": 718
    },
    {
      "epoch": 1.4134993446920052,
      "grad_norm": 0.24544461071491241,
      "learning_rate": 6.352101434976761e-06,
      "loss": 0.4336,
      "step": 719
    },
    {
      "epoch": 1.4154652686762779,
      "grad_norm": 0.2898389399051666,
      "learning_rate": 6.341067463200678e-06,
      "loss": 0.4439,
      "step": 720
    },
    {
      "epoch": 1.4174311926605505,
      "grad_norm": 0.28088799118995667,
      "learning_rate": 6.3300264497610295e-06,
      "loss": 0.4526,
      "step": 721
    },
    {
      "epoch": 1.4193971166448232,
      "grad_norm": 0.3206160366535187,
      "learning_rate": 6.318978452631859e-06,
      "loss": 0.48,
      "step": 722
    },
    {
      "epoch": 1.4213630406290956,
      "grad_norm": 0.2654613256454468,
      "learning_rate": 6.307923529823876e-06,
      "loss": 0.4594,
      "step": 723
    },
    {
      "epoch": 1.4233289646133682,
      "grad_norm": 0.31457147002220154,
      "learning_rate": 6.296861739384162e-06,
      "loss": 0.4333,
      "step": 724
    },
    {
      "epoch": 1.4252948885976409,
      "grad_norm": 0.25586414337158203,
      "learning_rate": 6.285793139395853e-06,
      "loss": 0.4354,
      "step": 725
    },
    {
      "epoch": 1.4272608125819135,
      "grad_norm": 0.27597638964653015,
      "learning_rate": 6.2747177879778424e-06,
      "loss": 0.4483,
      "step": 726
    },
    {
      "epoch": 1.4292267365661862,
      "grad_norm": 0.23994868993759155,
      "learning_rate": 6.263635743284475e-06,
      "loss": 0.4504,
      "step": 727
    },
    {
      "epoch": 1.4311926605504588,
      "grad_norm": 0.25358685851097107,
      "learning_rate": 6.252547063505241e-06,
      "loss": 0.4522,
      "step": 728
    },
    {
      "epoch": 1.4331585845347314,
      "grad_norm": 0.28718921542167664,
      "learning_rate": 6.241451806864465e-06,
      "loss": 0.4689,
      "step": 729
    },
    {
      "epoch": 1.4351245085190039,
      "grad_norm": 0.2732388973236084,
      "learning_rate": 6.230350031621014e-06,
      "loss": 0.4385,
      "step": 730
    },
    {
      "epoch": 1.4370904325032765,
      "grad_norm": 0.29866504669189453,
      "learning_rate": 6.219241796067974e-06,
      "loss": 0.4456,
      "step": 731
    },
    {
      "epoch": 1.4390563564875491,
      "grad_norm": 0.308805376291275,
      "learning_rate": 6.208127158532358e-06,
      "loss": 0.4529,
      "step": 732
    },
    {
      "epoch": 1.4410222804718218,
      "grad_norm": 0.26815065741539,
      "learning_rate": 6.197006177374793e-06,
      "loss": 0.4323,
      "step": 733
    },
    {
      "epoch": 1.4429882044560944,
      "grad_norm": 0.30783241987228394,
      "learning_rate": 6.1858789109892145e-06,
      "loss": 0.4334,
      "step": 734
    },
    {
      "epoch": 1.4449541284403669,
      "grad_norm": 0.25694021582603455,
      "learning_rate": 6.174745417802563e-06,
      "loss": 0.4469,
      "step": 735
    },
    {
      "epoch": 1.4469200524246395,
      "grad_norm": 0.2983214855194092,
      "learning_rate": 6.163605756274472e-06,
      "loss": 0.4634,
      "step": 736
    },
    {
      "epoch": 1.4488859764089121,
      "grad_norm": 0.2631153166294098,
      "learning_rate": 6.1524599848969635e-06,
      "loss": 0.4518,
      "step": 737
    },
    {
      "epoch": 1.4508519003931848,
      "grad_norm": 0.30863478779792786,
      "learning_rate": 6.141308162194141e-06,
      "loss": 0.4842,
      "step": 738
    },
    {
      "epoch": 1.4528178243774574,
      "grad_norm": 0.2634303569793701,
      "learning_rate": 6.130150346721888e-06,
      "loss": 0.4375,
      "step": 739
    },
    {
      "epoch": 1.45478374836173,
      "grad_norm": 0.2764803469181061,
      "learning_rate": 6.118986597067543e-06,
      "loss": 0.4523,
      "step": 740
    },
    {
      "epoch": 1.4567496723460027,
      "grad_norm": 0.25771403312683105,
      "learning_rate": 6.1078169718496164e-06,
      "loss": 0.4404,
      "step": 741
    },
    {
      "epoch": 1.4587155963302751,
      "grad_norm": 0.28037071228027344,
      "learning_rate": 6.096641529717459e-06,
      "loss": 0.4302,
      "step": 742
    },
    {
      "epoch": 1.4606815203145478,
      "grad_norm": 0.2815488576889038,
      "learning_rate": 6.085460329350975e-06,
      "loss": 0.4523,
      "step": 743
    },
    {
      "epoch": 1.4626474442988204,
      "grad_norm": 0.24113839864730835,
      "learning_rate": 6.074273429460296e-06,
      "loss": 0.4368,
      "step": 744
    },
    {
      "epoch": 1.464613368283093,
      "grad_norm": 0.2681233286857605,
      "learning_rate": 6.063080888785484e-06,
      "loss": 0.4313,
      "step": 745
    },
    {
      "epoch": 1.4665792922673657,
      "grad_norm": 0.26223933696746826,
      "learning_rate": 6.051882766096219e-06,
      "loss": 0.427,
      "step": 746
    },
    {
      "epoch": 1.4685452162516381,
      "grad_norm": 0.23828788101673126,
      "learning_rate": 6.040679120191491e-06,
      "loss": 0.4509,
      "step": 747
    },
    {
      "epoch": 1.470511140235911,
      "grad_norm": 0.2911582589149475,
      "learning_rate": 6.0294700098992944e-06,
      "loss": 0.4623,
      "step": 748
    },
    {
      "epoch": 1.4724770642201834,
      "grad_norm": 0.22850066423416138,
      "learning_rate": 6.018255494076309e-06,
      "loss": 0.4229,
      "step": 749
    },
    {
      "epoch": 1.474442988204456,
      "grad_norm": 0.28099560737609863,
      "learning_rate": 6.007035631607605e-06,
      "loss": 0.4398,
      "step": 750
    },
    {
      "epoch": 1.4764089121887287,
      "grad_norm": 0.24817194044589996,
      "learning_rate": 5.995810481406319e-06,
      "loss": 0.4438,
      "step": 751
    },
    {
      "epoch": 1.4783748361730014,
      "grad_norm": 0.29838526248931885,
      "learning_rate": 5.984580102413361e-06,
      "loss": 0.4374,
      "step": 752
    },
    {
      "epoch": 1.480340760157274,
      "grad_norm": 0.28657519817352295,
      "learning_rate": 5.9733445535970915e-06,
      "loss": 0.451,
      "step": 753
    },
    {
      "epoch": 1.4823066841415464,
      "grad_norm": 0.2528459429740906,
      "learning_rate": 5.962103893953016e-06,
      "loss": 0.4176,
      "step": 754
    },
    {
      "epoch": 1.484272608125819,
      "grad_norm": 0.29090726375579834,
      "learning_rate": 5.950858182503478e-06,
      "loss": 0.4679,
      "step": 755
    },
    {
      "epoch": 1.4862385321100917,
      "grad_norm": 0.2765166759490967,
      "learning_rate": 5.939607478297347e-06,
      "loss": 0.4687,
      "step": 756
    },
    {
      "epoch": 1.4882044560943644,
      "grad_norm": 0.28713834285736084,
      "learning_rate": 5.928351840409707e-06,
      "loss": 0.45,
      "step": 757
    },
    {
      "epoch": 1.490170380078637,
      "grad_norm": 0.2559838891029358,
      "learning_rate": 5.917091327941548e-06,
      "loss": 0.4383,
      "step": 758
    },
    {
      "epoch": 1.4921363040629096,
      "grad_norm": 0.24240902066230774,
      "learning_rate": 5.905826000019458e-06,
      "loss": 0.4527,
      "step": 759
    },
    {
      "epoch": 1.4941022280471823,
      "grad_norm": 0.3005048930644989,
      "learning_rate": 5.8945559157953035e-06,
      "loss": 0.4577,
      "step": 760
    },
    {
      "epoch": 1.4960681520314547,
      "grad_norm": 0.30774039030075073,
      "learning_rate": 5.883281134445932e-06,
      "loss": 0.4625,
      "step": 761
    },
    {
      "epoch": 1.4980340760157274,
      "grad_norm": 0.24888800084590912,
      "learning_rate": 5.8720017151728526e-06,
      "loss": 0.4438,
      "step": 762
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.2986248731613159,
      "learning_rate": 5.8607177172019245e-06,
      "loss": 0.4405,
      "step": 763
    },
    {
      "epoch": 1.5019659239842726,
      "grad_norm": 0.2639208436012268,
      "learning_rate": 5.849429199783054e-06,
      "loss": 0.4518,
      "step": 764
    },
    {
      "epoch": 1.5039318479685453,
      "grad_norm": 0.2760374844074249,
      "learning_rate": 5.838136222189874e-06,
      "loss": 0.4393,
      "step": 765
    },
    {
      "epoch": 1.5058977719528177,
      "grad_norm": 0.27932795882225037,
      "learning_rate": 5.826838843719437e-06,
      "loss": 0.4402,
      "step": 766
    },
    {
      "epoch": 1.5078636959370906,
      "grad_norm": 0.24393486976623535,
      "learning_rate": 5.8155371236919045e-06,
      "loss": 0.4394,
      "step": 767
    },
    {
      "epoch": 1.509829619921363,
      "grad_norm": 0.2864012122154236,
      "learning_rate": 5.804231121450235e-06,
      "loss": 0.4457,
      "step": 768
    },
    {
      "epoch": 1.5117955439056356,
      "grad_norm": 0.2616628408432007,
      "learning_rate": 5.79292089635987e-06,
      "loss": 0.446,
      "step": 769
    },
    {
      "epoch": 1.5137614678899083,
      "grad_norm": 0.27714791893959045,
      "learning_rate": 5.781606507808428e-06,
      "loss": 0.4543,
      "step": 770
    },
    {
      "epoch": 1.5157273918741807,
      "grad_norm": 0.2534174621105194,
      "learning_rate": 5.770288015205385e-06,
      "loss": 0.4229,
      "step": 771
    },
    {
      "epoch": 1.5176933158584536,
      "grad_norm": 0.23894338309764862,
      "learning_rate": 5.758965477981771e-06,
      "loss": 0.4376,
      "step": 772
    },
    {
      "epoch": 1.519659239842726,
      "grad_norm": 0.2569407522678375,
      "learning_rate": 5.747638955589847e-06,
      "loss": 0.4344,
      "step": 773
    },
    {
      "epoch": 1.5216251638269986,
      "grad_norm": 0.23768967390060425,
      "learning_rate": 5.736308507502805e-06,
      "loss": 0.4303,
      "step": 774
    },
    {
      "epoch": 1.5235910878112713,
      "grad_norm": 0.2613021731376648,
      "learning_rate": 5.724974193214448e-06,
      "loss": 0.4352,
      "step": 775
    },
    {
      "epoch": 1.525557011795544,
      "grad_norm": 0.27572882175445557,
      "learning_rate": 5.713636072238879e-06,
      "loss": 0.4289,
      "step": 776
    },
    {
      "epoch": 1.5275229357798166,
      "grad_norm": 0.2537841498851776,
      "learning_rate": 5.702294204110191e-06,
      "loss": 0.4559,
      "step": 777
    },
    {
      "epoch": 1.529488859764089,
      "grad_norm": 0.256172776222229,
      "learning_rate": 5.6909486483821485e-06,
      "loss": 0.452,
      "step": 778
    },
    {
      "epoch": 1.5314547837483619,
      "grad_norm": 0.2669544517993927,
      "learning_rate": 5.679599464627885e-06,
      "loss": 0.4309,
      "step": 779
    },
    {
      "epoch": 1.5334207077326343,
      "grad_norm": 0.24457275867462158,
      "learning_rate": 5.668246712439579e-06,
      "loss": 0.4395,
      "step": 780
    },
    {
      "epoch": 1.535386631716907,
      "grad_norm": 0.23864710330963135,
      "learning_rate": 5.656890451428147e-06,
      "loss": 0.4672,
      "step": 781
    },
    {
      "epoch": 1.5373525557011796,
      "grad_norm": 0.27357181906700134,
      "learning_rate": 5.645530741222931e-06,
      "loss": 0.4625,
      "step": 782
    },
    {
      "epoch": 1.5393184796854522,
      "grad_norm": 0.24743252992630005,
      "learning_rate": 5.634167641471383e-06,
      "loss": 0.4583,
      "step": 783
    },
    {
      "epoch": 1.5412844036697249,
      "grad_norm": 0.2562152147293091,
      "learning_rate": 5.622801211838753e-06,
      "loss": 0.437,
      "step": 784
    },
    {
      "epoch": 1.5432503276539973,
      "grad_norm": 0.23854993283748627,
      "learning_rate": 5.611431512007776e-06,
      "loss": 0.4269,
      "step": 785
    },
    {
      "epoch": 1.5452162516382701,
      "grad_norm": 0.24666976928710938,
      "learning_rate": 5.600058601678357e-06,
      "loss": 0.4478,
      "step": 786
    },
    {
      "epoch": 1.5471821756225426,
      "grad_norm": 0.25134962797164917,
      "learning_rate": 5.588682540567261e-06,
      "loss": 0.4367,
      "step": 787
    },
    {
      "epoch": 1.5491480996068152,
      "grad_norm": 0.24130164086818695,
      "learning_rate": 5.577303388407793e-06,
      "loss": 0.45,
      "step": 788
    },
    {
      "epoch": 1.5511140235910879,
      "grad_norm": 0.32679837942123413,
      "learning_rate": 5.5659212049494915e-06,
      "loss": 0.4541,
      "step": 789
    },
    {
      "epoch": 1.5530799475753603,
      "grad_norm": 0.26636821031570435,
      "learning_rate": 5.554536049957813e-06,
      "loss": 0.4467,
      "step": 790
    },
    {
      "epoch": 1.5550458715596331,
      "grad_norm": 0.28566592931747437,
      "learning_rate": 5.543147983213811e-06,
      "loss": 0.4362,
      "step": 791
    },
    {
      "epoch": 1.5570117955439056,
      "grad_norm": 0.2524894177913666,
      "learning_rate": 5.531757064513837e-06,
      "loss": 0.4416,
      "step": 792
    },
    {
      "epoch": 1.5589777195281782,
      "grad_norm": 0.2857241928577423,
      "learning_rate": 5.520363353669208e-06,
      "loss": 0.4417,
      "step": 793
    },
    {
      "epoch": 1.5609436435124509,
      "grad_norm": 0.25668948888778687,
      "learning_rate": 5.50896691050591e-06,
      "loss": 0.4505,
      "step": 794
    },
    {
      "epoch": 1.5629095674967235,
      "grad_norm": 0.2940739393234253,
      "learning_rate": 5.4975677948642704e-06,
      "loss": 0.4559,
      "step": 795
    },
    {
      "epoch": 1.5648754914809961,
      "grad_norm": 0.2603815793991089,
      "learning_rate": 5.48616606659865e-06,
      "loss": 0.4436,
      "step": 796
    },
    {
      "epoch": 1.5668414154652686,
      "grad_norm": 0.25054648518562317,
      "learning_rate": 5.474761785577133e-06,
      "loss": 0.4318,
      "step": 797
    },
    {
      "epoch": 1.5688073394495414,
      "grad_norm": 0.2663193643093109,
      "learning_rate": 5.4633550116812e-06,
      "loss": 0.4403,
      "step": 798
    },
    {
      "epoch": 1.5707732634338138,
      "grad_norm": 0.24768784642219543,
      "learning_rate": 5.451945804805425e-06,
      "loss": 0.4464,
      "step": 799
    },
    {
      "epoch": 1.5727391874180865,
      "grad_norm": 0.2361469715833664,
      "learning_rate": 5.440534224857158e-06,
      "loss": 0.4482,
      "step": 800
    },
    {
      "epoch": 1.5747051114023591,
      "grad_norm": 0.26781949400901794,
      "learning_rate": 5.429120331756208e-06,
      "loss": 0.4413,
      "step": 801
    },
    {
      "epoch": 1.5766710353866316,
      "grad_norm": 0.28918296098709106,
      "learning_rate": 5.417704185434531e-06,
      "loss": 0.4341,
      "step": 802
    },
    {
      "epoch": 1.5786369593709044,
      "grad_norm": 0.24010907113552094,
      "learning_rate": 5.4062858458359135e-06,
      "loss": 0.4259,
      "step": 803
    },
    {
      "epoch": 1.5806028833551768,
      "grad_norm": 0.26394084095954895,
      "learning_rate": 5.394865372915656e-06,
      "loss": 0.4376,
      "step": 804
    },
    {
      "epoch": 1.5825688073394495,
      "grad_norm": 0.25336146354675293,
      "learning_rate": 5.383442826640266e-06,
      "loss": 0.4472,
      "step": 805
    },
    {
      "epoch": 1.5845347313237221,
      "grad_norm": 0.25079286098480225,
      "learning_rate": 5.3720182669871334e-06,
      "loss": 0.452,
      "step": 806
    },
    {
      "epoch": 1.5865006553079948,
      "grad_norm": 0.27424266934394836,
      "learning_rate": 5.360591753944221e-06,
      "loss": 0.4225,
      "step": 807
    },
    {
      "epoch": 1.5884665792922674,
      "grad_norm": 0.2350778877735138,
      "learning_rate": 5.349163347509748e-06,
      "loss": 0.4247,
      "step": 808
    },
    {
      "epoch": 1.5904325032765398,
      "grad_norm": 0.2994675934314728,
      "learning_rate": 5.337733107691879e-06,
      "loss": 0.4335,
      "step": 809
    },
    {
      "epoch": 1.5923984272608127,
      "grad_norm": 0.23439203202724457,
      "learning_rate": 5.3263010945083994e-06,
      "loss": 0.42,
      "step": 810
    },
    {
      "epoch": 1.5943643512450851,
      "grad_norm": 0.2544015049934387,
      "learning_rate": 5.314867367986409e-06,
      "loss": 0.4446,
      "step": 811
    },
    {
      "epoch": 1.5963302752293578,
      "grad_norm": 0.29329681396484375,
      "learning_rate": 5.303431988162008e-06,
      "loss": 0.4397,
      "step": 812
    },
    {
      "epoch": 1.5982961992136304,
      "grad_norm": 0.26393359899520874,
      "learning_rate": 5.291995015079969e-06,
      "loss": 0.4418,
      "step": 813
    },
    {
      "epoch": 1.600262123197903,
      "grad_norm": 0.22339020669460297,
      "learning_rate": 5.2805565087934396e-06,
      "loss": 0.413,
      "step": 814
    },
    {
      "epoch": 1.6022280471821757,
      "grad_norm": 0.25190436840057373,
      "learning_rate": 5.269116529363613e-06,
      "loss": 0.4291,
      "step": 815
    },
    {
      "epoch": 1.6041939711664481,
      "grad_norm": 0.22340822219848633,
      "learning_rate": 5.257675136859415e-06,
      "loss": 0.4377,
      "step": 816
    },
    {
      "epoch": 1.606159895150721,
      "grad_norm": 0.23750069737434387,
      "learning_rate": 5.246232391357198e-06,
      "loss": 0.4284,
      "step": 817
    },
    {
      "epoch": 1.6081258191349934,
      "grad_norm": 0.25304466485977173,
      "learning_rate": 5.234788352940413e-06,
      "loss": 0.4414,
      "step": 818
    },
    {
      "epoch": 1.610091743119266,
      "grad_norm": 0.23254382610321045,
      "learning_rate": 5.223343081699302e-06,
      "loss": 0.4508,
      "step": 819
    },
    {
      "epoch": 1.6120576671035387,
      "grad_norm": 0.25423750281333923,
