{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 112.94117647058823,
  "eval_steps": 500,
  "global_step": 1440,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 3.280456066131592,
      "learning_rate": 3.4722222222222224e-06,
      "loss": 2.7006,
      "step": 10
    },
    {
      "epoch": 0.9411764705882353,
      "eval_accuracy": 0.11666666666666667,
      "eval_loss": 2.678151845932007,
      "eval_runtime": 4.6451,
      "eval_samples_per_second": 38.751,
      "eval_steps_per_second": 1.292,
      "step": 12
    },
    {
      "epoch": 1.5686274509803921,
      "grad_norm": 7.857480049133301,
      "learning_rate": 6.944444444444445e-06,
      "loss": 2.6863,
      "step": 20
    },
    {
      "epoch": 1.9607843137254903,
      "eval_accuracy": 0.16111111111111112,
      "eval_loss": 2.627171039581299,
      "eval_runtime": 3.8353,
      "eval_samples_per_second": 46.933,
      "eval_steps_per_second": 1.564,
      "step": 25
    },
    {
      "epoch": 2.3529411764705883,
      "grad_norm": 3.7707808017730713,
      "learning_rate": 1.0416666666666668e-05,
      "loss": 2.6437,
      "step": 30
    },
    {
      "epoch": 2.980392156862745,
      "eval_accuracy": 0.28888888888888886,
      "eval_loss": 2.5389277935028076,
      "eval_runtime": 3.8826,
      "eval_samples_per_second": 46.36,
      "eval_steps_per_second": 1.545,
      "step": 38
    },
    {
      "epoch": 3.1372549019607843,
      "grad_norm": 5.107277870178223,
      "learning_rate": 1.388888888888889e-05,
      "loss": 2.5839,
      "step": 40
    },
    {
      "epoch": 3.9215686274509802,
      "grad_norm": 6.703848361968994,
      "learning_rate": 1.736111111111111e-05,
      "loss": 2.4851,
      "step": 50
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.4111111111111111,
      "eval_loss": 2.411587953567505,
      "eval_runtime": 3.8828,
      "eval_samples_per_second": 46.359,
      "eval_steps_per_second": 1.545,
      "step": 51
    },
    {
      "epoch": 4.705882352941177,
      "grad_norm": 11.087530136108398,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 2.3732,
      "step": 60
    },
    {
      "epoch": 4.9411764705882355,
      "eval_accuracy": 0.4888888888888889,
      "eval_loss": 2.270714521408081,
      "eval_runtime": 3.8466,
      "eval_samples_per_second": 46.794,
      "eval_steps_per_second": 1.56,
      "step": 63
    },
    {
      "epoch": 5.490196078431373,
      "grad_norm": 7.360437393188477,
      "learning_rate": 2.4305555555555558e-05,
      "loss": 2.2546,
      "step": 70
    },
    {
      "epoch": 5.96078431372549,
      "eval_accuracy": 0.5722222222222222,
      "eval_loss": 2.0710320472717285,
      "eval_runtime": 3.9713,
      "eval_samples_per_second": 45.326,
      "eval_steps_per_second": 1.511,
      "step": 76
    },
    {
      "epoch": 6.2745098039215685,
      "grad_norm": 6.191979885101318,
      "learning_rate": 2.777777777777778e-05,
      "loss": 2.1023,
      "step": 80
    },
    {
      "epoch": 6.980392156862745,
      "eval_accuracy": 0.6166666666666667,
      "eval_loss": 1.8370894193649292,
      "eval_runtime": 3.8628,
      "eval_samples_per_second": 46.599,
      "eval_steps_per_second": 1.553,
      "step": 89
    },
    {
      "epoch": 7.0588235294117645,
      "grad_norm": 7.690328598022461,
      "learning_rate": 3.125e-05,
      "loss": 1.9156,
      "step": 90
    },
    {
      "epoch": 7.8431372549019605,
      "grad_norm": 7.919386863708496,
      "learning_rate": 3.472222222222222e-05,
      "loss": 1.7115,
      "step": 100
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.6111111111111112,
      "eval_loss": 1.6161085367202759,
      "eval_runtime": 3.8886,
      "eval_samples_per_second": 46.29,
      "eval_steps_per_second": 1.543,
      "step": 102
    },
    {
      "epoch": 8.627450980392156,
      "grad_norm": 12.924628257751465,
      "learning_rate": 3.8194444444444444e-05,
      "loss": 1.5295,
      "step": 110
    },
    {
      "epoch": 8.941176470588236,
      "eval_accuracy": 0.6277777777777778,
      "eval_loss": 1.4381340742111206,
      "eval_runtime": 3.8538,
      "eval_samples_per_second": 46.708,
      "eval_steps_per_second": 1.557,
      "step": 114
    },
    {
      "epoch": 9.411764705882353,
      "grad_norm": 14.41945743560791,