      "learning_rate": 5.211896637730582e-06,
      "loss": 0.4543,
      "step": 820
    },
    {
      "epoch": 1.6140235910878111,
      "grad_norm": 0.25801584124565125,
      "learning_rate": 5.200449081137124e-06,
      "loss": 0.4447,
      "step": 821
    },
    {
      "epoch": 1.615989515072084,
      "grad_norm": 0.2248302847146988,
      "learning_rate": 5.189000472027645e-06,
      "loss": 0.4302,
      "step": 822
    },
    {
      "epoch": 1.6179554390563564,
      "grad_norm": 0.27316683530807495,
      "learning_rate": 5.177550870516386e-06,
      "loss": 0.4333,
      "step": 823
    },
    {
      "epoch": 1.619921363040629,
      "grad_norm": 0.24663394689559937,
      "learning_rate": 5.1661003367228e-06,
      "loss": 0.4538,
      "step": 824
    },
    {
      "epoch": 1.6218872870249017,
      "grad_norm": 0.26696285605430603,
      "learning_rate": 5.1546489307712345e-06,
      "loss": 0.4366,
      "step": 825
    },
    {
      "epoch": 1.6238532110091743,
      "grad_norm": 0.26279351115226746,
      "learning_rate": 5.1431967127906156e-06,
      "loss": 0.4383,
      "step": 826
    },
    {
      "epoch": 1.625819134993447,
      "grad_norm": 0.2355157434940338,
      "learning_rate": 5.131743742914136e-06,
      "loss": 0.4334,
      "step": 827
    },
    {
      "epoch": 1.6277850589777194,
      "grad_norm": 0.2610161006450653,
      "learning_rate": 5.1202900812789346e-06,
      "loss": 0.4309,
      "step": 828
    },
    {
      "epoch": 1.6297509829619923,
      "grad_norm": 0.24807049334049225,
      "learning_rate": 5.108835788025782e-06,
      "loss": 0.4339,
      "step": 829
    },
    {
      "epoch": 1.6317169069462647,
      "grad_norm": 0.2702607214450836,
      "learning_rate": 5.097380923298767e-06,
      "loss": 0.4312,
      "step": 830
    },
    {
      "epoch": 1.6336828309305373,
      "grad_norm": 0.2486424446105957,
      "learning_rate": 5.085925547244978e-06,
      "loss": 0.4435,
      "step": 831
    },
    {
      "epoch": 1.63564875491481,
      "grad_norm": 0.2502179741859436,
      "learning_rate": 5.07446972001419e-06,
      "loss": 0.4404,
      "step": 832
    },
    {
      "epoch": 1.6376146788990824,
      "grad_norm": 0.29890352487564087,
      "learning_rate": 5.063013501758544e-06,
      "loss": 0.4325,
      "step": 833
    },
    {
      "epoch": 1.6395806028833553,
      "grad_norm": 0.2597431540489197,
      "learning_rate": 5.051556952632235e-06,
      "loss": 0.4438,
      "step": 834
    },
    {
      "epoch": 1.6415465268676277,
      "grad_norm": 0.23143181204795837,
      "learning_rate": 5.040100132791197e-06,
      "loss": 0.4387,
      "step": 835
    },
    {
      "epoch": 1.6435124508519003,
      "grad_norm": 0.2477666586637497,
      "learning_rate": 5.028643102392785e-06,
      "loss": 0.4247,
      "step": 836
    },
    {
      "epoch": 1.645478374836173,
      "grad_norm": 0.23772212862968445,
      "learning_rate": 5.0171859215954575e-06,
      "loss": 0.4314,
      "step": 837
    },
    {
      "epoch": 1.6474442988204456,
      "grad_norm": 0.2282267063856125,
      "learning_rate": 5.005728650558467e-06,
      "loss": 0.4519,
      "step": 838
    },
    {
      "epoch": 1.6494102228047183,
      "grad_norm": 0.27698102593421936,
      "learning_rate": 4.994271349441534e-06,
      "loss": 0.4283,
      "step": 839
    },
    {
      "epoch": 1.6513761467889907,
      "grad_norm": 0.22621089220046997,
      "learning_rate": 4.982814078404543e-06,
      "loss": 0.4115,
      "step": 840
    },
    {
      "epoch": 1.6533420707732636,
      "grad_norm": 0.2617005705833435,
      "learning_rate": 4.971356897607216e-06,
      "loss": 0.4468,
      "step": 841
    },
    {
      "epoch": 1.655307994757536,
      "grad_norm": 0.25365275144577026,
      "learning_rate": 4.959899867208805e-06,
      "loss": 0.4469,
      "step": 842
    },
    {
      "epoch": 1.6572739187418086,
      "grad_norm": 0.21561695635318756,
      "learning_rate": 4.948443047367767e-06,
      "loss": 0.4206,
      "step": 843
    },
    {
      "epoch": 1.6592398427260813,
      "grad_norm": 0.2597047686576843,
      "learning_rate": 4.936986498241458e-06,
      "loss": 0.4226,
      "step": 844
    },
    {
      "epoch": 1.661205766710354,
      "grad_norm": 0.27081021666526794,
      "learning_rate": 4.9255302799858125e-06,
      "loss": 0.4531,
      "step": 845
    },
    {
      "epoch": 1.6631716906946266,
      "grad_norm": 0.24182185530662537,
      "learning_rate": 4.9140744527550225e-06,
      "loss": 0.4275,
      "step": 846
    },
    {
      "epoch": 1.665137614678899,
      "grad_norm": 0.2677760720252991,
      "learning_rate": 4.9026190767012345e-06,
      "loss": 0.4383,
      "step": 847
    },
    {
      "epoch": 1.6671035386631718,
      "grad_norm": 0.24200788140296936,
      "learning_rate": 4.891164211974218e-06,
      "loss": 0.4246,
      "step": 848
    },
    {
      "epoch": 1.6690694626474443,
      "grad_norm": 0.27642157673835754,
      "learning_rate": 4.879709918721067e-06,
      "loss": 0.4454,
      "step": 849
    },
    {
      "epoch": 1.671035386631717,
      "grad_norm": 0.2820422351360321,
      "learning_rate": 4.868256257085866e-06,
      "loss": 0.4439,
      "step": 850
    },
    {
      "epoch": 1.6730013106159896,
      "grad_norm": 0.23756973445415497,
      "learning_rate": 4.856803287209385e-06,
      "loss": 0.4303,
      "step": 851
    },
    {
      "epoch": 1.674967234600262,
      "grad_norm": 0.2752552926540375,
      "learning_rate": 4.845351069228767e-06,
      "loss": 0.4196,
      "step": 852
    },
    {
      "epoch": 1.6769331585845348,
      "grad_norm": 0.2415931224822998,
      "learning_rate": 4.8338996632772014e-06,
      "loss": 0.4377,
      "step": 853
    },
    {
      "epoch": 1.6788990825688073,
      "grad_norm": 0.2568812668323517,
      "learning_rate": 4.822449129483616e-06,
      "loss": 0.4459,
      "step": 854
    },
    {
      "epoch": 1.68086500655308,
      "grad_norm": 0.2614029049873352,
      "learning_rate": 4.8109995279723556e-06,
      "loss": 0.4542,
      "step": 855
    },
    {
      "epoch": 1.6828309305373526,
      "grad_norm": 0.24816203117370605,
      "learning_rate": 4.799550918862877e-06,
      "loss": 0.452,
      "step": 856
    },
    {
      "epoch": 1.6847968545216252,
      "grad_norm": 0.24617765843868256,
      "learning_rate": 4.788103362269418e-06,
      "loss": 0.4341,
      "step": 857
    },
    {
      "epoch": 1.6867627785058978,
      "grad_norm": 0.23530834913253784,
      "learning_rate": 4.776656918300699e-06,
      "loss": 0.4248,
      "step": 858
    },
    {
      "epoch": 1.6887287024901703,
      "grad_norm": 0.263662725687027,
      "learning_rate": 4.76521164705959e-06,
      "loss": 0.4456,
      "step": 859
    },
    {
      "epoch": 1.6906946264744431,
      "grad_norm": 0.24673894047737122,
      "learning_rate": 4.7537676086428035e-06,
      "loss": 0.4545,
      "step": 860
    },
    {
      "epoch": 1.6926605504587156,
      "grad_norm": 0.24264618754386902,
      "learning_rate": 4.742324863140587e-06,
      "loss": 0.4328,
      "step": 861
    },
    {
      "epoch": 1.6946264744429882,
      "grad_norm": 0.2592974305152893,
      "learning_rate": 4.73088347063639e-06,
      "loss": 0.4512,
      "step": 862
    },
    {
      "epoch": 1.6965923984272608,
      "grad_norm": 0.2544708549976349,
      "learning_rate": 4.719443491206562e-06,
      "loss": 0.4415,
      "step": 863
    },
    {
      "epoch": 1.6985583224115333,
      "grad_norm": 0.2565777897834778,
      "learning_rate": 4.70800498492003e-06,
      "loss": 0.434,
      "step": 864
    },
    {
      "epoch": 1.7005242463958061,
      "grad_norm": 0.24272798001766205,
      "learning_rate": 4.696568011837994e-06,
      "loss": 0.4404,
      "step": 865
    },
    {
      "epoch": 1.7024901703800785,
      "grad_norm": 0.2778138816356659,
      "learning_rate": 4.685132632013592e-06,
      "loss": 0.4377,
      "step": 866
    },
    {
      "epoch": 1.7044560943643512,
      "grad_norm": 0.24986910820007324,
      "learning_rate": 4.673698905491602e-06,
      "loss": 0.4316,
      "step": 867
    },
    {
      "epoch": 1.7064220183486238,
      "grad_norm": 0.2514886260032654,
      "learning_rate": 4.6622668923081235e-06,
      "loss": 0.4352,
      "step": 868
    },
    {
      "epoch": 1.7083879423328965,
      "grad_norm": 0.24508267641067505,
      "learning_rate": 4.6508366524902525e-06,
      "loss": 0.4434,
      "step": 869
    },
    {
      "epoch": 1.7103538663171691,
      "grad_norm": 0.2508834898471832,
      "learning_rate": 4.639408246055781e-06,
      "loss": 0.4242,
      "step": 870
    },
    {
      "epoch": 1.7123197903014415,
      "grad_norm": 0.2465783655643463,
      "learning_rate": 4.627981733012868e-06,
      "loss": 0.4348,
      "step": 871
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 0.25469180941581726,
      "learning_rate": 4.616557173359736e-06,
      "loss": 0.4439,
      "step": 872
    },
    {
      "epoch": 1.7162516382699868,
      "grad_norm": 0.2471664994955063,
      "learning_rate": 4.605134627084345e-06,
      "loss": 0.4394,
      "step": 873
    },
    {
      "epoch": 1.7182175622542595,
      "grad_norm": 0.27033987641334534,
      "learning_rate": 4.593714154164088e-06,
      "loss": 0.453,
      "step": 874
    },
    {
      "epoch": 1.7201834862385321,
      "grad_norm": 0.22780223190784454,
      "learning_rate": 4.58229581456547e-06,
      "loss": 0.4449,
      "step": 875
    },
    {
      "epoch": 1.7221494102228048,
      "grad_norm": 0.24192917346954346,
      "learning_rate": 4.570879668243792e-06,
      "loss": 0.4547,
      "step": 876
    },
    {
      "epoch": 1.7241153342070774,
      "grad_norm": 0.259541392326355,
      "learning_rate": 4.559465775142843e-06,
      "loss": 0.4416,
      "step": 877
    },
    {
      "epoch": 1.7260812581913498,
      "grad_norm": 0.24240902066230774,
      "learning_rate": 4.548054195194576e-06,
      "loss": 0.437,
      "step": 878
    },
    {
      "epoch": 1.7280471821756227,
      "grad_norm": 0.23845937848091125,
      "learning_rate": 4.536644988318802e-06,
      "loss": 0.4482,
      "step": 879
    },
    {
      "epoch": 1.7300131061598951,
      "grad_norm": 0.2592415511608124,
      "learning_rate": 4.5252382144228696e-06,
      "loss": 0.4249,
      "step": 880
    },
    {
      "epoch": 1.7319790301441678,
      "grad_norm": 0.2503202557563782,
      "learning_rate": 4.5138339334013505e-06,
      "loss": 0.4435,
      "step": 881
    },
    {
      "epoch": 1.7339449541284404,
      "grad_norm": 0.23188181221485138,
      "learning_rate": 4.502432205135731e-06,
      "loss": 0.4361,
      "step": 882
    },
    {
      "epoch": 1.7359108781127128,
      "grad_norm": 0.27275577187538147,
      "learning_rate": 4.491033089494091e-06,
      "loss": 0.4644,
      "step": 883
    },
    {
      "epoch": 1.7378768020969857,
      "grad_norm": 0.27509182691574097,
      "learning_rate": 4.479636646330793e-06,
      "loss": 0.4496,
      "step": 884
    },
    {
      "epoch": 1.7398427260812581,
      "grad_norm": 0.2707621157169342,
      "learning_rate": 4.468242935486164e-06,
      "loss": 0.4384,
      "step": 885
    },
    {
      "epoch": 1.7418086500655308,
      "grad_norm": 0.24250587821006775,
      "learning_rate": 4.45685201678619e-06,
      "loss": 0.4458,
      "step": 886
    },
    {
      "epoch": 1.7437745740498034,
      "grad_norm": 0.25717586278915405,
      "learning_rate": 4.445463950042191e-06,
      "loss": 0.4288,
      "step": 887
    },
    {
      "epoch": 1.745740498034076,
      "grad_norm": 0.2693054676055908,
      "learning_rate": 4.434078795050509e-06,
      "loss": 0.4534,
      "step": 888
    },
    {
      "epoch": 1.7477064220183487,
      "grad_norm": 0.2557573616504669,
      "learning_rate": 4.4226966115922096e-06,
      "loss": 0.4385,
      "step": 889
    },
    {
      "epoch": 1.7496723460026211,
      "grad_norm": 0.224521666765213,
      "learning_rate": 4.411317459432741e-06,
      "loss": 0.4176,
      "step": 890
    },
    {
      "epoch": 1.751638269986894,
      "grad_norm": 0.27973490953445435,
      "learning_rate": 4.3999413983216434e-06,
      "loss": 0.448,
      "step": 891
    },
    {
      "epoch": 1.7536041939711664,
      "grad_norm": 0.2818668484687805,
      "learning_rate": 4.388568487992225e-06,
      "loss": 0.4555,
      "step": 892
    },
    {
      "epoch": 1.755570117955439,
      "grad_norm": 0.25364866852760315,
      "learning_rate": 4.3771987881612484e-06,
      "loss": 0.4399,
      "step": 893
    },
    {
      "epoch": 1.7575360419397117,
      "grad_norm": 0.24888482689857483,
      "learning_rate": 4.365832358528618e-06,
      "loss": 0.4504,
      "step": 894
    },
    {
      "epoch": 1.7595019659239841,
      "grad_norm": 0.2617346942424774,
      "learning_rate": 4.35446925877707e-06,
      "loss": 0.4355,
      "step": 895
    },
    {
      "epoch": 1.761467889908257,
      "grad_norm": 0.2641500234603882,
      "learning_rate": 4.343109548571855e-06,
      "loss": 0.4545,
      "step": 896
    },
    {
      "epoch": 1.7634338138925294,
      "grad_norm": 0.2263491451740265,
      "learning_rate": 4.331753287560423e-06,
      "loss": 0.4433,
      "step": 897
    },
    {
      "epoch": 1.765399737876802,
      "grad_norm": 0.2606113851070404,
      "learning_rate": 4.320400535372117e-06,
      "loss": 0.429,
      "step": 898
    },
    {
      "epoch": 1.7673656618610747,
      "grad_norm": 0.25996842980384827,
      "learning_rate": 4.3090513516178514e-06,
      "loss": 0.446,
      "step": 899
    },
    {
      "epoch": 1.7693315858453473,
      "grad_norm": 0.23425088822841644,
      "learning_rate": 4.29770579588981e-06,
      "loss": 0.4664,
      "step": 900
    },
    {
      "epoch": 1.77129750982962,
      "grad_norm": 0.2421378493309021,
      "learning_rate": 4.286363927761122e-06,
      "loss": 0.4585,
      "step": 901
    },
    {
      "epoch": 1.7732634338138924,
      "grad_norm": 0.2616322636604309,
      "learning_rate": 4.275025806785554e-06,
      "loss": 0.4585,
      "step": 902
    },
    {
      "epoch": 1.7752293577981653,
      "grad_norm": 0.23578320443630219,
      "learning_rate": 4.263691492497197e-06,
      "loss": 0.4362,
      "step": 903
    },
    {
      "epoch": 1.7771952817824377,
      "grad_norm": 0.22001796960830688,
      "learning_rate": 4.252361044410154e-06,
      "loss": 0.4431,
      "step": 904
    },
    {
      "epoch": 1.7791612057667103,
      "grad_norm": 0.24889324605464935,
      "learning_rate": 4.241034522018232e-06,
      "loss": 0.4493,
      "step": 905
    },
    {
      "epoch": 1.781127129750983,
      "grad_norm": 0.2580612897872925,
      "learning_rate": 4.229711984794614e-06,
      "loss": 0.4433,
      "step": 906
    },
    {
      "epoch": 1.7830930537352556,
      "grad_norm": 0.2380072921514511,
      "learning_rate": 4.218393492191573e-06,
      "loss": 0.4507,
      "step": 907
    },
    {
      "epoch": 1.7850589777195283,
      "grad_norm": 0.24413016438484192,
      "learning_rate": 4.207079103640129e-06,
      "loss": 0.4496,
      "step": 908
    },
    {
      "epoch": 1.7870249017038007,
      "grad_norm": 0.2281772643327713,
      "learning_rate": 4.195768878549766e-06,
      "loss": 0.4202,
      "step": 909
    },
    {
      "epoch": 1.7889908256880735,
      "grad_norm": 0.2313961535692215,
      "learning_rate": 4.184462876308097e-06,
      "loss": 0.4306,
      "step": 910
    },
    {
      "epoch": 1.790956749672346,
      "grad_norm": 0.24212126433849335,
      "learning_rate": 4.173161156280564e-06,
      "loss": 0.4366,
      "step": 911
    },
    {
      "epoch": 1.7929226736566186,
      "grad_norm": 0.2456527203321457,
      "learning_rate": 4.161863777810128e-06,
      "loss": 0.4594,
      "step": 912
    },
    {
      "epoch": 1.7948885976408913,
      "grad_norm": 0.24351178109645844,
      "learning_rate": 4.150570800216946e-06,
      "loss": 0.4497,
      "step": 913
    },
    {
      "epoch": 1.7968545216251637,
      "grad_norm": 0.2541344463825226,
      "learning_rate": 4.139282282798076e-06,
      "loss": 0.454,
      "step": 914
    },
    {
      "epoch": 1.7988204456094365,
      "grad_norm": 0.2662005126476288,
      "learning_rate": 4.127998284827148e-06,
      "loss": 0.4479,
      "step": 915
    },
    {
      "epoch": 1.800786369593709,
      "grad_norm": 0.26165616512298584,
      "learning_rate": 4.11671886555407e-06,
      "loss": 0.4249,
      "step": 916
    },
    {
      "epoch": 1.8027522935779816,
      "grad_norm": 0.2643074095249176,
      "learning_rate": 4.105444084204699e-06,
      "loss": 0.4396,
      "step": 917
    },
    {
      "epoch": 1.8047182175622543,
      "grad_norm": 0.2607715129852295,
      "learning_rate": 4.094173999980544e-06,
      "loss": 0.4366,
      "step": 918
    },
    {
      "epoch": 1.806684141546527,
      "grad_norm": 0.2415345460176468,
      "learning_rate": 4.082908672058453e-06,
      "loss": 0.4458,
      "step": 919
    },
    {
      "epoch": 1.8086500655307995,
      "grad_norm": 0.2511383891105652,
      "learning_rate": 4.071648159590294e-06,
      "loss": 0.4428,
      "step": 920
    },
    {
      "epoch": 1.810615989515072,
      "grad_norm": 0.24544315040111542,
      "learning_rate": 4.060392521702655e-06,
      "loss": 0.4462,
      "step": 921
    },
    {
      "epoch": 1.8125819134993448,
      "grad_norm": 0.24017304182052612,
      "learning_rate": 4.0491418174965225e-06,
      "loss": 0.4395,
      "step": 922
    },
    {
      "epoch": 1.8145478374836173,
      "grad_norm": 0.23082610964775085,
      "learning_rate": 4.037896106046986e-06,
      "loss": 0.4313,
      "step": 923
    },
    {
      "epoch": 1.81651376146789,
      "grad_norm": 0.24073214828968048,
      "learning_rate": 4.026655446402912e-06,
      "loss": 0.458,
      "step": 924
    },
    {
      "epoch": 1.8184796854521625,
      "grad_norm": 0.24371448159217834,
      "learning_rate": 4.01541989758664e-06,
      "loss": 0.4429,
      "step": 925
    },
    {
      "epoch": 1.820445609436435,
      "grad_norm": 0.24334824085235596,
      "learning_rate": 4.004189518593683e-06,
      "loss": 0.4395,
      "step": 926
    },
    {
      "epoch": 1.8224115334207078,
      "grad_norm": 0.26907652616500854,
      "learning_rate": 3.9929643683923965e-06,
      "loss": 0.4482,
      "step": 927
    },
    {
      "epoch": 1.8243774574049803,
      "grad_norm": 0.23593313992023468,
      "learning_rate": 3.981744505923692e-06,
      "loss": 0.4453,
      "step": 928
    },
    {
      "epoch": 1.826343381389253,
      "grad_norm": 0.2463892549276352,
      "learning_rate": 3.970529990100706e-06,
      "loss": 0.4295,
      "step": 929
    },
    {
      "epoch": 1.8283093053735255,
      "grad_norm": 0.2889455556869507,
      "learning_rate": 3.9593208798085094e-06,
      "loss": 0.4455,
      "step": 930
    },
    {
      "epoch": 1.8302752293577982,
      "grad_norm": 0.25848588347435,
      "learning_rate": 3.948117233903781e-06,
      "loss": 0.4381,
      "step": 931
    },
    {
      "epoch": 1.8322411533420708,
      "grad_norm": 0.2491040676832199,
      "learning_rate": 3.936919111214518e-06,
      "loss": 0.4493,
      "step": 932
    },
    {
      "epoch": 1.8342070773263432,
      "grad_norm": 0.27089598774909973,
      "learning_rate": 3.9257265705397065e-06,
      "loss": 0.433,
      "step": 933
    },
    {
      "epoch": 1.8361730013106161,
      "grad_norm": 0.23713093996047974,
      "learning_rate": 3.914539670649026e-06,
      "loss": 0.4403,
      "step": 934
    },
    {
      "epoch": 1.8381389252948885,
      "grad_norm": 0.2511511445045471,
      "learning_rate": 3.903358470282542e-06,
      "loss": 0.4355,
      "step": 935
    },
    {
      "epoch": 1.8401048492791612,
      "grad_norm": 0.23158514499664307,
      "learning_rate": 3.892183028150384e-06,
      "loss": 0.4426,
      "step": 936
    },
    {
      "epoch": 1.8420707732634338,
      "grad_norm": 0.23655346035957336,
      "learning_rate": 3.881013402932458e-06,
      "loss": 0.4652,
      "step": 937
    },
    {
      "epoch": 1.8440366972477065,
      "grad_norm": 0.24980229139328003,
      "learning_rate": 3.869849653278114e-06,
      "loss": 0.4421,
      "step": 938
    },
    {
      "epoch": 1.8460026212319791,
      "grad_norm": 0.24362362921237946,
      "learning_rate": 3.8586918378058595e-06,
      "loss": 0.4269,
      "step": 939
    },
    {
      "epoch": 1.8479685452162515,
      "grad_norm": 0.23607191443443298,
      "learning_rate": 3.847540015103038e-06,
      "loss": 0.4618,
      "step": 940
    },
    {
      "epoch": 1.8499344692005244,
      "grad_norm": 0.22779516875743866,
      "learning_rate": 3.836394243725529e-06,
      "loss": 0.4473,
      "step": 941
    },
    {
      "epoch": 1.8519003931847968,
      "grad_norm": 0.2740325629711151,
      "learning_rate": 3.8252545821974385e-06,
      "loss": 0.4491,
      "step": 942
    },
    {
      "epoch": 1.8538663171690695,
      "grad_norm": 0.26140812039375305,
      "learning_rate": 3.814121089010786e-06,
      "loss": 0.4311,
      "step": 943
    },
    {
      "epoch": 1.855832241153342,
      "grad_norm": 0.2269832193851471,
      "learning_rate": 3.8029938226252095e-06,
      "loss": 0.4228,
      "step": 944
    },
    {
      "epoch": 1.8577981651376145,
      "grad_norm": 0.2418784499168396,
      "learning_rate": 3.791872841467643e-06,
      "loss": 0.4206,
      "step": 945
    },
    {
      "epoch": 1.8597640891218874,
      "grad_norm": 0.25120002031326294,
      "learning_rate": 3.780758203932028e-06,
      "loss": 0.4367,
      "step": 946
    },
    {
      "epoch": 1.8617300131061598,
      "grad_norm": 0.2378482073545456,
      "learning_rate": 3.769649968378989e-06,
      "loss": 0.4405,
      "step": 947
    },
    {
      "epoch": 1.8636959370904325,
      "grad_norm": 0.2553141713142395,
      "learning_rate": 3.758548193135536e-06,
      "loss": 0.4472,
      "step": 948
    },
    {
      "epoch": 1.865661861074705,
      "grad_norm": 0.23460035026073456,
      "learning_rate": 3.747452936494761e-06,
      "loss": 0.4426,
      "step": 949
    },
    {
      "epoch": 1.8676277850589778,
      "grad_norm": 0.2531510293483734,
      "learning_rate": 3.7363642567155254e-06,
      "loss": 0.4687,
      "step": 950
    },
    {
      "epoch": 1.8695937090432504,
      "grad_norm": 0.21824617683887482,
      "learning_rate": 3.7252822120221592e-06,
      "loss": 0.4408,
      "step": 951
    },
    {
      "epoch": 1.8715596330275228,
      "grad_norm": 0.2500864863395691,
      "learning_rate": 3.714206860604148e-06,
      "loss": 0.4329,
      "step": 952
    },
    {
      "epoch": 1.8735255570117957,
      "grad_norm": 0.2598780691623688,
      "learning_rate": 3.7031382606158396e-06,
      "loss": 0.459,
      "step": 953
    },
    {
      "epoch": 1.875491480996068,
      "grad_norm": 0.25134357810020447,
      "learning_rate": 3.6920764701761263e-06,
      "loss": 0.4465,
      "step": 954
    },
    {
      "epoch": 1.8774574049803407,
      "grad_norm": 0.24129661917686462,
      "learning_rate": 3.6810215473681433e-06,
      "loss": 0.4462,
      "step": 955
    },
    {
      "epoch": 1.8794233289646134,
      "grad_norm": 0.24308183789253235,
      "learning_rate": 3.6699735502389734e-06,
      "loss": 0.4255,
      "step": 956
    },
    {
      "epoch": 1.8813892529488858,
      "grad_norm": 0.27380120754241943,
      "learning_rate": 3.6589325367993243e-06,
      "loss": 0.4181,
      "step": 957
    },
    {
      "epoch": 1.8833551769331587,
      "grad_norm": 0.22356578707695007,
      "learning_rate": 3.6478985650232413e-06,
      "loss": 0.437,
      "step": 958
    },
    {
      "epoch": 1.885321100917431,
      "grad_norm": 0.22350762784481049,
      "learning_rate": 3.636871692847791e-06,
      "loss": 0.4359,
      "step": 959
    },
    {
      "epoch": 1.8872870249017037,
      "grad_norm": 0.21726961433887482,
      "learning_rate": 3.625851978172765e-06,
      "loss": 0.4358,
      "step": 960
    },
    {
      "epoch": 1.8892529488859764,
      "grad_norm": 0.2217891365289688,
      "learning_rate": 3.614839478860369e-06,
      "loss": 0.434,
      "step": 961
    },
    {
      "epoch": 1.891218872870249,
      "grad_norm": 0.2290656715631485,
      "learning_rate": 3.6038342527349225e-06,
      "loss": 0.4239,
      "step": 962
    },
    {
      "epoch": 1.8931847968545217,
      "grad_norm": 0.21448542177677155,
      "learning_rate": 3.59283635758256e-06,
      "loss": 0.4377,
      "step": 963
    },
    {
      "epoch": 1.895150720838794,
      "grad_norm": 0.2176762968301773,
      "learning_rate": 3.5818458511509135e-06,
      "loss": 0.4366,
      "step": 964
    },
    {
      "epoch": 1.897116644823067,
      "grad_norm": 0.23071660101413727,
      "learning_rate": 3.57086279114883e-06,
      "loss": 0.4415,
      "step": 965
    },
    {
      "epoch": 1.8990825688073394,
      "grad_norm": 0.23253513872623444,
      "learning_rate": 3.5598872352460457e-06,
      "loss": 0.4366,
      "step": 966
    },
    {
      "epoch": 1.901048492791612,
      "grad_norm": 0.22595295310020447,
      "learning_rate": 3.548919241072901e-06,
      "loss": 0.4212,
      "step": 967
    },
    {
      "epoch": 1.9030144167758847,
      "grad_norm": 0.22117412090301514,
      "learning_rate": 3.537958866220031e-06,
      "loss": 0.4363,
      "step": 968
    },
    {
      "epoch": 1.9049803407601573,
      "grad_norm": 0.23329798877239227,
      "learning_rate": 3.527006168238061e-06,
      "loss": 0.4164,
      "step": 969
    },
    {
      "epoch": 1.90694626474443,
      "grad_norm": 0.21395321190357208,
      "learning_rate": 3.5160612046373067e-06,
      "loss": 0.435,
      "step": 970
    },
    {
      "epoch": 1.9089121887287024,
      "grad_norm": 0.2612607777118683,
      "learning_rate": 3.505124032887471e-06,
      "loss": 0.4505,
      "step": 971
    },
    {
      "epoch": 1.9108781127129753,
      "grad_norm": 0.2517973482608795,
      "learning_rate": 3.4941947104173514e-06,
      "loss": 0.4362,
      "step": 972
    },
    {
      "epoch": 1.9128440366972477,
      "grad_norm": 0.25207528471946716,
      "learning_rate": 3.4832732946145187e-06,
      "loss": 0.4461,
      "step": 973
    },
    {
      "epoch": 1.9148099606815203,
      "grad_norm": 0.2599436938762665,
      "learning_rate": 3.472359842825037e-06,
      "loss": 0.4332,
      "step": 974
    },
    {
      "epoch": 1.916775884665793,
      "grad_norm": 0.24764251708984375,
      "learning_rate": 3.4614544123531476e-06,
      "loss": 0.4217,
      "step": 975
    },
    {
      "epoch": 1.9187418086500654,
      "grad_norm": 0.2398913949728012,
      "learning_rate": 3.450557060460975e-06,
      "loss": 0.4417,
      "step": 976
    },
    {
      "epoch": 1.9207077326343382,
      "grad_norm": 0.2356083244085312,
      "learning_rate": 3.4396678443682274e-06,
      "loss": 0.4305,
      "step": 977
    },
    {
      "epoch": 1.9226736566186107,
      "grad_norm": 0.2680135667324066,
      "learning_rate": 3.428786821251888e-06,
      "loss": 0.4484,
      "step": 978
    },
    {
      "epoch": 1.9246395806028833,
      "grad_norm": 0.22987043857574463,
      "learning_rate": 3.417914048245927e-06,
      "loss": 0.4395,
      "step": 979
    },
    {
      "epoch": 1.926605504587156,
      "grad_norm": 0.2541668117046356,
      "learning_rate": 3.4070495824409876e-06,
      "loss": 0.4183,
      "step": 980
    },
    {
      "epoch": 1.9285714285714286,
      "grad_norm": 0.2168930470943451,
      "learning_rate": 3.3961934808841023e-06,
      "loss": 0.4343,
      "step": 981
    },
    {
      "epoch": 1.9305373525557012,
      "grad_norm": 0.23768441379070282,
      "learning_rate": 3.385345800578376e-06,
      "loss": 0.4511,
      "step": 982
    },
    {
      "epoch": 1.9325032765399737,
      "grad_norm": 0.24167458713054657,
      "learning_rate": 3.374506598482703e-06,
      "loss": 0.4505,
      "step": 983
    },
    {
      "epoch": 1.9344692005242465,
      "grad_norm": 0.22421284019947052,
      "learning_rate": 3.363675931511455e-06,
      "loss": 0.4556,
      "step": 984
    },
    {
      "epoch": 1.936435124508519,
      "grad_norm": 0.23368152976036072,
      "learning_rate": 3.3528538565341885e-06,
      "loss": 0.4276,
      "step": 985
    },
    {
      "epoch": 1.9384010484927916,
      "grad_norm": 0.26514336466789246,
      "learning_rate": 3.3420404303753485e-06,
      "loss": 0.4469,
      "step": 986
    },
    {
      "epoch": 1.9403669724770642,
      "grad_norm": 0.24340538680553436,
      "learning_rate": 3.331235709813962e-06,
      "loss": 0.4406,
      "step": 987
    },
    {
      "epoch": 1.9423328964613367,
      "grad_norm": 0.2298521101474762,
      "learning_rate": 3.3204397515833494e-06,
      "loss": 0.437,
      "step": 988
    },
    {
      "epoch": 1.9442988204456095,
      "grad_norm": 0.2267940640449524,
      "learning_rate": 3.309652612370816e-06,
      "loss": 0.4432,
      "step": 989
    },
    {
      "epoch": 1.946264744429882,
      "grad_norm": 0.24631616473197937,
      "learning_rate": 3.29887434881737e-06,
      "loss": 0.4255,
      "step": 990
    },
    {
      "epoch": 1.9482306684141546,
      "grad_norm": 0.22750426828861237,
      "learning_rate": 3.288105017517405e-06,
      "loss": 0.4252,
      "step": 991
    },
    {
      "epoch": 1.9501965923984272,
      "grad_norm": 0.22862617671489716,
      "learning_rate": 3.277344675018417e-06,
      "loss": 0.4496,
      "step": 992
    },
    {
      "epoch": 1.9521625163826999,
      "grad_norm": 0.21800747513771057,
      "learning_rate": 3.2665933778207082e-06,
      "loss": 0.4136,
      "step": 993
    },
    {
      "epoch": 1.9541284403669725,
      "grad_norm": 0.2186320722103119,
      "learning_rate": 3.255851182377077e-06,
      "loss": 0.4328,
      "step": 994
    },
    {
      "epoch": 1.956094364351245,
      "grad_norm": 0.23287145793437958,
      "learning_rate": 3.2451181450925406e-06,
      "loss": 0.4273,
      "step": 995
    },
    {
      "epoch": 1.9580602883355178,
      "grad_norm": 0.2228022962808609,
      "learning_rate": 3.234394322324019e-06,
      "loss": 0.4402,
      "step": 996
    },
    {
      "epoch": 1.9600262123197902,
      "grad_norm": 0.2278522402048111,
      "learning_rate": 3.223679770380057e-06,
      "loss": 0.4347,
      "step": 997
    },
    {
      "epoch": 1.9619921363040629,
      "grad_norm": 0.21856601536273956,
      "learning_rate": 3.2129745455205135e-06,
      "loss": 0.4267,
      "step": 998
    },
    {
      "epoch": 1.9639580602883355,
      "grad_norm": 0.21911118924617767,
      "learning_rate": 3.2022787039562745e-06,
      "loss": 0.4325,
      "step": 999
    },
    {
      "epoch": 1.9659239842726082,
      "grad_norm": 0.2267688512802124,
      "learning_rate": 3.191592301848961e-06,
      "loss": 0.4536,
      "step": 1000
    },
    {
      "epoch": 1.9678899082568808,
      "grad_norm": 0.21859389543533325,
      "learning_rate": 3.180915395310623e-06,
      "loss": 0.4686,
      "step": 1001
    },
    {
      "epoch": 1.9698558322411532,
      "grad_norm": 0.22693945467472076,
      "learning_rate": 3.170248040403457e-06,
      "loss": 0.426,
      "step": 1002
    },
    {
      "epoch": 1.971821756225426,
      "grad_norm": 0.23281870782375336,
      "learning_rate": 3.1595902931394983e-06,
      "loss": 0.4416,
      "step": 1003
    },
    {
      "epoch": 1.9737876802096985,
      "grad_norm": 0.22778624296188354,
      "learning_rate": 3.1489422094803458e-06,
      "loss": 0.443,
      "step": 1004
    },
    {
      "epoch": 1.9757536041939712,
      "grad_norm": 0.214397594332695,
      "learning_rate": 3.138303845336844e-06,
      "loss": 0.4338,
      "step": 1005
    },
    {
      "epoch": 1.9777195281782438,
      "grad_norm": 0.2683956027030945,
      "learning_rate": 3.127675256568813e-06,
      "loss": 0.4484,
      "step": 1006
    },
    {
      "epoch": 1.9796854521625162,
      "grad_norm": 0.23611973226070404,
      "learning_rate": 3.1170564989847374e-06,
      "loss": 0.4338,
      "step": 1007
    },
    {
      "epoch": 1.981651376146789,
      "grad_norm": 0.22654104232788086,
      "learning_rate": 3.1064476283414818e-06,
      "loss": 0.4365,
      "step": 1008
    },
    {
      "epoch": 1.9836173001310615,
      "grad_norm": 0.2249293178319931,
      "learning_rate": 3.095848700344001e-06,
      "loss": 0.426,
      "step": 1009
    },
    {
      "epoch": 1.9855832241153342,
      "grad_norm": 0.2095581293106079,
      "learning_rate": 3.085259770645036e-06,
      "loss": 0.4408,
      "step": 1010
    },
    {
      "epoch": 1.9875491480996068,
      "grad_norm": 0.2307659089565277,
      "learning_rate": 3.074680894844837e-06,
      "loss": 0.4314,
      "step": 1011
    },
    {
      "epoch": 1.9895150720838795,
      "grad_norm": 0.2108234465122223,
      "learning_rate": 3.064112128490853e-06,
      "loss": 0.4483,
      "step": 1012
    },
    {
      "epoch": 1.991480996068152,
      "grad_norm": 0.2225804477930069,
      "learning_rate": 3.0535535270774624e-06,
      "loss": 0.4384,
      "step": 1013
    },
    {
      "epoch": 1.9934469200524245,
      "grad_norm": 0.20851469039916992,
      "learning_rate": 3.04300514604566e-06,
      "loss": 0.427,
      "step": 1014
    },
    {
      "epoch": 1.9954128440366974,
      "grad_norm": 0.24080389738082886,
      "learning_rate": 3.0324670407827788e-06,
      "loss": 0.4377,
      "step": 1015
    },
    {
      "epoch": 1.9973787680209698,
      "grad_norm": 0.22984230518341064,
      "learning_rate": 3.0219392666221975e-06,
      "loss": 0.441,
      "step": 1016
    },
    {
      "epoch": 1.9993446920052425,
      "grad_norm": 0.25892481207847595,
      "learning_rate": 3.011421878843044e-06,
      "loss": 0.5071,
      "step": 1017
    },
    {
      "epoch": 2.001310615989515,
      "grad_norm": 0.2732914984226227,
      "learning_rate": 3.000914932669917e-06,
      "loss": 0.5019,
      "step": 1018
    },
    {
      "epoch": 2.0032765399737875,
      "grad_norm": 0.22422873973846436,
      "learning_rate": 2.990418483272579e-06,
      "loss": 0.4274,
      "step": 1019
    },
    {
      "epoch": 2.0052424639580604,
      "grad_norm": 0.2819391191005707,
      "learning_rate": 2.9799325857656856e-06,
      "loss": 0.4444,
      "step": 1020
    },
    {
      "epoch": 2.007208387942333,
      "grad_norm": 0.23747709393501282,
      "learning_rate": 2.96945729520848e-06,
      "loss": 0.4291,
      "step": 1021
    },
    {
      "epoch": 2.0091743119266057,
      "grad_norm": 0.21126174926757812,
      "learning_rate": 2.958992666604511e-06,
      "loss": 0.4128,
      "step": 1022
    },
    {
      "epoch": 2.011140235910878,