      "learning_rate": 4.166666666666667e-05,
      "loss": 1.3366,
      "step": 120
    },
    {
      "epoch": 9.96078431372549,
      "eval_accuracy": 0.65,
      "eval_loss": 1.2539671659469604,
      "eval_runtime": 3.9661,
      "eval_samples_per_second": 45.385,
      "eval_steps_per_second": 1.513,
      "step": 127
    },
    {
      "epoch": 10.196078431372548,
      "grad_norm": 17.627479553222656,
      "learning_rate": 4.5138888888888894e-05,
      "loss": 1.2377,
      "step": 130
    },
    {
      "epoch": 10.980392156862745,
      "grad_norm": 13.988055229187012,
      "learning_rate": 4.8611111111111115e-05,
      "loss": 1.0556,
      "step": 140
    },
    {
      "epoch": 10.980392156862745,
      "eval_accuracy": 0.6611111111111111,
      "eval_loss": 1.1632429361343384,
      "eval_runtime": 3.9305,
      "eval_samples_per_second": 45.796,
      "eval_steps_per_second": 1.527,
      "step": 140
    },
    {
      "epoch": 11.764705882352942,
      "grad_norm": 12.996641159057617,
      "learning_rate": 4.976851851851852e-05,
      "loss": 0.9657,
      "step": 150
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.7,
      "eval_loss": 1.0600230693817139,
      "eval_runtime": 3.9116,
      "eval_samples_per_second": 46.016,
      "eval_steps_per_second": 1.534,
      "step": 153
    },
    {
      "epoch": 12.549019607843137,
      "grad_norm": 26.70894432067871,
      "learning_rate": 4.938271604938271e-05,
      "loss": 0.8703,
      "step": 160
    },
    {
      "epoch": 12.941176470588236,
      "eval_accuracy": 0.7222222222222222,
      "eval_loss": 0.9983330368995667,
      "eval_runtime": 3.823,
      "eval_samples_per_second": 47.084,
      "eval_steps_per_second": 1.569,
      "step": 165
    },
    {
      "epoch": 13.333333333333334,
      "grad_norm": 24.295700073242188,
      "learning_rate": 4.899691358024692e-05,
      "loss": 0.8007,
      "step": 170
    },
    {
      "epoch": 13.96078431372549,
      "eval_accuracy": 0.7277777777777777,
      "eval_loss": 0.9474301934242249,
      "eval_runtime": 4.0708,
      "eval_samples_per_second": 44.218,
      "eval_steps_per_second": 1.474,
      "step": 178
    },
    {
      "epoch": 14.117647058823529,
      "grad_norm": 19.43092918395996,
      "learning_rate": 4.8611111111111115e-05,
      "loss": 0.7257,
      "step": 180
    },
    {
      "epoch": 14.901960784313726,
      "grad_norm": 22.165098190307617,
      "learning_rate": 4.8225308641975306e-05,
      "loss": 0.6398,
      "step": 190
    },
    {
      "epoch": 14.980392156862745,
      "eval_accuracy": 0.75,
      "eval_loss": 0.8633670210838318,
      "eval_runtime": 4.1208,
      "eval_samples_per_second": 43.681,
      "eval_steps_per_second": 1.456,
      "step": 191
    },
    {
      "epoch": 15.686274509803921,
      "grad_norm": 24.815359115600586,
      "learning_rate": 4.783950617283951e-05,
      "loss": 0.6023,
      "step": 200
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.7277777777777777,
      "eval_loss": 0.8527319431304932,
      "eval_runtime": 3.869,
      "eval_samples_per_second": 46.524,
      "eval_steps_per_second": 1.551,
      "step": 204
    },
    {
      "epoch": 16.470588235294116,
      "grad_norm": 28.00403594970703,
      "learning_rate": 4.745370370370371e-05,
      "loss": 0.583,
      "step": 210
    },
    {
      "epoch": 16.941176470588236,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.7927896976470947,
      "eval_runtime": 3.8557,
      "eval_samples_per_second": 46.684,
      "eval_steps_per_second": 1.556,
      "step": 216
    },
    {
      "epoch": 17.254901960784313,
      "grad_norm": 30.892383575439453,
      "learning_rate": 4.70679012345679e-05,
      "loss": 0.5279,
      "step": 220
    },
    {
      "epoch": 17.96078431372549,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.7896744012832642,
      "eval_runtime": 3.7921,
      "eval_samples_per_second": 47.468,
      "eval_steps_per_second": 1.582,
      "step": 229
    },
    {
      "epoch": 18.03921568627451,
      "grad_norm": 19.16618537902832,
      "learning_rate": 4.66820987654321e-05,
      "loss": 0.5084,
      "step": 230
    },
    {
      "epoch": 18.823529411764707,
      "grad_norm": 13.686722755432129,
      "learning_rate": 4.62962962962963e-05,