      "grad_norm": 0.2312362641096115,
      "learning_rate": 2.948538754901349e-06,
      "loss": 0.4242,
      "step": 1023
    },
    {
      "epoch": 2.0131061598951505,
      "grad_norm": 0.25565868616104126,
      "learning_rate": 2.938095614990285e-06,
      "loss": 0.4331,
      "step": 1024
    },
    {
      "epoch": 2.0150720838794234,
      "grad_norm": 0.23785267770290375,
      "learning_rate": 2.9276633017060563e-06,
      "loss": 0.4268,
      "step": 1025
    },
    {
      "epoch": 2.017038007863696,
      "grad_norm": 0.23399539291858673,
      "learning_rate": 2.917241869826545e-06,
      "loss": 0.4207,
      "step": 1026
    },
    {
      "epoch": 2.0190039318479687,
      "grad_norm": 0.23603691160678864,
      "learning_rate": 2.906831374072504e-06,
      "loss": 0.4207,
      "step": 1027
    },
    {
      "epoch": 2.020969855832241,
      "grad_norm": 0.23575784265995026,
      "learning_rate": 2.89643186910726e-06,
      "loss": 0.416,
      "step": 1028
    },
    {
      "epoch": 2.022935779816514,
      "grad_norm": 0.2484109103679657,
      "learning_rate": 2.8860434095364266e-06,
      "loss": 0.4299,
      "step": 1029
    },
    {
      "epoch": 2.0249017038007864,
      "grad_norm": 0.22861795127391815,
      "learning_rate": 2.875666049907625e-06,
      "loss": 0.4118,
      "step": 1030
    },
    {
      "epoch": 2.026867627785059,
      "grad_norm": 0.23161958158016205,
      "learning_rate": 2.86529984471019e-06,
      "loss": 0.4158,
      "step": 1031
    },
    {
      "epoch": 2.0288335517693317,
      "grad_norm": 0.22446328401565552,
      "learning_rate": 2.8549448483748888e-06,
      "loss": 0.4041,
      "step": 1032
    },
    {
      "epoch": 2.030799475753604,
      "grad_norm": 0.27122098207473755,
      "learning_rate": 2.8446011152736295e-06,
      "loss": 0.4351,
      "step": 1033
    },
    {
      "epoch": 2.032765399737877,
      "grad_norm": 0.23165744543075562,
      "learning_rate": 2.834268699719187e-06,
      "loss": 0.4329,
      "step": 1034
    },
    {
      "epoch": 2.0347313237221494,
      "grad_norm": 0.23846569657325745,
      "learning_rate": 2.8239476559649013e-06,
      "loss": 0.4179,
      "step": 1035
    },
    {
      "epoch": 2.036697247706422,
      "grad_norm": 0.2745479941368103,
      "learning_rate": 2.8136380382044036e-06,
      "loss": 0.4211,
      "step": 1036
    },
    {
      "epoch": 2.0386631716906947,
      "grad_norm": 0.2304745316505432,
      "learning_rate": 2.803339900571337e-06,
      "loss": 0.4348,
      "step": 1037
    },
    {
      "epoch": 2.040629095674967,
      "grad_norm": 0.22313368320465088,
      "learning_rate": 2.7930532971390543e-06,
      "loss": 0.4273,
      "step": 1038
    },
    {
      "epoch": 2.04259501965924,
      "grad_norm": 0.2543697953224182,
      "learning_rate": 2.7827782819203497e-06,
      "loss": 0.4419,
      "step": 1039
    },
    {
      "epoch": 2.0445609436435124,
      "grad_norm": 0.2175356149673462,
      "learning_rate": 2.7725149088671733e-06,
      "loss": 0.4189,
      "step": 1040
    },
    {
      "epoch": 2.0465268676277852,
      "grad_norm": 0.24138113856315613,
      "learning_rate": 2.762263231870339e-06,
      "loss": 0.4346,
      "step": 1041
    },
    {
      "epoch": 2.0484927916120577,
      "grad_norm": 0.23216262459754944,
      "learning_rate": 2.752023304759248e-06,
      "loss": 0.4362,
      "step": 1042
    },
    {
      "epoch": 2.05045871559633,
      "grad_norm": 0.21769055724143982,
      "learning_rate": 2.7417951813016097e-06,
      "loss": 0.4262,
      "step": 1043
    },
    {
      "epoch": 2.052424639580603,
      "grad_norm": 0.23920607566833496,
      "learning_rate": 2.7315789152031504e-06,
      "loss": 0.4207,
      "step": 1044
    },
    {
      "epoch": 2.0543905635648754,
      "grad_norm": 0.24695385992527008,
      "learning_rate": 2.721374560107336e-06,
      "loss": 0.4227,
      "step": 1045
    },
    {
      "epoch": 2.0563564875491482,
      "grad_norm": 0.23072242736816406,
      "learning_rate": 2.7111821695950957e-06,
      "loss": 0.4289,
      "step": 1046
    },
    {
      "epoch": 2.0583224115334207,
      "grad_norm": 0.22974249720573425,
      "learning_rate": 2.7010017971845267e-06,
      "loss": 0.429,
      "step": 1047
    },
    {
      "epoch": 2.0602883355176935,
      "grad_norm": 0.2350340336561203,
      "learning_rate": 2.6908334963306305e-06,
      "loss": 0.4434,
      "step": 1048
    },
    {
      "epoch": 2.062254259501966,
      "grad_norm": 0.25064942240715027,
      "learning_rate": 2.6806773204250148e-06,
      "loss": 0.4253,
      "step": 1049
    },
    {
      "epoch": 2.0642201834862384,
      "grad_norm": 0.23788711428642273,
      "learning_rate": 2.6705333227956304e-06,
      "loss": 0.4119,
      "step": 1050
    },
    {
      "epoch": 2.0661861074705112,
      "grad_norm": 0.2238015979528427,
      "learning_rate": 2.6604015567064756e-06,
      "loss": 0.4324,
      "step": 1051
    },
    {
      "epoch": 2.0681520314547837,
      "grad_norm": 0.21394391357898712,
      "learning_rate": 2.650282075357325e-06,
      "loss": 0.4102,
      "step": 1052
    },
    {
      "epoch": 2.0701179554390565,
      "grad_norm": 0.23077036440372467,
      "learning_rate": 2.6401749318834528e-06,
      "loss": 0.4204,
      "step": 1053
    },
    {
      "epoch": 2.072083879423329,
      "grad_norm": 0.23175548017024994,
      "learning_rate": 2.6300801793553433e-06,
      "loss": 0.462,
      "step": 1054
    },
    {
      "epoch": 2.0740498034076014,
      "grad_norm": 0.22535480558872223,
      "learning_rate": 2.619997870778424e-06,
      "loss": 0.4178,
      "step": 1055
    },
    {
      "epoch": 2.0760157273918742,
      "grad_norm": 0.2462986558675766,
      "learning_rate": 2.609928059092779e-06,
      "loss": 0.4117,
      "step": 1056
    },
    {
      "epoch": 2.0779816513761467,
      "grad_norm": 0.26453328132629395,
      "learning_rate": 2.599870797172872e-06,
      "loss": 0.4261,
      "step": 1057
    },
    {
      "epoch": 2.0799475753604195,
      "grad_norm": 0.2081374228000641,
      "learning_rate": 2.589826137827277e-06,
      "loss": 0.4206,
      "step": 1058
    },
    {
      "epoch": 2.081913499344692,
      "grad_norm": 0.27211225032806396,
      "learning_rate": 2.579794133798388e-06,
      "loss": 0.4513,
      "step": 1059
    },
    {
      "epoch": 2.083879423328965,
      "grad_norm": 0.24931125342845917,
      "learning_rate": 2.5697748377621505e-06,
      "loss": 0.4104,
      "step": 1060
    },
    {
      "epoch": 2.0858453473132372,
      "grad_norm": 0.23902453482151031,
      "learning_rate": 2.559768302327783e-06,
      "loss": 0.4114,
      "step": 1061
    },
    {
      "epoch": 2.0878112712975097,
      "grad_norm": 0.20612815022468567,
      "learning_rate": 2.549774580037504e-06,
      "loss": 0.4109,
      "step": 1062
    },
    {
      "epoch": 2.0897771952817825,
      "grad_norm": 0.2153496891260147,
      "learning_rate": 2.539793723366247e-06,
      "loss": 0.3957,
      "step": 1063
    },
    {
      "epoch": 2.091743119266055,
      "grad_norm": 0.24023601412773132,
      "learning_rate": 2.529825784721397e-06,
      "loss": 0.4136,
      "step": 1064
    },
    {
      "epoch": 2.093709043250328,
      "grad_norm": 0.25313615798950195,
      "learning_rate": 2.5198708164425046e-06,
      "loss": 0.4243,
      "step": 1065
    },
    {
      "epoch": 2.0956749672346002,
      "grad_norm": 0.22436189651489258,
      "learning_rate": 2.509928870801015e-06,
      "loss": 0.4136,
      "step": 1066
    },
    {
      "epoch": 2.097640891218873,
      "grad_norm": 0.22276759147644043,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.4469,
      "step": 1067
    },
    {
      "epoch": 2.0996068152031455,
      "grad_norm": 0.22407156229019165,
      "learning_rate": 2.4900842561738736e-06,
      "loss": 0.4213,
      "step": 1068
    },
    {
      "epoch": 2.101572739187418,
      "grad_norm": 0.244294673204422,
      "learning_rate": 2.4801816913881242e-06,
      "loss": 0.4247,
      "step": 1069
    },
    {
      "epoch": 2.103538663171691,
      "grad_norm": 0.2237965315580368,
      "learning_rate": 2.4702923576390377e-06,
      "loss": 0.4201,
      "step": 1070
    },
    {
      "epoch": 2.1055045871559632,
      "grad_norm": 0.22248247265815735,
      "learning_rate": 2.4604163068534313e-06,
      "loss": 0.437,
      "step": 1071
    },
    {
      "epoch": 2.107470511140236,
      "grad_norm": 0.21586407721042633,
      "learning_rate": 2.4505535908883714e-06,
      "loss": 0.423,
      "step": 1072
    },
    {
      "epoch": 2.1094364351245085,
      "grad_norm": 0.22616347670555115,
      "learning_rate": 2.4407042615309066e-06,
      "loss": 0.4139,
      "step": 1073
    },
    {
      "epoch": 2.111402359108781,
      "grad_norm": 0.23774725198745728,
      "learning_rate": 2.4308683704978e-06,
      "loss": 0.4141,
      "step": 1074
    },
    {
      "epoch": 2.113368283093054,
      "grad_norm": 0.22781816124916077,
      "learning_rate": 2.421045969435246e-06,
      "loss": 0.4261,
      "step": 1075
    },
    {
      "epoch": 2.1153342070773262,
      "grad_norm": 0.21420340240001678,
      "learning_rate": 2.411237109918613e-06,
      "loss": 0.4065,
      "step": 1076
    },
    {
      "epoch": 2.117300131061599,
      "grad_norm": 0.21004629135131836,
      "learning_rate": 2.401441843452159e-06,
      "loss": 0.433,
      "step": 1077
    },
    {
      "epoch": 2.1192660550458715,
      "grad_norm": 0.2432694286108017,
      "learning_rate": 2.391660221468773e-06,
      "loss": 0.4172,
      "step": 1078
    },
    {
      "epoch": 2.1212319790301444,
      "grad_norm": 0.23806829750537872,
      "learning_rate": 2.3818922953296937e-06,
      "loss": 0.444,
      "step": 1079
    },
    {
      "epoch": 2.123197903014417,
      "grad_norm": 0.21620041131973267,
      "learning_rate": 2.372138116324254e-06,
      "loss": 0.43,
      "step": 1080
    },
    {
      "epoch": 2.125163826998689,
      "grad_norm": 0.21120987832546234,
      "learning_rate": 2.3623977356695977e-06,
      "loss": 0.4167,
      "step": 1081
    },
    {
      "epoch": 2.127129750982962,
      "grad_norm": 0.2339550107717514,
      "learning_rate": 2.352671204510415e-06,
      "loss": 0.4321,
      "step": 1082
    },
    {
      "epoch": 2.1290956749672345,
      "grad_norm": 0.22796887159347534,
      "learning_rate": 2.342958573918682e-06,
      "loss": 0.4121,
      "step": 1083
    },
    {
      "epoch": 2.1310615989515074,
      "grad_norm": 0.2140340507030487,
      "learning_rate": 2.333259894893378e-06,
      "loss": 0.4411,
      "step": 1084
    },
    {
      "epoch": 2.13302752293578,
      "grad_norm": 0.2166963368654251,
      "learning_rate": 2.323575218360234e-06,
      "loss": 0.4089,
      "step": 1085
    },
    {
      "epoch": 2.134993446920052,
      "grad_norm": 0.21713577210903168,
      "learning_rate": 2.3139045951714473e-06,
      "loss": 0.4237,
      "step": 1086
    },
    {
      "epoch": 2.136959370904325,
      "grad_norm": 0.21196705102920532,
      "learning_rate": 2.304248076105434e-06,
      "loss": 0.4223,
      "step": 1087
    },
    {
      "epoch": 2.1389252948885975,
      "grad_norm": 0.21942131221294403,
      "learning_rate": 2.2946057118665436e-06,
      "loss": 0.4268,
      "step": 1088
    },
    {
      "epoch": 2.1408912188728704,
      "grad_norm": 0.21689389646053314,
      "learning_rate": 2.2849775530848057e-06,
      "loss": 0.4242,
      "step": 1089
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 0.2458450198173523,
      "learning_rate": 2.2753636503156624e-06,
      "loss": 0.418,
      "step": 1090
    },
    {
      "epoch": 2.1448230668414157,
      "grad_norm": 0.21314334869384766,
      "learning_rate": 2.2657640540396935e-06,
      "loss": 0.4093,
      "step": 1091
    },
    {
      "epoch": 2.146788990825688,
      "grad_norm": 0.20487774908542633,
      "learning_rate": 2.256178814662368e-06,
      "loss": 0.4275,
      "step": 1092
    },
    {
      "epoch": 2.1487549148099605,
      "grad_norm": 0.21627502143383026,
      "learning_rate": 2.2466079825137595e-06,
      "loss": 0.4207,
      "step": 1093
    },
    {
      "epoch": 2.1507208387942334,
      "grad_norm": 0.21706241369247437,
      "learning_rate": 2.2370516078483014e-06,
      "loss": 0.4117,
      "step": 1094
    },
    {
      "epoch": 2.152686762778506,
      "grad_norm": 0.22046822309494019,
      "learning_rate": 2.227509740844508e-06,
      "loss": 0.428,
      "step": 1095
    },
    {
      "epoch": 2.1546526867627787,
      "grad_norm": 0.21805192530155182,
      "learning_rate": 2.217982431604719e-06,
      "loss": 0.429,
      "step": 1096
    },
    {
      "epoch": 2.156618610747051,
      "grad_norm": 0.21168456971645355,
      "learning_rate": 2.2084697301548336e-06,
      "loss": 0.419,
      "step": 1097
    },
    {
      "epoch": 2.1585845347313235,
      "grad_norm": 0.22348026931285858,
      "learning_rate": 2.198971686444047e-06,
      "loss": 0.418,
      "step": 1098
    },
    {
      "epoch": 2.1605504587155964,
      "grad_norm": 0.23373623192310333,
      "learning_rate": 2.189488350344596e-06,
      "loss": 0.4159,
      "step": 1099
    },
    {
      "epoch": 2.162516382699869,
      "grad_norm": 0.21262992918491364,
      "learning_rate": 2.1800197716514824e-06,
      "loss": 0.4343,
      "step": 1100
    },
    {
      "epoch": 2.1644823066841417,
      "grad_norm": 0.21812313795089722,
      "learning_rate": 2.1705660000822286e-06,
      "loss": 0.421,
      "step": 1101
    },
    {
      "epoch": 2.166448230668414,
      "grad_norm": 0.2137649804353714,
      "learning_rate": 2.1611270852766e-06,
      "loss": 0.4233,
      "step": 1102
    },
    {
      "epoch": 2.168414154652687,
      "grad_norm": 0.219873309135437,
      "learning_rate": 2.151703076796356e-06,
      "loss": 0.4232,
      "step": 1103
    },
    {
      "epoch": 2.1703800786369594,
      "grad_norm": 0.20803530514240265,
      "learning_rate": 2.1422940241249875e-06,
      "loss": 0.4146,
      "step": 1104
    },
    {
      "epoch": 2.172346002621232,
      "grad_norm": 0.2098754644393921,
      "learning_rate": 2.1328999766674504e-06,
      "loss": 0.4335,
      "step": 1105
    },
    {
      "epoch": 2.1743119266055047,
      "grad_norm": 0.20901520550251007,
      "learning_rate": 2.1235209837499186e-06,
      "loss": 0.4263,
      "step": 1106
    },
    {
      "epoch": 2.176277850589777,
      "grad_norm": 0.2198159247636795,
      "learning_rate": 2.1141570946195106e-06,
      "loss": 0.4132,
      "step": 1107
    },
    {
      "epoch": 2.17824377457405,
      "grad_norm": 0.21341699361801147,
      "learning_rate": 2.104808358444044e-06,
      "loss": 0.4188,
      "step": 1108
    },
    {
      "epoch": 2.1802096985583224,
      "grad_norm": 0.20872806012630463,
      "learning_rate": 2.095474824311769e-06,
      "loss": 0.4293,
      "step": 1109
    },
    {
      "epoch": 2.1821756225425952,
      "grad_norm": 0.21234822273254395,
      "learning_rate": 2.086156541231109e-06,
      "loss": 0.4352,
      "step": 1110
    },
    {
      "epoch": 2.1841415465268676,
      "grad_norm": 0.2104458063840866,
      "learning_rate": 2.0768535581304154e-06,
      "loss": 0.4454,
      "step": 1111
    },
    {
      "epoch": 2.18610747051114,
      "grad_norm": 0.2399250715970993,
      "learning_rate": 2.0675659238576955e-06,
      "loss": 0.4357,
      "step": 1112
    },
    {
      "epoch": 2.188073394495413,
      "grad_norm": 0.21785323321819305,
      "learning_rate": 2.0582936871803692e-06,
      "loss": 0.4272,
      "step": 1113
    },
    {
      "epoch": 2.1900393184796854,
      "grad_norm": 0.19603775441646576,
      "learning_rate": 2.049036896785002e-06,
      "loss": 0.3989,
      "step": 1114
    },
    {
      "epoch": 2.1920052424639582,
      "grad_norm": 0.20027326047420502,
      "learning_rate": 2.0397956012770555e-06,
      "loss": 0.4016,
      "step": 1115
    },
    {
      "epoch": 2.1939711664482306,
      "grad_norm": 0.19911032915115356,
      "learning_rate": 2.0305698491806297e-06,
      "loss": 0.4156,
      "step": 1116
    },
    {
      "epoch": 2.195937090432503,
      "grad_norm": 0.21624262630939484,
      "learning_rate": 2.0213596889382153e-06,
      "loss": 0.422,
      "step": 1117
    },
    {
      "epoch": 2.197903014416776,
      "grad_norm": 0.21816721558570862,
      "learning_rate": 2.0121651689104264e-06,
      "loss": 0.4191,
      "step": 1118
    },
    {
      "epoch": 2.1998689384010484,
      "grad_norm": 0.20420682430267334,
      "learning_rate": 2.0029863373757553e-06,
      "loss": 0.4315,
      "step": 1119
    },
    {
      "epoch": 2.2018348623853212,
      "grad_norm": 0.20504304766654968,
      "learning_rate": 1.9938232425303215e-06,
      "loss": 0.43,
      "step": 1120
    },
    {
      "epoch": 2.2038007863695936,
      "grad_norm": 0.19833733141422272,
      "learning_rate": 1.984675932487608e-06,
      "loss": 0.4155,
      "step": 1121
    },
    {
      "epoch": 2.2057667103538665,
      "grad_norm": 0.21118539571762085,
      "learning_rate": 1.9755444552782228e-06,
      "loss": 0.4167,
      "step": 1122
    },
    {
      "epoch": 2.207732634338139,
      "grad_norm": 0.1972157508134842,
      "learning_rate": 1.96642885884963e-06,
      "loss": 0.4247,
      "step": 1123
    },
    {
      "epoch": 2.2096985583224114,
      "grad_norm": 0.2121991664171219,
      "learning_rate": 1.957329191065916e-06,
      "loss": 0.4154,
      "step": 1124
    },
    {
      "epoch": 2.211664482306684,
      "grad_norm": 0.2141561657190323,
      "learning_rate": 1.948245499707523e-06,
      "loss": 0.4172,
      "step": 1125
    },
    {
      "epoch": 2.2136304062909566,
      "grad_norm": 0.20581470429897308,
      "learning_rate": 1.939177832471004e-06,
      "loss": 0.4043,
      "step": 1126
    },
    {
      "epoch": 2.2155963302752295,
      "grad_norm": 0.21256645023822784,
      "learning_rate": 1.9301262369687794e-06,
      "loss": 0.4369,
      "step": 1127
    },
    {
      "epoch": 2.217562254259502,
      "grad_norm": 0.224394753575325,
      "learning_rate": 1.9210907607288728e-06,
      "loss": 0.415,
      "step": 1128
    },
    {
      "epoch": 2.219528178243775,
      "grad_norm": 0.22776895761489868,
      "learning_rate": 1.9120714511946746e-06,
      "loss": 0.4314,
      "step": 1129
    },
    {
      "epoch": 2.221494102228047,
      "grad_norm": 0.19702021777629852,
      "learning_rate": 1.9030683557246815e-06,
      "loss": 0.4223,
      "step": 1130
    },
    {
      "epoch": 2.2234600262123196,
      "grad_norm": 0.20599868893623352,
      "learning_rate": 1.8940815215922609e-06,
      "loss": 0.4204,
      "step": 1131
    },
    {
      "epoch": 2.2254259501965925,
      "grad_norm": 0.21677346527576447,
      "learning_rate": 1.8851109959853885e-06,
      "loss": 0.4087,
      "step": 1132
    },
    {
      "epoch": 2.227391874180865,
      "grad_norm": 0.20800592005252838,
      "learning_rate": 1.8761568260064085e-06,
      "loss": 0.4462,
      "step": 1133
    },
    {
      "epoch": 2.229357798165138,
      "grad_norm": 0.2114061564207077,
      "learning_rate": 1.867219058671791e-06,
      "loss": 0.4209,
      "step": 1134
    },
    {
      "epoch": 2.23132372214941,
      "grad_norm": 0.1985040009021759,
      "learning_rate": 1.858297740911872e-06,
      "loss": 0.4145,
      "step": 1135
    },
    {
      "epoch": 2.2332896461336826,
      "grad_norm": 0.19175900518894196,
      "learning_rate": 1.8493929195706178e-06,
      "loss": 0.4304,
      "step": 1136
    },
    {
      "epoch": 2.2352555701179555,
      "grad_norm": 0.20461994409561157,
      "learning_rate": 1.8405046414053728e-06,
      "loss": 0.4237,
      "step": 1137
    },
    {
      "epoch": 2.237221494102228,
      "grad_norm": 0.19885732233524323,
      "learning_rate": 1.8316329530866228e-06,
      "loss": 0.4195,
      "step": 1138
    },
    {
      "epoch": 2.239187418086501,
      "grad_norm": 0.20640921592712402,
      "learning_rate": 1.822777901197738e-06,
      "loss": 0.4476,
      "step": 1139
    },
    {
      "epoch": 2.241153342070773,
      "grad_norm": 0.22110776603221893,
      "learning_rate": 1.8139395322347335e-06,
      "loss": 0.4341,
      "step": 1140
    },
    {
      "epoch": 2.243119266055046,
      "grad_norm": 0.19489331543445587,
      "learning_rate": 1.805117892606032e-06,
      "loss": 0.4084,
      "step": 1141
    },
    {
      "epoch": 2.2450851900393185,
      "grad_norm": 0.20213156938552856,
      "learning_rate": 1.7963130286322066e-06,
      "loss": 0.4177,
      "step": 1142
    },
    {
      "epoch": 2.247051114023591,
      "grad_norm": 0.2150600552558899,
      "learning_rate": 1.787524986545753e-06,
      "loss": 0.416,
      "step": 1143
    },
    {
      "epoch": 2.249017038007864,
      "grad_norm": 0.1949455887079239,
      "learning_rate": 1.7787538124908293e-06,
      "loss": 0.4329,
      "step": 1144
    },
    {
      "epoch": 2.250982961992136,
      "grad_norm": 0.2082454264163971,
      "learning_rate": 1.769999552523033e-06,
      "loss": 0.4101,
      "step": 1145
    },
    {
      "epoch": 2.252948885976409,
      "grad_norm": 0.20170274376869202,
      "learning_rate": 1.7612622526091406e-06,
      "loss": 0.4226,
      "step": 1146
    },
    {
      "epoch": 2.2549148099606815,
      "grad_norm": 0.21375320851802826,
      "learning_rate": 1.7525419586268816e-06,
      "loss": 0.421,
      "step": 1147
    },
    {
      "epoch": 2.2568807339449544,
      "grad_norm": 0.23509110510349274,
      "learning_rate": 1.7438387163646868e-06,
      "loss": 0.4237,
      "step": 1148
    },
    {
      "epoch": 2.258846657929227,
      "grad_norm": 0.20422513782978058,
      "learning_rate": 1.7351525715214512e-06,
      "loss": 0.4149,
      "step": 1149
    },
    {
      "epoch": 2.260812581913499,
      "grad_norm": 0.2060721516609192,
      "learning_rate": 1.7264835697063009e-06,
      "loss": 0.4165,
      "step": 1150
    },
    {
      "epoch": 2.262778505897772,
      "grad_norm": 0.21105261147022247,
      "learning_rate": 1.7178317564383396e-06,
      "loss": 0.4315,
      "step": 1151
    },
    {
      "epoch": 2.2647444298820445,
      "grad_norm": 0.20209087431430817,
      "learning_rate": 1.709197177146425e-06,
      "loss": 0.3992,
      "step": 1152
    },
    {
      "epoch": 2.2667103538663174,
      "grad_norm": 0.22701884806156158,
      "learning_rate": 1.700579877168918e-06,
      "loss": 0.422,
      "step": 1153
    },
    {
      "epoch": 2.26867627785059,
      "grad_norm": 0.21007543802261353,
      "learning_rate": 1.6919799017534505e-06,
      "loss": 0.4172,
      "step": 1154
    },
    {
      "epoch": 2.270642201834862,
      "grad_norm": 0.20358331501483917,
      "learning_rate": 1.6833972960566868e-06,
      "loss": 0.428,
      "step": 1155
    },
    {
      "epoch": 2.272608125819135,
      "grad_norm": 0.20329374074935913,
      "learning_rate": 1.6748321051440853e-06,
      "loss": 0.4287,
      "step": 1156
    },
    {
      "epoch": 2.2745740498034075,
      "grad_norm": 0.2196989506483078,
      "learning_rate": 1.6662843739896678e-06,
      "loss": 0.42,
      "step": 1157
    },
    {
      "epoch": 2.2765399737876804,
      "grad_norm": 0.22448624670505524,
      "learning_rate": 1.6577541474757712e-06,
      "loss": 0.4254,
      "step": 1158
    },
    {
      "epoch": 2.278505897771953,
      "grad_norm": 0.19686675071716309,
      "learning_rate": 1.6492414703928277e-06,
      "loss": 0.4204,
      "step": 1159
    },
    {
      "epoch": 2.280471821756225,
      "grad_norm": 0.20652048289775848,
      "learning_rate": 1.640746387439112e-06,
      "loss": 0.4214,
      "step": 1160
    },
    {
      "epoch": 2.282437745740498,
      "grad_norm": 0.21331384778022766,
      "learning_rate": 1.6322689432205252e-06,
      "loss": 0.4178,
      "step": 1161
    },
    {
      "epoch": 2.2844036697247705,
      "grad_norm": 0.2033684104681015,
      "learning_rate": 1.6238091822503426e-06,
      "loss": 0.4195,
      "step": 1162
    },
    {
      "epoch": 2.2863695937090434,
      "grad_norm": 0.20949597656726837,
      "learning_rate": 1.6153671489489925e-06,
      "loss": 0.4339,
      "step": 1163
    },
    {
      "epoch": 2.288335517693316,
      "grad_norm": 0.20488624274730682,
      "learning_rate": 1.6069428876438203e-06,
      "loss": 0.4191,
      "step": 1164
    },
    {
      "epoch": 2.2903014416775886,
      "grad_norm": 0.19698691368103027,
      "learning_rate": 1.5985364425688505e-06,
      "loss": 0.4094,
      "step": 1165
    },
    {
      "epoch": 2.292267365661861,
      "grad_norm": 0.20300184190273285,
      "learning_rate": 1.590147857864563e-06,
      "loss": 0.4183,
      "step": 1166
    },
    {
      "epoch": 2.2942332896461335,
      "grad_norm": 0.22306329011917114,
      "learning_rate": 1.5817771775776508e-06,
      "loss": 0.3995,
      "step": 1167
    },
    {
      "epoch": 2.2961992136304064,
      "grad_norm": 0.20931847393512726,
      "learning_rate": 1.5734244456608023e-06,
      "loss": 0.4185,
      "step": 1168
    },
    {
      "epoch": 2.2981651376146788,
      "grad_norm": 0.20768919587135315,
      "learning_rate": 1.5650897059724545e-06,
      "loss": 0.4257,
      "step": 1169
    },
    {
      "epoch": 2.3001310615989516,
      "grad_norm": 0.23441515862941742,
      "learning_rate": 1.5567730022765753e-06,
      "loss": 0.4214,
      "step": 1170
    },
    {
      "epoch": 2.302096985583224,
      "grad_norm": 0.20758725702762604,
      "learning_rate": 1.5484743782424317e-06,
      "loss": 0.421,
      "step": 1171
    },
    {
      "epoch": 2.304062909567497,
      "grad_norm": 0.20621424913406372,
      "learning_rate": 1.540193877444353e-06,
      "loss": 0.4153,
      "step": 1172
    },
    {
      "epoch": 2.3060288335517694,
      "grad_norm": 0.20830129086971283,
      "learning_rate": 1.5319315433615101e-06,
      "loss": 0.4295,
      "step": 1173
    },
    {
      "epoch": 2.3079947575360418,
      "grad_norm": 0.2129228115081787,
      "learning_rate": 1.5236874193776824e-06,
      "loss": 0.4108,
      "step": 1174
    },
    {
      "epoch": 2.3099606815203146,
      "grad_norm": 0.22020192444324493,
      "learning_rate": 1.515461548781036e-06,
      "loss": 0.4269,
      "step": 1175
    },
    {
      "epoch": 2.311926605504587,
      "grad_norm": 0.21290135383605957,
      "learning_rate": 1.5072539747638887e-06,
      "loss": 0.4373,
      "step": 1176
    },
    {
      "epoch": 2.31389252948886,
      "grad_norm": 0.2099660485982895,
      "learning_rate": 1.4990647404224856e-06,
      "loss": 0.4349,
      "step": 1177
    },
    {
      "epoch": 2.3158584534731324,
      "grad_norm": 0.19619333744049072,
      "learning_rate": 1.4908938887567797e-06,
      "loss": 0.3947,
      "step": 1178
    },
    {
      "epoch": 2.3178243774574048,
      "grad_norm": 0.21211563050746918,
      "learning_rate": 1.482741462670193e-06,
      "loss": 0.4304,
      "step": 1179
    },
    {
      "epoch": 2.3197903014416776,
      "grad_norm": 0.2202829271554947,
      "learning_rate": 1.4746075049694065e-06,
      "loss": 0.4376,
      "step": 1180
    },
    {
      "epoch": 2.32175622542595,
      "grad_norm": 0.20805491507053375,
      "learning_rate": 1.4664920583641196e-06,
      "loss": 0.4045,
      "step": 1181
    },
    {
      "epoch": 2.323722149410223,
      "grad_norm": 0.20760352909564972,
      "learning_rate": 1.4583951654668416e-06,
      "loss": 0.4068,
      "step": 1182
    },
    {
      "epoch": 2.3256880733944953,
      "grad_norm": 0.20598104596138,
      "learning_rate": 1.4503168687926533e-06,
      "loss": 0.4134,
      "step": 1183
    },
    {
      "epoch": 2.327653997378768,
      "grad_norm": 0.19814953207969666,
      "learning_rate": 1.4422572107589965e-06,
      "loss": 0.4028,
      "step": 1184
    },
    {
      "epoch": 2.3296199213630406,
      "grad_norm": 0.20932331681251526,
      "learning_rate": 1.434216233685441e-06,
      "loss": 0.4084,
      "step": 1185
    },
    {
      "epoch": 2.331585845347313,
      "grad_norm": 0.21522469818592072,
      "learning_rate": 1.426193979793467e-06,
      "loss": 0.416,
      "step": 1186
    },
    {
      "epoch": 2.333551769331586,
      "grad_norm": 0.20631645619869232,
      "learning_rate": 1.4181904912062482e-06,
      "loss": 0.4192,
      "step": 1187
    },
    {
      "epoch": 2.3355176933158583,
      "grad_norm": 0.2081977128982544,
      "learning_rate": 1.4102058099484188e-06,
      "loss": 0.4247,
      "step": 1188
    },
    {
      "epoch": 2.337483617300131,
      "grad_norm": 0.20724564790725708,
      "learning_rate": 1.4022399779458656e-06,
      "loss": 0.4513,
      "step": 1189
    },
    {
      "epoch": 2.3394495412844036,
      "grad_norm": 0.2072247415781021,
      "learning_rate": 1.3942930370254982e-06,
      "loss": 0.425,
      "step": 1190
    },
    {
      "epoch": 2.3414154652686765,
      "grad_norm": 0.22311511635780334,
      "learning_rate": 1.3863650289150338e-06,
      "loss": 0.4204,
      "step": 1191
    },
    {
      "epoch": 2.343381389252949,
      "grad_norm": 0.2175440937280655,
      "learning_rate": 1.3784559952427761e-06,
      "loss": 0.4231,
      "step": 1192
    },
    {
      "epoch": 2.3453473132372213,
      "grad_norm": 0.2023177295923233,
      "learning_rate": 1.3705659775374036e-06,
      "loss": 0.4218,
      "step": 1193
    },
    {
      "epoch": 2.347313237221494,
      "grad_norm": 0.1970416158437729,
      "learning_rate": 1.3626950172277398e-06,
      "loss": 0.4111,
      "step": 1194
    },
    {
      "epoch": 2.3492791612057666,
      "grad_norm": 0.20438626408576965,
      "learning_rate": 1.3548431556425423e-06,
      "loss": 0.4123,
      "step": 1195
    },
    {
      "epoch": 2.3512450851900395,
      "grad_norm": 0.19982948899269104,
      "learning_rate": 1.34701043401029e-06,
      "loss": 0.4094,
      "step": 1196
    },
    {
      "epoch": 2.353211009174312,
      "grad_norm": 0.21173065900802612,
      "learning_rate": 1.3391968934589573e-06,
      "loss": 0.4217,
      "step": 1197
    },
    {
      "epoch": 2.3551769331585843,
      "grad_norm": 0.2196841835975647,
      "learning_rate": 1.331402575015806e-06,
      "loss": 0.4232,
      "step": 1198
    },
    {
      "epoch": 2.357142857142857,
      "grad_norm": 0.2042812556028366,
      "learning_rate": 1.3236275196071641e-06,
      "loss": 0.4172,
      "step": 1199
    },
    {
      "epoch": 2.3591087811271296,
      "grad_norm": 0.21468588709831238,
      "learning_rate": 1.3158717680582128e-06,
      "loss": 0.4206,
      "step": 1200
    },
    {
      "epoch": 2.3610747051114025,
      "grad_norm": 0.2045600861310959,
      "learning_rate": 1.3081353610927777e-06,
      "loss": 0.4238,
      "step": 1201
    },
    {
      "epoch": 2.363040629095675,
      "grad_norm": 0.20586079359054565,
      "learning_rate": 1.3004183393331038e-06,
      "loss": 0.4233,
      "step": 1202
    },
    {
      "epoch": 2.3650065530799473,
      "grad_norm": 0.19963957369327545,
      "learning_rate": 1.292720743299654e-06,
      "loss": 0.404,
      "step": 1203
    },
    {
      "epoch": 2.36697247706422,
      "grad_norm": 0.22821871936321259,
      "learning_rate": 1.2850426134108856e-06,
      "loss": 0.4136,
      "step": 1204
    },
    {
      "epoch": 2.3689384010484926,
      "grad_norm": 0.20745152235031128,
      "learning_rate": 1.2773839899830487e-06,
      "loss": 0.4189,
      "step": 1205
    },
    {
      "epoch": 2.3709043250327655,
      "grad_norm": 0.18824845552444458,
      "learning_rate": 1.2697449132299649e-06,
      "loss": 0.4157,
      "step": 1206
    },
    {
      "epoch": 2.372870249017038,
      "grad_norm": 0.19603678584098816,
      "learning_rate": 1.2621254232628199e-06,
      "loss": 0.4237,
      "step": 1207
    },
    {
      "epoch": 2.374836173001311,
      "grad_norm": 0.2191622406244278,
      "learning_rate": 1.2545255600899587e-06,
      "loss": 0.4076,
      "step": 1208
    },
    {
      "epoch": 2.376802096985583,
      "grad_norm": 0.20817476511001587,
      "learning_rate": 1.2469453636166645e-06,
      "loss": 0.4118,
      "step": 1209
    },
    {
      "epoch": 2.378768020969856,
      "grad_norm": 0.21630269289016724,
      "learning_rate": 1.2393848736449559e-06,
      "loss": 0.4236,
      "step": 1210
    },
    {
      "epoch": 2.3807339449541285,
      "grad_norm": 0.19972224533557892,
      "learning_rate": 1.2318441298733796e-06,
      "loss": 0.4086,
      "step": 1211
    },
    {
      "epoch": 2.382699868938401,
      "grad_norm": 0.1989041119813919,
      "learning_rate": 1.224323171896797e-06,
      "loss": 0.4269,
      "step": 1212
    },
    {
      "epoch": 2.3846657929226738,
      "grad_norm": 0.20862899720668793,
      "learning_rate": 1.2168220392061775e-06,
      "loss": 0.4347,
      "step": 1213
    },
    {
      "epoch": 2.386631716906946,
      "grad_norm": 0.19993065297603607,
      "learning_rate": 1.2093407711883926e-06,
      "loss": 0.4258,
      "step": 1214
    },
    {
      "epoch": 2.388597640891219,
      "grad_norm": 0.20331786572933197,
      "learning_rate": 1.201879407126012e-06,
      "loss": 0.4497,
      "step": 1215
    },
    {
      "epoch": 2.3905635648754915,
      "grad_norm": 0.19923487305641174,
      "learning_rate": 1.1944379861970884e-06,
      "loss": 0.4437,
      "step": 1216
    },
    {
      "epoch": 2.392529488859764,
      "grad_norm": 0.20787520706653595,
      "learning_rate": 1.187016547474963e-06,
      "loss": 0.427,
      "step": 1217
    },
    {
      "epoch": 2.3944954128440368,
      "grad_norm": 0.20175479352474213,
      "learning_rate": 1.1796151299280483e-06,
      "loss": 0.4245,
      "step": 1218
    },
    {
      "epoch": 2.396461336828309,
      "grad_norm": 0.194331094622612,
      "learning_rate": 1.1722337724196365e-06,
      "loss": 0.4293,
      "step": 1219
    },
    {
      "epoch": 2.398427260812582,
      "grad_norm": 0.19052304327487946,
      "learning_rate": 1.1648725137076822e-06,
      "loss": 0.4242,
      "step": 1220
    },
    {
      "epoch": 2.4003931847968545,
      "grad_norm": 0.20737259089946747,
      "learning_rate": 1.1575313924446123e-06,
      "loss": 0.4301,
      "step": 1221
    },
    {
      "epoch": 2.402359108781127,
      "grad_norm": 0.21076960861682892,
      "learning_rate": 1.1502104471771109e-06,
      "loss": 0.4299,
      "step": 1222
    },
    {
      "epoch": 2.4043250327653998,
      "grad_norm": 0.19797001779079437,
      "learning_rate": 1.1429097163459219e-06,
      "loss": 0.4172,
      "step": 1223
    },
    {
      "epoch": 2.406290956749672,
      "grad_norm": 0.197958305478096,
      "learning_rate": 1.1356292382856531e-06,
      "loss": 0.4286,
      "step": 1224
    },
    {
      "epoch": 2.408256880733945,
      "grad_norm": 0.2084892839193344,
      "learning_rate": 1.1283690512245621e-06,
      "loss": 0.4317,
      "step": 1225
    },
    {
      "epoch": 2.4102228047182175,
      "grad_norm": 0.2041456699371338,
      "learning_rate": 1.1211291932843687e-06,
      "loss": 0.4218,
      "step": 1226
    },
    {
      "epoch": 2.4121887287024903,
      "grad_norm": 0.1887100487947464,
      "learning_rate": 1.113909702480046e-06,
      "loss": 0.4307,
      "step": 1227
    },
    {
      "epoch": 2.4141546526867628,
      "grad_norm": 0.21170374751091003,
      "learning_rate": 1.1067106167196217e-06,
      "loss": 0.4119,
      "step": 1228
    },
    {
      "epoch": 2.4161205766710356,
      "grad_norm": 0.2029750794172287,
      "learning_rate": 1.0995319738039855e-06,
      "loss": 0.4309,
      "step": 1229
    },
    {
      "epoch": 2.418086500655308,
      "grad_norm": 0.21280771493911743,
      "learning_rate": 1.0923738114266824e-06,
      "loss": 0.4182,
      "step": 1230
    },
    {
      "epoch": 2.4200524246395805,
      "grad_norm": 0.19209042191505432,
      "learning_rate": 1.08523616717372e-06,
      "loss": 0.417,
      "step": 1231
    },
    {
      "epoch": 2.4220183486238533,
      "grad_norm": 0.19855521619319916,
      "learning_rate": 1.078119078523367e-06,
      "loss": 0.4164,
      "step": 1232
    },
    {
      "epoch": 2.4239842726081258,
      "grad_norm": 0.2216549813747406,
      "learning_rate": 1.0710225828459642e-06,
      "loss": 0.422,
      "step": 1233
    },
    {
      "epoch": 2.4259501965923986,
      "grad_norm": 0.19830770790576935,
      "learning_rate": 1.0639467174037165e-06,
      "loss": 0.4192,
      "step": 1234
    },
    {
      "epoch": 2.427916120576671,
      "grad_norm": 0.20297428965568542,
      "learning_rate": 1.0568915193505103e-06,
      "loss": 0.4325,
      "step": 1235
    },
    {
      "epoch": 2.4298820445609435,
      "grad_norm": 0.2170720249414444,
      "learning_rate": 1.0498570257317075e-06,
      "loss": 0.4345,
      "step": 1236
    },
    {
      "epoch": 2.4318479685452163,
      "grad_norm": 0.20226174592971802,
      "learning_rate": 1.0428432734839543e-06,
      "loss": 0.4273,
      "step": 1237
    },
    {
      "epoch": 2.4338138925294888,
      "grad_norm": 0.19869181513786316,
      "learning_rate": 1.0358502994349945e-06,
      "loss": 0.4268,
      "step": 1238
    },
    {
      "epoch": 2.4357798165137616,
      "grad_norm": 0.20573928952217102,
      "learning_rate": 1.028878140303462e-06,
      "loss": 0.4256,
      "step": 1239
    },
    {
      "epoch": 2.437745740498034,
      "grad_norm": 0.1978793442249298,
      "learning_rate": 1.0219268326987035e-06,
      "loss": 0.4215,
      "step": 1240
    },
    {
      "epoch": 2.4397116644823065,
      "grad_norm": 0.19740813970565796,
      "learning_rate": 1.0149964131205724e-06,
      "loss": 0.415,
      "step": 1241
    },
    {
      "epoch": 2.4416775884665793,
      "grad_norm": 0.19393078982830048,
      "learning_rate": 1.008086917959249e-06,
      "loss": 0.4194,
      "step": 1242
    },
    {
      "epoch": 2.4436435124508518,
      "grad_norm": 0.20559275150299072,
      "learning_rate": 1.0011983834950389e-06,
      "loss": 0.4165,
      "step": 1243
    },
    {
      "epoch": 2.4456094364351246,
      "grad_norm": 0.19457489252090454,
      "learning_rate": 9.943308458981892e-07,
      "loss": 0.409,
      "step": 1244
    },
    {
      "epoch": 2.447575360419397,
      "grad_norm": 0.19760730862617493,
      "learning_rate": 9.874843412286994e-07,
      "loss": 0.4249,
      "step": 1245
    },
    {
      "epoch": 2.44954128440367,
      "grad_norm": 0.19084718823432922,
      "learning_rate": 9.806589054361255e-07,
      "loss": 0.4205,
      "step": 1246
    },
    {
      "epoch": 2.4515072083879423,
      "grad_norm": 0.2036636471748352,
      "learning_rate": 9.738545743593991e-07,
      "loss": 0.416,
      "step": 1247
    },
    {
      "epoch": 2.4534731323722148,
      "grad_norm": 0.1984458714723587,
      "learning_rate": 9.670713837266322e-07,
      "loss": 0.4235,
      "step": 1248
    },
    {
      "epoch": 2.4554390563564876,
      "grad_norm": 0.2098272740840912,
      "learning_rate": 9.603093691549348e-07,
      "loss": 0.4344,
      "step": 1249
    },
    {
      "epoch": 2.45740498034076,
      "grad_norm": 0.206814706325531,
      "learning_rate": 9.535685661502248e-07,
      "loss": 0.4232,
      "step": 1250
    },
    {
      "epoch": 2.459370904325033,
      "grad_norm": 0.2012605369091034,
      "learning_rate": 9.46849010107041e-07,
      "loss": 0.4076,
      "step": 1251
    },
    {
      "epoch": 2.4613368283093053,
      "grad_norm": 0.204511821269989,
      "learning_rate": 9.401507363083634e-07,
      "loss": 0.4219,
      "step": 1252
    },
    {
      "epoch": 2.463302752293578,
      "grad_norm": 0.19177298247814178,
      "learning_rate": 9.334737799254195e-07,
      "loss": 0.4103,
      "step": 1253
    },
    {
      "epoch": 2.4652686762778506,
      "grad_norm": 0.20843657851219177,
      "learning_rate": 9.26818176017506e-07,
      "loss": 0.4244,
      "step": 1254
    },
    {
      "epoch": 2.467234600262123,
      "grad_norm": 0.21537087857723236,
      "learning_rate": 9.201839595317991e-07,
      "loss": 0.4357,
      "step": 1255
    },
    {
      "epoch": 2.469200524246396,
      "grad_norm": 0.19909898936748505,
      "learning_rate": 9.135711653031781e-07,
      "loss": 0.4234,
      "step": 1256
    },
    {
      "epoch": 2.4711664482306683,
      "grad_norm": 0.19548694789409637,
      "learning_rate": 9.069798280540348e-07,
      "loss": 0.4144,
      "step": 1257
    },
    {
      "epoch": 2.473132372214941,
      "grad_norm": 0.2011128067970276,
      "learning_rate": 9.004099823940982e-07,
      "loss": 0.4165,
      "step": 1258
    },
    {
      "epoch": 2.4750982961992136,
      "grad_norm": 0.19573193788528442,
      "learning_rate": 8.938616628202478e-07,
      "loss": 0.4129,
      "step": 1259
    },
    {
      "epoch": 2.477064220183486,
      "grad_norm": 0.19089443981647491,
      "learning_rate": 8.87334903716332e-07,
      "loss": 0.4006,
      "step": 1260
    },
    {
      "epoch": 2.479030144167759,
      "grad_norm": 0.19790777564048767,
      "learning_rate": 8.808297393529946e-07,
      "loss": 0.4227,
      "step": 1261
    },
    {
      "epoch": 2.4809960681520313,
      "grad_norm": 0.21096652746200562,
      "learning_rate": 8.743462038874856e-07,
      "loss": 0.4232,
      "step": 1262
    },
    {
      "epoch": 2.482961992136304,
      "grad_norm": 0.20578713715076447,
      "learning_rate": 8.678843313634894e-07,
      "loss": 0.4274,
      "step": 1263
    },
    {
      "epoch": 2.4849279161205766,
      "grad_norm": 0.20468361675739288,
      "learning_rate": 8.614441557109388e-07,
      "loss": 0.4269,
      "step": 1264
    },
    {
      "epoch": 2.486893840104849,
      "grad_norm": 0.19431470334529877,
      "learning_rate": 8.550257107458471e-07,
      "loss": 0.4174,
      "step": 1265
    },
    {
      "epoch": 2.488859764089122,
      "grad_norm": 0.1942060887813568,
      "learning_rate": 8.486290301701183e-07,
      "loss": 0.4107,
      "step": 1266
    },
    {
      "epoch": 2.4908256880733943,
      "grad_norm": 0.20655293762683868,
      "learning_rate": 8.422541475713785e-07,
      "loss": 0.4332,
      "step": 1267
    },
    {
      "epoch": 2.492791612057667,
      "grad_norm": 0.19286088645458221,
      "learning_rate": 8.35901096422797e-07,
      "loss": 0.4246,
      "step": 1268
    },
    {
      "epoch": 2.4947575360419396,
      "grad_norm": 0.1953059434890747,
      "learning_rate": 8.295699100829124e-07,
      "loss": 0.4258,
      "step": 1269
    },
    {
      "epoch": 2.4967234600262125,
      "grad_norm": 0.21908029913902283,
      "learning_rate": 8.232606217954536e-07,
      "loss": 0.4419,
      "step": 1270
    },
    {
      "epoch": 2.498689384010485,
      "grad_norm": 0.1864355057477951,
      "learning_rate": 8.169732646891665e-07,
      "loss": 0.4069,
      "step": 1271
    },
    {
      "epoch": 2.5006553079947578,
      "grad_norm": 0.19375306367874146,
      "learning_rate": 8.107078717776457e-07,
      "loss": 0.4138,
      "step": 1272
    },
    {
      "epoch": 2.50262123197903,
      "grad_norm": 0.18619294464588165,
      "learning_rate": 8.044644759591519e-07,
      "loss": 0.4108,
      "step": 1273
    },
    {
      "epoch": 2.5045871559633026,
      "grad_norm": 0.20051321387290955,
      "learning_rate": 7.982431100164439e-07,
      "loss": 0.4241,
      "step": 1274
    },
    {
      "epoch": 2.5065530799475755,
      "grad_norm": 0.195013627409935,
      "learning_rate": 7.920438066166097e-07,
      "loss": 0.4239,
      "step": 1275
    },
    {
      "epoch": 2.508519003931848,
      "grad_norm": 0.19501401484012604,
      "learning_rate": 7.858665983108871e-07,
      "loss": 0.4446,
      "step": 1276
    },
    {
      "epoch": 2.5104849279161208,
      "grad_norm": 0.190773144364357,
      "learning_rate": 7.797115175345021e-07,
      "loss": 0.4097,
      "step": 1277
    },
    {
      "epoch": 2.512450851900393,
      "grad_norm": 0.18783928453922272,
      "learning_rate": 7.735785966064885e-07,
      "loss": 0.4115,
      "step": 1278
    },
    {
      "epoch": 2.5144167758846656,
      "grad_norm": 0.1892012506723404,
      "learning_rate": 7.674678677295277e-07,
      "loss": 0.4126,
      "step": 1279
    },
    {
      "epoch": 2.5163826998689385,
      "grad_norm": 0.20406071841716766,
      "learning_rate": 7.613793629897732e-07,
      "loss": 0.439,
      "step": 1280
    },
    {
      "epoch": 2.518348623853211,
      "grad_norm": 0.20593975484371185,
      "learning_rate": 7.553131143566822e-07,
      "loss": 0.4392,
      "step": 1281
    },
    {
      "epoch": 2.5203145478374838,
      "grad_norm": 0.19002759456634521,
      "learning_rate": 7.492691536828556e-07,
      "loss": 0.4296,
      "step": 1282
    },
    {
      "epoch": 2.522280471821756,
      "grad_norm": 0.19039657711982727,
      "learning_rate": 7.432475127038591e-07,
      "loss": 0.4028,
      "step": 1283
    },
    {
      "epoch": 2.5242463958060286,
      "grad_norm": 0.1925535649061203,
      "learning_rate": 7.372482230380657e-07,
      "loss": 0.4276,
      "step": 1284
    },
    {
      "epoch": 2.5262123197903015,
      "grad_norm": 0.19870179891586304,
      "learning_rate": 7.312713161864854e-07,
      "loss": 0.4313,
      "step": 1285
    },
    {
      "epoch": 2.528178243774574,
      "grad_norm": 0.20361679792404175,
      "learning_rate": 7.253168235325992e-07,
      "loss": 0.4296,
      "step": 1286
    },
    {
      "epoch": 2.5301441677588468,
      "grad_norm": 0.19246189296245575,
      "learning_rate": 7.193847763421991e-07,
      "loss": 0.4178,
      "step": 1287
    },
    {
      "epoch": 2.532110091743119,
      "grad_norm": 0.18752935528755188,
      "learning_rate": 7.134752057632188e-07,
      "loss": 0.4044,
      "step": 1288
    },
    {
      "epoch": 2.5340760157273916,
      "grad_norm": 0.1949397772550583,
      "learning_rate": 7.07588142825571e-07,
      "loss": 0.4116,
      "step": 1289
    },
    {
      "epoch": 2.5360419397116645,
      "grad_norm": 0.19233138859272003,
      "learning_rate": 7.017236184409859e-07,
      "loss": 0.4207,
      "step": 1290
    },
    {
      "epoch": 2.5380078636959373,
      "grad_norm": 0.1937728077173233,
      "learning_rate": 6.95881663402852e-07,
      "loss": 0.3934,
      "step": 1291
    },
    {
      "epoch": 2.5399737876802098,
      "grad_norm": 0.19518445432186127,
      "learning_rate": 6.900623083860453e-07,
      "loss": 0.4228,
      "step": 1292
    },
    {
      "epoch": 2.541939711664482,
      "grad_norm": 0.19948603212833405,
      "learning_rate": 6.842655839467787e-07,
      "loss": 0.4223,
      "step": 1293
    },
    {
      "epoch": 2.543905635648755,
      "grad_norm": 0.1915956288576126,
      "learning_rate": 6.78491520522433e-07,
      "loss": 0.4169,
      "step": 1294
    },
    {
      "epoch": 2.5458715596330275,
      "grad_norm": 0.19227120280265808,
      "learning_rate": 6.72740148431405e-07,
      "loss": 0.4069,
      "step": 1295
    },
    {
      "epoch": 2.5478374836173003,
      "grad_norm": 0.21147392690181732,
      "learning_rate": 6.670114978729392e-07,
      "loss": 0.4388,
      "step": 1296
    },
    {
      "epoch": 2.5498034076015728,
      "grad_norm": 0.19551649689674377,
      "learning_rate": 6.613055989269762e-07,
      "loss": 0.4349,
      "step": 1297
    },
    {
      "epoch": 2.551769331585845,
      "grad_norm": 0.1957281529903412,
      "learning_rate": 6.556224815539946e-07,
      "loss": 0.4079,
      "step": 1298
    },
    {
      "epoch": 2.553735255570118,
      "grad_norm": 0.19849686324596405,
      "learning_rate": 6.499621755948487e-07,
      "loss": 0.4161,
      "step": 1299
    },
    {
      "epoch": 2.5557011795543905,
      "grad_norm": 0.19393165409564972,
      "learning_rate": 6.443247107706174e-07,
      "loss": 0.4336,
      "step": 1300
    },
    {
      "epoch": 2.5576671035386633,
      "grad_norm": 0.21561844646930695,
      "learning_rate": 6.387101166824422e-07,
      "loss": 0.4394,
      "step": 1301
    },
    {
      "epoch": 2.5596330275229358,
      "grad_norm": 0.1955285668373108,
      "learning_rate": 6.331184228113801e-07,
      "loss": 0.4319,
      "step": 1302
    },
    {
      "epoch": 2.561598951507208,
      "grad_norm": 0.21019408106803894,
      "learning_rate": 6.2754965851824e-07,
      "loss": 0.4322,
      "step": 1303
    },
    {
      "epoch": 2.563564875491481,
      "grad_norm": 0.19174997508525848,
      "learning_rate": 6.220038530434319e-07,
      "loss": 0.4167,
      "step": 1304
    },
    {
      "epoch": 2.5655307994757535,
      "grad_norm": 0.2004367858171463,
      "learning_rate": 6.164810355068179e-07,
      "loss": 0.4347,
      "step": 1305
    },
    {
      "epoch": 2.5674967234600263,
      "grad_norm": 0.211960107088089,
      "learning_rate": 6.109812349075517e-07,
      "loss": 0.4186,
      "step": 1306
    },
    {
      "epoch": 2.5694626474442988,
      "grad_norm": 0.20421504974365234,
      "learning_rate": 6.055044801239313e-07,
      "loss": 0.4085,
      "step": 1307
    },
    {
      "epoch": 2.571428571428571,
      "grad_norm": 0.19226616621017456,
      "learning_rate": 6.000507999132444e-07,
      "loss": 0.4308,
      "step": 1308
    },
    {
      "epoch": 2.573394495412844,
      "grad_norm": 0.1891256868839264,
      "learning_rate": 5.946202229116227e-07,
      "loss": 0.4318,
      "step": 1309
    },
    {
      "epoch": 2.575360419397117,
      "grad_norm": 0.189310222864151,
      "learning_rate": 5.892127776338841e-07,
      "loss": 0.3998,
      "step": 1310
    },
    {
      "epoch": 2.5773263433813893,
      "grad_norm": 0.19134820997714996,
      "learning_rate": 5.838284924733866e-07,
      "loss": 0.4302,
      "step": 1311
    },
    {
      "epoch": 2.5792922673656618,
      "grad_norm": 0.19616124033927917,
      "learning_rate": 5.784673957018833e-07,
      "loss": 0.4245,
      "step": 1312
    },
    {
      "epoch": 2.5812581913499346,
      "grad_norm": 0.18709275126457214,
      "learning_rate": 5.731295154693644e-07,
      "loss": 0.4081,
      "step": 1313
    },
    {
      "epoch": 2.583224115334207,
      "grad_norm": 0.1914079487323761,
      "learning_rate": 5.678148798039213e-07,
      "loss": 0.4051,
      "step": 1314
    },
    {
      "epoch": 2.58519003931848,
      "grad_norm": 0.19706860184669495,
      "learning_rate": 5.625235166115855e-07,
      "loss": 0.4243,
      "step": 1315
    },
    {
      "epoch": 2.5871559633027523,
      "grad_norm": 0.19415414333343506,
      "learning_rate": 5.572554536761976e-07,
      "loss": 0.4348,
      "step": 1316
    },
    {
      "epoch": 2.5891218872870247,
      "grad_norm": 0.1907108873128891,
      "learning_rate": 5.520107186592477e-07,
      "loss": 0.4183,
      "step": 1317
    },
    {
      "epoch": 2.5910878112712976,
      "grad_norm": 0.20467859506607056,
      "learning_rate": 5.467893390997369e-07,
      "loss": 0.4101,
      "step": 1318
    },
    {
      "epoch": 2.59305373525557,
      "grad_norm": 0.19916951656341553,
      "learning_rate": 5.41591342414034e-07,
      "loss": 0.4148,
      "step": 1319
    },
    {
      "epoch": 2.595019659239843,
      "grad_norm": 0.1953822374343872,
      "learning_rate": 5.364167558957267e-07,
      "loss": 0.4188,
      "step": 1320
    },
    {
      "epoch": 2.5969855832241153,
      "grad_norm": 0.1931547224521637,
      "learning_rate": 5.312656067154831e-07,
      "loss": 0.4185,
      "step": 1321
    },
    {
      "epoch": 2.5989515072083877,
      "grad_norm": 0.19639557600021362,
      "learning_rate": 5.261379219209045e-07,
      "loss": 0.4142,
      "step": 1322
    },
    {
      "epoch": 2.6009174311926606,
      "grad_norm": 0.20015166699886322,
      "learning_rate": 5.210337284363876e-07,
      "loss": 0.4174,
      "step": 1323
    },
    {
      "epoch": 2.602883355176933,
      "grad_norm": 0.20684877038002014,
      "learning_rate": 5.159530530629803e-07,
      "loss": 0.4344,
      "step": 1324
    },
    {
      "epoch": 2.604849279161206,
      "grad_norm": 0.1928563117980957,
      "learning_rate": 5.108959224782406e-07,
      "loss": 0.4256,
      "step": 1325
    },
    {
      "epoch": 2.6068152031454783,
      "grad_norm": 0.19201450049877167,
      "learning_rate": 5.058623632361004e-07,
      "loss": 0.4047,
      "step": 1326
    },
    {
      "epoch": 2.6087811271297507,
      "grad_norm": 0.20671480894088745,
      "learning_rate": 5.008524017667204e-07,
      "loss": 0.4457,
      "step": 1327
    },
    {
      "epoch": 2.6107470511140236,
      "grad_norm": 0.19633173942565918,
      "learning_rate": 4.958660643763574e-07,
      "loss": 0.4013,
      "step": 1328
    },
    {
      "epoch": 2.6127129750982965,
      "grad_norm": 0.2064533680677414,
      "learning_rate": 4.909033772472204e-07,
      "loss": 0.4332,
      "step": 1329
    },
    {
      "epoch": 2.614678899082569,
      "grad_norm": 0.21506528556346893,
      "learning_rate": 4.859643664373387e-07,
      "loss": 0.4516,
      "step": 1330
    },
    {
      "epoch": 2.6166448230668413,
      "grad_norm": 0.19731992483139038,
      "learning_rate": 4.810490578804195e-07,
      "loss": 0.4359,
      "step": 1331
    },
    {
      "epoch": 2.618610747051114,
      "grad_norm": 0.18773064017295837,
      "learning_rate": 4.7615747738571636e-07,
      "loss": 0.4152,
      "step": 1332
    },
    {
      "epoch": 2.6205766710353866,
      "grad_norm": 0.1858416497707367,
      "learning_rate": 4.7128965063789054e-07,
      "loss": 0.4246,
      "step": 1333
    },
    {
      "epoch": 2.6225425950196595,
      "grad_norm": 0.18619361519813538,
      "learning_rate": 4.664456031968773e-07,
      "loss": 0.4199,
      "step": 1334
    },
    {
      "epoch": 2.624508519003932,
      "grad_norm": 0.19577224552631378,
      "learning_rate": 4.6162536049775387e-07,
      "loss": 0.4125,
      "step": 1335
    },
    {
      "epoch": 2.6264744429882043,
      "grad_norm": 0.1945699155330658,
      "learning_rate": 4.5682894785059995e-07,
      "loss": 0.4206,
      "step": 1336
    },
    {
      "epoch": 2.628440366972477,
      "grad_norm": 0.19429698586463928,
      "learning_rate": 4.520563904403735e-07,
      "loss": 0.4014,
      "step": 1337
    },
    {
      "epoch": 2.6304062909567496,
      "grad_norm": 0.2008766531944275,
      "learning_rate": 4.473077133267684e-07,
      "loss": 0.426,
      "step": 1338
    },
    {
      "epoch": 2.6323722149410225,
      "grad_norm": 0.20195744931697845,
      "learning_rate": 4.42582941444093e-07,
      "loss": 0.4237,
      "step": 1339
    },
    {
      "epoch": 2.634338138925295,
      "grad_norm": 0.189142107963562,
      "learning_rate": 4.378820996011307e-07,
      "loss": 0.4153,
      "step": 1340
    },
    {
      "epoch": 2.6363040629095673,
      "grad_norm": 0.18883055448532104,
      "learning_rate": 4.3320521248101487e-07,
      "loss": 0.3998,
      "step": 1341
    },
    {
      "epoch": 2.63826998689384,
      "grad_norm": 0.18498775362968445,
      "learning_rate": 4.2855230464109775e-07,
      "loss": 0.4135,
      "step": 1342
    },
    {
      "epoch": 2.6402359108781126,
      "grad_norm": 0.19491244852542877,
      "learning_rate": 4.239234005128212e-07,
      "loss": 0.4074,
      "step": 1343
    },
    {
      "epoch": 2.6422018348623855,
      "grad_norm": 0.20609928667545319,
      "learning_rate": 4.193185244015879e-07,
      "loss": 0.4149,
      "step": 1344
    },
    {
      "epoch": 2.644167758846658,
      "grad_norm": 0.1882026344537735,
      "learning_rate": 4.1473770048663487e-07,
      "loss": 0.416,
      "step": 1345
    },
    {
      "epoch": 2.6461336828309303,
      "grad_norm": 0.19239036738872528,
      "learning_rate": 4.1018095282090775e-07,
      "loss": 0.434,
      "step": 1346
    },
    {
      "epoch": 2.648099606815203,
      "grad_norm": 0.188425675034523,
      "learning_rate": 4.0564830533093014e-07,
      "loss": 0.4144,
      "step": 1347
    },
    {
      "epoch": 2.6500655307994756,
      "grad_norm": 0.1889129877090454,
      "learning_rate": 4.011397818166818e-07,
      "loss": 0.41,
      "step": 1348
    },
    {
      "epoch": 2.6520314547837485,
      "grad_norm": 0.19384776055812836,
      "learning_rate": 3.9665540595147376e-07,
      "loss": 0.4265,
      "step": 1349
    },
    {
      "epoch": 2.653997378768021,
      "grad_norm": 0.18682250380516052,
      "learning_rate": 3.9219520128182087e-07,
      "loss": 0.4165,
      "step": 1350
    },
    {
      "epoch": 2.6559633027522933,
      "grad_norm": 0.19491790235042572,
      "learning_rate": 3.877591912273215e-07,
      "loss": 0.42,
      "step": 1351
    },
    {
      "epoch": 2.657929226736566,
      "grad_norm": 0.19894134998321533,
      "learning_rate": 3.8334739908053196e-07,
      "loss": 0.4155,
      "step": 1352
    },
    {
      "epoch": 2.659895150720839,
      "grad_norm": 0.18585951626300812,
      "learning_rate": 3.789598480068479e-07,
      "loss": 0.4262,
      "step": 1353
    },
    {
      "epoch": 2.6618610747051115,
      "grad_norm": 0.19301895797252655,
      "learning_rate": 3.745965610443769e-07,
      "loss": 0.415,
      "step": 1354
    },
    {
      "epoch": 2.663826998689384,
      "grad_norm": 0.1924465447664261,
      "learning_rate": 3.702575611038217e-07,
      "loss": 0.4182,
      "step": 1355
    },
    {
      "epoch": 2.6657929226736568,
      "grad_norm": 0.18831956386566162,
      "learning_rate": 3.659428709683621e-07,
      "loss": 0.4143,
      "step": 1356
    },
    {
      "epoch": 2.667758846657929,
      "grad_norm": 0.19500558078289032,
      "learning_rate": 3.616525132935267e-07,
      "loss": 0.4196,
      "step": 1357
    },
    {
      "epoch": 2.669724770642202,
      "grad_norm": 0.1873614639043808,
      "learning_rate": 3.573865106070851e-07,
      "loss": 0.4095,
      "step": 1358
    },
    {
      "epoch": 2.6716906946264745,
      "grad_norm": 0.18997088074684143,
      "learning_rate": 3.531448853089192e-07,
      "loss": 0.4059,
      "step": 1359
    },
    {
      "epoch": 2.673656618610747,
      "grad_norm": 0.19585131108760834,
      "learning_rate": 3.489276596709146e-07,
      "loss": 0.4334,
      "step": 1360
    },
    {
      "epoch": 2.6756225425950197,
      "grad_norm": 0.19634681940078735,
      "learning_rate": 3.4473485583683576e-07,
      "loss": 0.4085,
      "step": 1361
    },
    {
      "epoch": 2.677588466579292,
      "grad_norm": 0.18681412935256958,
      "learning_rate": 3.40566495822216e-07,
      "loss": 0.4217,
      "step": 1362
    },
    {
      "epoch": 2.679554390563565,
      "grad_norm": 0.20143257081508636,
      "learning_rate": 3.364226015142369e-07,
      "loss": 0.4248,
      "step": 1363
    },
    {
      "epoch": 2.6815203145478375,
      "grad_norm": 0.2103082686662674,
      "learning_rate": 3.323031946716182e-07,
      "loss": 0.4346,
      "step": 1364
    },
    {
      "epoch": 2.68348623853211,
      "grad_norm": 0.19275066256523132,
      "learning_rate": 3.2820829692449984e-07,
      "loss": 0.4544,
      "step": 1365
    },
    {
      "epoch": 2.6854521625163827,
      "grad_norm": 0.18487761914730072,
      "learning_rate": 3.2413792977432856e-07,
      "loss": 0.3919,
      "step": 1366
    },
    {
      "epoch": 2.687418086500655,
      "grad_norm": 0.18378794193267822,
      "learning_rate": 3.2009211459374913e-07,
      "loss": 0.4259,
      "step": 1367
    },
    {
      "epoch": 2.689384010484928,
      "grad_norm": 0.18482978641986847,
      "learning_rate": 3.160708726264855e-07,
      "loss": 0.4222,
      "step": 1368
    },
    {
      "epoch": 2.6913499344692005,
      "grad_norm": 0.19463540613651276,
      "learning_rate": 3.1207422498723663e-07,
      "loss": 0.4228,
      "step": 1369
    },
    {
      "epoch": 2.693315858453473,
      "grad_norm": 0.19638504087924957,
      "learning_rate": 3.081021926615585e-07,
      "loss": 0.4058,
      "step": 1370
    },
    {
      "epoch": 2.6952817824377457,
      "grad_norm": 0.19137264788150787,
      "learning_rate": 3.0415479650575783e-07,
      "loss": 0.4124,
      "step": 1371
    },
    {
      "epoch": 2.6972477064220186,
      "grad_norm": 0.1765514612197876,
      "learning_rate": 3.0023205724678483e-07,
      "loss": 0.405,
      "step": 1372
    },
    {
      "epoch": 2.699213630406291,
      "grad_norm": 0.1941332072019577,
      "learning_rate": 2.9633399548211707e-07,
      "loss": 0.4081,
      "step": 1373
    },
    {
      "epoch": 2.7011795543905635,
      "grad_norm": 0.19398586452007294,
      "learning_rate": 2.9246063167965963e-07,
      "loss": 0.4279,
      "step": 1374
    },
    {
      "epoch": 2.7031454783748363,
      "grad_norm": 0.19305099546909332,
      "learning_rate": 2.8861198617763154e-07,
      "loss": 0.4139,
      "step": 1375
    },
    {
      "epoch": 2.7051114023591087,
      "grad_norm": 0.18326827883720398,
      "learning_rate": 2.8478807918446315e-07,
      "loss": 0.4204,
      "step": 1376
    },
    {
      "epoch": 2.7070773263433816,
      "grad_norm": 0.1823471635580063,
      "learning_rate": 2.809889307786856e-07,
      "loss": 0.4086,
      "step": 1377
    },
    {
      "epoch": 2.709043250327654,
      "grad_norm": 0.18850870430469513,
      "learning_rate": 2.7721456090882893e-07,
      "loss": 0.4252,
      "step": 1378
    },
    {
      "epoch": 2.7110091743119265,
      "grad_norm": 0.18475760519504547,
      "learning_rate": 2.734649893933178e-07,
      "loss": 0.4186,
      "step": 1379
    },
    {
      "epoch": 2.7129750982961993,
      "grad_norm": 0.19123856723308563,
      "learning_rate": 2.697402359203638e-07,
      "loss": 0.4288,
      "step": 1380
    },
    {
      "epoch": 2.7149410222804717,
      "grad_norm": 0.1787601113319397,
      "learning_rate": 2.6604032004786563e-07,
      "loss": 0.4028,
      "step": 1381
    },
    {
      "epoch": 2.7169069462647446,
      "grad_norm": 0.1857636570930481,
      "learning_rate": 2.6236526120330395e-07,
      "loss": 0.4107,
      "step": 1382
    },
    {
      "epoch": 2.718872870249017,
      "grad_norm": 0.18619926273822784,
      "learning_rate": 2.587150786836407e-07,
      "loss": 0.4234,
      "step": 1383
    },
    {
      "epoch": 2.7208387942332894,
      "grad_norm": 0.1910834163427353,
      "learning_rate": 2.550897916552181e-07,
      "loss": 0.4183,
      "step": 1384
    },
    {
      "epoch": 2.7228047182175623,
      "grad_norm": 0.18822459876537323,
      "learning_rate": 2.51489419153656e-07,
      "loss": 0.3991,
      "step": 1385
    },
    {
      "epoch": 2.7247706422018347,
      "grad_norm": 0.19424988329410553,
      "learning_rate": 2.4791398008375545e-07,
      "loss": 0.4327,
      "step": 1386
    },
    {
      "epoch": 2.7267365661861076,
      "grad_norm": 0.2632913887500763,
      "learning_rate": 2.4436349321939447e-07,
      "loss": 0.4187,
      "step": 1387
    },
    {
      "epoch": 2.72870249017038,
      "grad_norm": 0.18020807206630707,
      "learning_rate": 2.408379772034353e-07,
      "loss": 0.4197,
      "step": 1388
    },
    {
      "epoch": 2.7306684141546524,
      "grad_norm": 0.185395285487175,
      "learning_rate": 2.3733745054762059e-07,
      "loss": 0.4084,
      "step": 1389
    },
    {
      "epoch": 2.7326343381389253,
      "grad_norm": 0.19016891717910767,
      "learning_rate": 2.3386193163248193e-07,
      "loss": 0.4444,
      "step": 1390
    },
    {
      "epoch": 2.734600262123198,
      "grad_norm": 0.18271036446094513,
      "learning_rate": 2.3041143870723925e-07,
      "loss": 0.4086,
      "step": 1391
    },
    {
      "epoch": 2.7365661861074706,
      "grad_norm": 0.18831001222133636,
      "learning_rate": 2.2698598988970422e-07,
      "loss": 0.4097,
      "step": 1392
    },
    {
      "epoch": 2.738532110091743,
      "grad_norm": 0.18966880440711975,
      "learning_rate": 2.2358560316619093e-07,
      "loss": 0.4094,
      "step": 1393
    },
    {
      "epoch": 2.740498034076016,
      "grad_norm": 0.19208042323589325,
      "learning_rate": 2.2021029639141435e-07,
      "loss": 0.4268,
      "step": 1394
    },
    {
      "epoch": 2.7424639580602883,
      "grad_norm": 0.1806749701499939,
      "learning_rate": 2.1686008728840301e-07,
      "loss": 0.4164,
      "step": 1395
    },
    {
      "epoch": 2.744429882044561,
      "grad_norm": 0.19171354174613953,
      "learning_rate": 2.135349934483999e-07,
      "loss": 0.4241,
      "step": 1396
    },
    {
      "epoch": 2.7463958060288336,
      "grad_norm": 0.1913602203130722,
      "learning_rate": 2.102350323307756e-07,
      "loss": 0.4216,
      "step": 1397
    },
    {
      "epoch": 2.748361730013106,
      "grad_norm": 0.1981024146080017,
      "learning_rate": 2.0696022126293126e-07,
      "loss": 0.4092,
      "step": 1398
    },
    {
      "epoch": 2.750327653997379,
      "grad_norm": 0.18847931921482086,
      "learning_rate": 2.0371057744021315e-07,
      "loss": 0.4213,
      "step": 1399
    },
    {
      "epoch": 2.7522935779816513,
      "grad_norm": 0.1934966892004013,
      "learning_rate": 2.004861179258183e-07,
      "loss": 0.4432,
      "step": 1400
    },
    {
      "epoch": 2.754259501965924,
      "grad_norm": 0.18944783508777618,
      "learning_rate": 1.9728685965070604e-07,
      "loss": 0.4146,
      "step": 1401
    },
    {
      "epoch": 2.7562254259501966,
      "grad_norm": 0.19792866706848145,
      "learning_rate": 1.9411281941351001e-07,
      "loss": 0.4145,
      "step": 1402
    },
    {
      "epoch": 2.758191349934469,
      "grad_norm": 0.1858528107404709,
      "learning_rate": 1.9096401388044695e-07,
      "loss": 0.4184,
      "step": 1403
    },
    {
      "epoch": 2.760157273918742,
      "grad_norm": 0.1945904940366745,
      "learning_rate": 1.8784045958523623e-07,
      "loss": 0.4196,
      "step": 1404
    },
    {
      "epoch": 2.7621231979030143,
      "grad_norm": 0.18836303055286407,
      "learning_rate": 1.8474217292900275e-07,
      "loss": 0.4256,
      "step": 1405
    },
    {
      "epoch": 2.764089121887287,
      "grad_norm": 0.18192465603351593,
      "learning_rate": 1.816691701802009e-07,
      "loss": 0.4093,
      "step": 1406
    },
    {
      "epoch": 2.7660550458715596,
      "grad_norm": 0.19344103336334229,
      "learning_rate": 1.786214674745218e-07,
      "loss": 0.4229,
      "step": 1407
    },
    {
      "epoch": 2.768020969855832,
      "grad_norm": 0.19816488027572632,
      "learning_rate": 1.7559908081481225e-07,
      "loss": 0.4329,
      "step": 1408
    },
    {
      "epoch": 2.769986893840105,
      "grad_norm": 0.1782555729150772,
      "learning_rate": 1.7260202607098985e-07,
      "loss": 0.3958,
      "step": 1409
    },
    {
      "epoch": 2.7719528178243773,
      "grad_norm": 0.19153274595737457,
      "learning_rate": 1.6963031897995863e-07,
      "loss": 0.4101,
      "step": 1410
    },
    {
      "epoch": 2.77391874180865,
      "grad_norm": 0.1951267421245575,
      "learning_rate": 1.6668397514553013e-07,
      "loss": 0.4227,
      "step": 1411
    },
    {
      "epoch": 2.7758846657929226,
      "grad_norm": 0.20096047222614288,
      "learning_rate": 1.6376301003833583e-07,
      "loss": 0.4401,
      "step": 1412
    },
    {
      "epoch": 2.777850589777195,
      "grad_norm": 0.18297040462493896,
      "learning_rate": 1.6086743899575042e-07,
      "loss": 0.4181,
      "step": 1413
    },
    {
      "epoch": 2.779816513761468,
      "grad_norm": 0.18248505890369415,
      "learning_rate": 1.5799727722180858e-07,
      "loss": 0.4128,
      "step": 1414
    },
    {
      "epoch": 2.7817824377457407,
      "grad_norm": 0.1905166059732437,
      "learning_rate": 1.551525397871273e-07,
      "loss": 0.4121,
      "step": 1415
    },
    {
      "epoch": 2.783748361730013,
      "grad_norm": 0.18732059001922607,
      "learning_rate": 1.523332416288259e-07,
      "loss": 0.4315,
      "step": 1416
    },
    {
      "epoch": 2.7857142857142856,
      "grad_norm": 0.18855880200862885,
      "learning_rate": 1.4953939755044556e-07,
      "loss": 0.4245,
      "step": 1417
    },
    {
      "epoch": 2.7876802096985585,
      "grad_norm": 0.18331807851791382,
      "learning_rate": 1.46771022221876e-07,
      "loss": 0.4042,
      "step": 1418
    },
    {
      "epoch": 2.789646133682831,
      "grad_norm": 0.19505196809768677,
      "learning_rate": 1.4402813017927396e-07,
      "loss": 0.4266,
      "step": 1419
    },
    {
      "epoch": 2.7916120576671037,
      "grad_norm": 0.19084881246089935,
      "learning_rate": 1.413107358249899e-07,
      "loss": 0.4289,
      "step": 1420
    },
    {
      "epoch": 2.793577981651376,
      "grad_norm": 0.18647164106369019,
      "learning_rate": 1.386188534274896e-07,
      "loss": 0.4322,
      "step": 1421
    },
    {
      "epoch": 2.7955439056356486,
      "grad_norm": 0.19274601340293884,
      "learning_rate": 1.3595249712128334e-07,
      "loss": 0.4357,
      "step": 1422
    },
    {
      "epoch": 2.7975098296199215,
      "grad_norm": 0.18728584051132202,
      "learning_rate": 1.3331168090684742e-07,
      "loss": 0.4352,
      "step": 1423
    },
    {
      "epoch": 2.799475753604194,
      "grad_norm": 0.18775223195552826,
      "learning_rate": 1.3069641865055328e-07,
      "loss": 0.4336,
      "step": 1424
    },
    {
      "epoch": 2.8014416775884667,
      "grad_norm": 0.18465016782283783,
      "learning_rate": 1.28106724084594e-07,
      "loss": 0.4149,
      "step": 1425
    },
    {
      "epoch": 2.803407601572739,
      "grad_norm": 0.18570895493030548,
      "learning_rate": 1.2554261080691076e-07,
      "loss": 0.44,
      "step": 1426
    },
    {
      "epoch": 2.8053735255570116,
      "grad_norm": 0.18735641241073608,
      "learning_rate": 1.2300409228112541e-07,
      "loss": 0.3975,
      "step": 1427
    },
    {
      "epoch": 2.8073394495412844,
      "grad_norm": 0.1925758272409439,
      "learning_rate": 1.2049118183646403e-07,
      "loss": 0.4266,
      "step": 1428
    },
    {
      "epoch": 2.809305373525557,
      "grad_norm": 0.1925830841064453,
      "learning_rate": 1.1800389266769242e-07,
      "loss": 0.4065,
      "step": 1429
    },
    {
      "epoch": 2.8112712975098297,
      "grad_norm": 0.18923647701740265,
      "learning_rate": 1.1554223783504348e-07,
      "loss": 0.4028,
      "step": 1430
    },
    {
      "epoch": 2.813237221494102,
      "grad_norm": 0.1853543221950531,
      "learning_rate": 1.1310623026414891e-07,
      "loss": 0.4208,
      "step": 1431
    },
    {
      "epoch": 2.8152031454783746,
      "grad_norm": 0.18889391422271729,
      "learning_rate": 1.1069588274597365e-07,
      "loss": 0.4289,
      "step": 1432
    },
    {
      "epoch": 2.8171690694626474,
      "grad_norm": 0.19322317838668823,
      "learning_rate": 1.0831120793674598e-07,
      "loss": 0.4118,
      "step": 1433
    },
    {
      "epoch": 2.8191349934469203,
      "grad_norm": 0.18789125978946686,
      "learning_rate": 1.059522183578926e-07,
      "loss": 0.4263,
      "step": 1434
    },
    {
      "epoch": 2.8211009174311927,
      "grad_norm": 0.1874559223651886,
      "learning_rate": 1.0361892639597193e-07,
      "loss": 0.4064,
      "step": 1435
    },
    {
      "epoch": 2.823066841415465,
      "grad_norm": 0.18816040456295013,
      "learning_rate": 1.0131134430261036e-07,
      "loss": 0.4146,
      "step": 1436
    },
    {
      "epoch": 2.825032765399738,
      "grad_norm": 0.18052305281162262,
      "learning_rate": 9.902948419443669e-08,
      "loss": 0.414,
      "step": 1437
    },
    {
      "epoch": 2.8269986893840104,
      "grad_norm": 0.18447034060955048,
      "learning_rate": 9.67733580530189e-08,
      "loss": 0.4122,
      "step": 1438
    },
    {
      "epoch": 2.8289646133682833,
      "grad_norm": 0.19152402877807617,
      "learning_rate": 9.454297772480137e-08,
      "loss": 0.4309,
      "step": 1439
    },
    {
      "epoch": 2.8309305373525557,
      "grad_norm": 0.18504272401332855,
      "learning_rate": 9.233835492104326e-08,
      "loss": 0.4138,
      "step": 1440
    },
    {
      "epoch": 2.832896461336828,
      "grad_norm": 0.19270551204681396,
      "learning_rate": 9.015950121775474e-08,
      "loss": 0.4307,
      "step": 1441
    },
    {
      "epoch": 2.834862385321101,
      "grad_norm": 0.189494326710701,
      "learning_rate": 8.800642805563975e-08,
      "loss": 0.4145,
      "step": 1442
    },
    {
      "epoch": 2.8368283093053734,
      "grad_norm": 0.18586067855358124,
      "learning_rate": 8.587914674003384e-08,
      "loss": 0.4264,
      "step": 1443
    },
    {
      "epoch": 2.8387942332896463,
      "grad_norm": 0.18750880658626556,
      "learning_rate": 8.377766844084311e-08,
      "loss": 0.4302,
      "step": 1444
    },
    {
      "epoch": 2.8407601572739187,
      "grad_norm": 0.1872173547744751,
      "learning_rate": 8.170200419248931e-08,
      "loss": 0.4175,
      "step": 1445
    },
    {
      "epoch": 2.842726081258191,
      "grad_norm": 0.19608041644096375,
      "learning_rate": 7.965216489384919e-08,
      "loss": 0.4297,
      "step": 1446
    },
    {
      "epoch": 2.844692005242464,
      "grad_norm": 0.18357668817043304,
      "learning_rate": 7.762816130819862e-08,
      "loss": 0.4288,
      "step": 1447
    },
    {
      "epoch": 2.8466579292267364,
      "grad_norm": 0.18124645948410034,
      "learning_rate": 7.563000406315579e-08,
      "loss": 0.4069,
      "step": 1448
    },
    {
      "epoch": 2.8486238532110093,
      "grad_norm": 0.19631488621234894,
      "learning_rate": 7.365770365062308e-08,
      "loss": 0.4277,
      "step": 1449
    },
    {
      "epoch": 2.8505897771952817,
      "grad_norm": 0.18377605080604553,
      "learning_rate": 7.171127042673753e-08,
      "loss": 0.406,
      "step": 1450
    },
    {
      "epoch": 2.852555701179554,
      "grad_norm": 0.18855401873588562,
      "learning_rate": 6.979071461181042e-08,
      "loss": 0.4212,
      "step": 1451
    },
    {
      "epoch": 2.854521625163827,
      "grad_norm": 0.19027647376060486,
      "learning_rate": 6.789604629027614e-08,
      "loss": 0.4386,
      "step": 1452
    },
    {
      "epoch": 2.8564875491481,
      "grad_norm": 0.18420332670211792,
      "learning_rate": 6.602727541064114e-08,
      "loss": 0.4125,
      "step": 1453
    },
    {
      "epoch": 2.8584534731323723,
      "grad_norm": 0.19108708202838898,
      "learning_rate": 6.418441178542845e-08,
      "loss": 0.4264,
      "step": 1454
    },
    {
      "epoch": 2.8604193971166447,
      "grad_norm": 0.19463279843330383,
      "learning_rate": 6.236746509112824e-08,
      "loss": 0.4392,
      "step": 1455
    },
    {
      "epoch": 2.8623853211009176,
      "grad_norm": 0.21023961901664734,
      "learning_rate": 6.057644486814507e-08,
      "loss": 0.4326,
      "step": 1456
    },
    {
      "epoch": 2.86435124508519,
      "grad_norm": 0.18252646923065186,
      "learning_rate": 5.881136052075076e-08,
      "loss": 0.4186,
      "step": 1457
    },
    {
      "epoch": 2.866317169069463,
      "grad_norm": 0.19727899134159088,
      "learning_rate": 5.707222131703216e-08,
      "loss": 0.4393,
      "step": 1458
    },
    {
      "epoch": 2.8682830930537353,
      "grad_norm": 0.18230728805065155,
      "learning_rate": 5.535903638884399e-08,
      "loss": 0.4199,
      "step": 1459
    },
    {
      "epoch": 2.8702490170380077,
      "grad_norm": 0.18964985013008118,
      "learning_rate": 5.367181473176053e-08,
      "loss": 0.4114,
      "step": 1460
    },
    {
      "epoch": 2.8722149410222806,
      "grad_norm": 0.1857885718345642,
      "learning_rate": 5.201056520502734e-08,
      "loss": 0.4117,
      "step": 1461
    },
    {
      "epoch": 2.874180865006553,
      "grad_norm": 0.18905998766422272,
      "learning_rate": 5.0375296531518516e-08,
      "loss": 0.4263,
      "step": 1462
    },
    {
      "epoch": 2.876146788990826,
      "grad_norm": 0.18672683835029602,
      "learning_rate": 4.876601729768504e-08,
      "loss": 0.403,
      "step": 1463
    },
    {
      "epoch": 2.8781127129750983,
      "grad_norm": 0.18833793699741364,
      "learning_rate": 4.718273595351486e-08,
      "loss": 0.4229,
      "step": 1464
    },
    {
      "epoch": 2.8800786369593707,
      "grad_norm": 0.18816186487674713,
      "learning_rate": 4.5625460812485644e-08,
      "loss": 0.4273,
      "step": 1465
    },
    {
      "epoch": 2.8820445609436436,
      "grad_norm": 0.1809256672859192,
      "learning_rate": 4.409420005152154e-08,
      "loss": 0.4101,
      "step": 1466
    },
    {
      "epoch": 2.884010484927916,
      "grad_norm": 0.19573166966438293,
      "learning_rate": 4.25889617109515e-08,
      "loss": 0.4299,
      "step": 1467
    },
    {
      "epoch": 2.885976408912189,
      "grad_norm": 0.19006499648094177,
      "learning_rate": 4.1109753694466014e-08,
      "loss": 0.4442,
      "step": 1468
    },
    {
      "epoch": 2.8879423328964613,
      "grad_norm": 0.18832284212112427,
      "learning_rate": 3.965658376907544e-08,
      "loss": 0.4221,
      "step": 1469
    },
    {
      "epoch": 2.8899082568807337,
      "grad_norm": 0.189764603972435,
      "learning_rate": 3.8229459565070074e-08,
      "loss": 0.4172,
      "step": 1470
    },
    {
      "epoch": 2.8918741808650066,
      "grad_norm": 0.18678680062294006,
      "learning_rate": 3.682838857597959e-08,
      "loss": 0.44,
      "step": 1471
    },
    {
      "epoch": 2.893840104849279,
      "grad_norm": 0.19102615118026733,
      "learning_rate": 3.545337815853256e-08,
      "loss": 0.4117,
      "step": 1472
    },
    {
      "epoch": 2.895806028833552,
      "grad_norm": 0.18888624012470245,
      "learning_rate": 3.410443553262033e-08,
      "loss": 0.4203,
      "step": 1473
    },
    {
      "epoch": 2.8977719528178243,
      "grad_norm": 0.18853890895843506,
      "learning_rate": 3.2781567781258185e-08,
      "loss": 0.4149,
      "step": 1474
    },
    {
      "epoch": 2.8997378768020967,
      "grad_norm": 0.19942954182624817,
      "learning_rate": 3.1484781850545375e-08,
      "loss": 0.4416,
      "step": 1475
    },
    {
      "epoch": 2.9017038007863696,
      "grad_norm": 0.18686319887638092,
      "learning_rate": 3.0214084549632925e-08,
      "loss": 0.4301,
      "step": 1476
    },
    {
      "epoch": 2.9036697247706424,
      "grad_norm": 0.18849965929985046,
      "learning_rate": 2.8969482550686435e-08,
      "loss": 0.4187,
      "step": 1477
    },
    {
      "epoch": 2.905635648754915,
      "grad_norm": 0.18626683950424194,
      "learning_rate": 2.7750982388848324e-08,
      "loss": 0.4053,
      "step": 1478
    },
    {
      "epoch": 2.9076015727391873,
      "grad_norm": 0.1893082559108734,
      "learning_rate": 2.6558590462207322e-08,
      "loss": 0.4257,
      "step": 1479
    },
    {
      "epoch": 2.90956749672346,
      "grad_norm": 0.18734963238239288,
      "learning_rate": 2.539231303176293e-08,
      "loss": 0.4164,
      "step": 1480
    },
    {
      "epoch": 2.9115334207077326,
      "grad_norm": 0.18158259987831116,
      "learning_rate": 2.4252156221393764e-08,
      "loss": 0.4407,
      "step": 1481
    },
    {
      "epoch": 2.9134993446920054,
      "grad_norm": 0.19056791067123413,
      "learning_rate": 2.3138126017822614e-08,
      "loss": 0.4275,
      "step": 1482
    },
    {
      "epoch": 2.915465268676278,
      "grad_norm": 0.17401531338691711,
      "learning_rate": 2.205022827058867e-08,
      "loss": 0.3968,
      "step": 1483
    },
    {
      "epoch": 2.9174311926605503,
      "grad_norm": 0.1860005408525467,
      "learning_rate": 2.0988468692014208e-08,
      "loss": 0.4114,
      "step": 1484
    },
    {
      "epoch": 2.919397116644823,
      "grad_norm": 0.18477967381477356,
      "learning_rate": 1.99528528571763e-08,
      "loss": 0.4195,
      "step": 1485
    },
    {
      "epoch": 2.9213630406290956,
      "grad_norm": 0.18349048495292664,
      "learning_rate": 1.8943386203875702e-08,
      "loss": 0.4166,
      "step": 1486
    },
    {
      "epoch": 2.9233289646133684,
      "grad_norm": 0.18444041907787323,
      "learning_rate": 1.7960074032610243e-08,
      "loss": 0.3953,
      "step": 1487
    },
    {
      "epoch": 2.925294888597641,
      "grad_norm": 0.18297380208969116,
      "learning_rate": 1.7002921506544812e-08,
      "loss": 0.4158,
      "step": 1488
    },
    {
      "epoch": 2.9272608125819133,
      "grad_norm": 0.18708384037017822,
      "learning_rate": 1.607193365148696e-08,
      "loss": 0.4354,
      "step": 1489
    },
    {
      "epoch": 2.929226736566186,
      "grad_norm": 0.2040756493806839,
      "learning_rate": 1.516711535585802e-08,
      "loss": 0.4298,
      "step": 1490
    },
    {
      "epoch": 2.9311926605504586,
      "grad_norm": 0.18153029680252075,
      "learning_rate": 1.4288471370669244e-08,
      "loss": 0.4141,
      "step": 1491
    },
    {
      "epoch": 2.9331585845347314,
      "grad_norm": 0.1862143576145172,
      "learning_rate": 1.3436006309495152e-08,
      "loss": 0.4239,
      "step": 1492
    },
    {
      "epoch": 2.935124508519004,
      "grad_norm": 0.20297810435295105,
      "learning_rate": 1.2609724648450228e-08,
      "loss": 0.4275,
      "step": 1493
    },
    {
      "epoch": 2.9370904325032763,
      "grad_norm": 0.18510276079177856,
      "learning_rate": 1.1809630726167808e-08,
      "loss": 0.4167,
      "step": 1494
    },
    {
      "epoch": 2.939056356487549,
      "grad_norm": 0.18417201936244965,
      "learning_rate": 1.1035728743771235e-08,
      "loss": 0.4285,
      "step": 1495
    },
    {
      "epoch": 2.941022280471822,
      "grad_norm": 0.1946430206298828,
      "learning_rate": 1.0288022764857741e-08,
      "loss": 0.4275,
      "step": 1496
    },
    {
      "epoch": 2.9429882044560944,
      "grad_norm": 0.19136624038219452,
      "learning_rate": 9.566516715474594e-09,
      "loss": 0.4281,
      "step": 1497
    },
    {
      "epoch": 2.944954128440367,
      "grad_norm": 0.20006603002548218,
      "learning_rate": 8.871214384097993e-09,
      "loss": 0.4273,
      "step": 1498
    },
    {
      "epoch": 2.9469200524246397,
      "grad_norm": 0.18556608259677887,
      "learning_rate": 8.202119421615306e-09,
      "loss": 0.4261,
      "step": 1499
    },
    {
      "epoch": 2.948885976408912,
      "grad_norm": 0.18632759153842926,
      "learning_rate": 7.559235341302872e-09,
      "loss": 0.4167,
      "step": 1500
    },
    {
      "epoch": 2.950851900393185,
      "grad_norm": 0.1911812722682953,
      "learning_rate": 6.942565518810451e-09,
      "loss": 0.4116,
      "step": 1501
    },
    {
      "epoch": 2.9528178243774574,
      "grad_norm": 0.18124179542064667,
      "learning_rate": 6.352113192141241e-09,
      "loss": 0.3971,
      "step": 1502
    },
    {
      "epoch": 2.95478374836173,
      "grad_norm": 0.18406546115875244,
      "learning_rate": 5.787881461636891e-09,
      "loss": 0.4305,
      "step": 1503
    },
    {
      "epoch": 2.9567496723460027,
      "grad_norm": 0.18451999127864838,
      "learning_rate": 5.24987328995974e-09,
      "loss": 0.4107,
      "step": 1504
    },
    {
      "epoch": 2.958715596330275,
      "grad_norm": 0.1855209767818451,
      "learning_rate": 4.738091502077269e-09,
      "loss": 0.4312,
      "step": 1505
    },
    {
      "epoch": 2.960681520314548,
      "grad_norm": 0.19284525513648987,
      "learning_rate": 4.252538785248228e-09,
      "loss": 0.4122,
      "step": 1506
    },
    {
      "epoch": 2.9626474442988204,
      "grad_norm": 0.1882369965314865,
      "learning_rate": 3.793217689008199e-09,
      "loss": 0.4288,
      "step": 1507
    },
    {
      "epoch": 2.964613368283093,
      "grad_norm": 0.181296244263649,
      "learning_rate": 3.360130625155722e-09,
      "loss": 0.4153,
      "step": 1508
    },
    {
      "epoch": 2.9665792922673657,
      "grad_norm": 0.19471704959869385,
      "learning_rate": 2.9532798677395226e-09,
      "loss": 0.4211,
      "step": 1509
    },
    {
      "epoch": 2.968545216251638,
      "grad_norm": 0.18209101259708405,
      "learning_rate": 2.5726675530479695e-09,
      "loss": 0.4123,
      "step": 1510
    },
    {
      "epoch": 2.970511140235911,
      "grad_norm": 0.19035595655441284,
      "learning_rate": 2.21829567959686e-09,
      "loss": 0.4341,
      "step": 1511
    },
    {
      "epoch": 2.9724770642201834,
      "grad_norm": 0.18587549030780792,
      "learning_rate": 1.8901661081172084e-09,
      "loss": 0.4095,
      "step": 1512
    },
    {
      "epoch": 2.974442988204456,
      "grad_norm": 0.1827526092529297,
      "learning_rate": 1.5882805615496931e-09,
      "loss": 0.4119,
      "step": 1513
    },
    {
      "epoch": 2.9764089121887287,
      "grad_norm": 0.18282373249530792,
      "learning_rate": 1.312640625030781e-09,
      "loss": 0.4085,
      "step": 1514
    },
    {
      "epoch": 2.9783748361730016,
      "grad_norm": 0.19089441001415253,
      "learning_rate": 1.0632477458888401e-09,
      "loss": 0.4281,
      "step": 1515
    },
    {
      "epoch": 2.980340760157274,
      "grad_norm": 0.18670634925365448,
      "learning_rate": 8.401032336330384e-10,
      "loss": 0.4218,
      "step": 1516
    },
    {
      "epoch": 2.9823066841415464,
      "grad_norm": 0.20464399456977844,
      "learning_rate": 6.432082599489021e-10,
      "loss": 0.4187,
      "step": 1517
    },
    {
      "epoch": 2.9842726081258193,
      "grad_norm": 0.1876835972070694,
      "learning_rate": 4.725638586894344e-10,
      "loss": 0.429,
      "step": 1518
    },
    {
      "epoch": 2.9862385321100917,
      "grad_norm": 0.18850575387477875,
      "learning_rate": 3.2817092587345e-10,
      "loss": 0.4134,
      "step": 1519
    },
    {
      "epoch": 2.9882044560943646,
      "grad_norm": 0.19078615307807922,
      "learning_rate": 2.1003021967780369e-10,
      "loss": 0.4259,
      "step": 1520
    },
    {
      "epoch": 2.990170380078637,
      "grad_norm": 0.17989006638526917,
      "learning_rate": 1.1814236043405924e-10,
      "loss": 0.4127,
      "step": 1521
    },
    {
      "epoch": 2.9921363040629094,
      "grad_norm": 0.18568894267082214,
      "learning_rate": 5.250783062682452e-11,
      "loss": 0.4284,
      "step": 1522
    },
    {
      "epoch": 2.9941022280471823,
      "grad_norm": 0.19495601952075958,
      "learning_rate": 1.3126974888200139e-11,
      "loss": 0.4265,
      "step": 1523
    },
    {
      "epoch": 2.9960681520314547,
      "grad_norm": 0.19474370777606964,
      "learning_rate": 0.0,
      "loss": 0.4213,
      "step": 1524
    },
    {
      "epoch": 2.9960681520314547,
      "step": 1524,
      "total_flos": 4792734141382656.0,
      "train_loss": 0.46163445664203073,
      "train_runtime": 82407.0132,
      "train_samples_per_second": 1.777,
      "train_steps_per_second": 0.018
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1524,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4792734141382656.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}