      "loss": 0.4643,
      "step": 240
    },
    {
      "epoch": 18.980392156862745,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.7885976433753967,
      "eval_runtime": 3.8507,
      "eval_samples_per_second": 46.745,
      "eval_steps_per_second": 1.558,
      "step": 242
    },
    {
      "epoch": 19.607843137254903,
      "grad_norm": 17.4645938873291,
      "learning_rate": 4.591049382716049e-05,
      "loss": 0.4296,
      "step": 250
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.7328829169273376,
      "eval_runtime": 3.8093,
      "eval_samples_per_second": 47.252,
      "eval_steps_per_second": 1.575,
      "step": 255
    },
    {
      "epoch": 20.392156862745097,
      "grad_norm": 16.678590774536133,
      "learning_rate": 4.5524691358024696e-05,
      "loss": 0.41,
      "step": 260
    },
    {
      "epoch": 20.941176470588236,
      "eval_accuracy": 0.7611111111111111,
      "eval_loss": 0.7316663861274719,
      "eval_runtime": 3.8668,
      "eval_samples_per_second": 46.55,
      "eval_steps_per_second": 1.552,
      "step": 267
    },
    {
      "epoch": 21.176470588235293,
      "grad_norm": 12.328489303588867,
      "learning_rate": 4.5138888888888894e-05,
      "loss": 0.3663,
      "step": 270
    },
    {
      "epoch": 21.96078431372549,
      "grad_norm": 23.868070602416992,
      "learning_rate": 4.4753086419753084e-05,
      "loss": 0.3674,
      "step": 280
    },
    {
      "epoch": 21.96078431372549,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.7170845866203308,
      "eval_runtime": 3.956,
      "eval_samples_per_second": 45.5,
      "eval_steps_per_second": 1.517,
      "step": 280
    },
    {
      "epoch": 22.745098039215687,
      "grad_norm": 21.286258697509766,
      "learning_rate": 4.436728395061729e-05,
      "loss": 0.3285,
      "step": 290
    },
    {
      "epoch": 22.980392156862745,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.7005434036254883,
      "eval_runtime": 3.9134,
      "eval_samples_per_second": 45.996,
      "eval_steps_per_second": 1.533,
      "step": 293
    },
    {
      "epoch": 23.529411764705884,
      "grad_norm": 12.573646545410156,
      "learning_rate": 4.3981481481481486e-05,
      "loss": 0.2978,
      "step": 300
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.7888888888888889,
      "eval_loss": 0.6576042771339417,
      "eval_runtime": 3.8597,
      "eval_samples_per_second": 46.636,
      "eval_steps_per_second": 1.555,
      "step": 306
    },
    {
      "epoch": 24.313725490196077,
      "grad_norm": 18.386383056640625,
      "learning_rate": 4.359567901234568e-05,
      "loss": 0.293,
      "step": 310
    },
    {
      "epoch": 24.941176470588236,
      "eval_accuracy": 0.8,
      "eval_loss": 0.644997239112854,
      "eval_runtime": 4.0291,
      "eval_samples_per_second": 44.675,
      "eval_steps_per_second": 1.489,
      "step": 318
    },
    {
      "epoch": 25.098039215686274,
      "grad_norm": 18.57107925415039,
      "learning_rate": 4.3209876543209875e-05,
      "loss": 0.2665,
      "step": 320
    },
    {
      "epoch": 25.88235294117647,
      "grad_norm": 16.507802963256836,
      "learning_rate": 4.282407407407408e-05,
      "loss": 0.2724,
      "step": 330
    },
    {
      "epoch": 25.96078431372549,
      "eval_accuracy": 0.7888888888888889,
      "eval_loss": 0.6764713525772095,
      "eval_runtime": 3.8073,
      "eval_samples_per_second": 47.278,
      "eval_steps_per_second": 1.576,
      "step": 331
    },
    {
      "epoch": 26.666666666666668,
      "grad_norm": 18.386646270751953,
      "learning_rate": 4.243827160493827e-05,
      "loss": 0.2494,
      "step": 340
    },
    {
      "epoch": 26.980392156862745,
      "eval_accuracy": 0.8055555555555556,
      "eval_loss": 0.6826486587524414,
      "eval_runtime": 3.9305,
      "eval_samples_per_second": 45.796,
      "eval_steps_per_second": 1.527,
      "step": 344
    },
    {
      "epoch": 27.45098039215686,
      "grad_norm": 16.895566940307617,
      "learning_rate": 4.205246913580247e-05,
      "loss": 0.2504,
      "step": 350
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.8055555555555556,
      "eval_loss": 0.6710352301597595,
      "eval_runtime": 3.8264,
      "eval_samples_per_second": 47.041,
      "eval_steps_per_second": 1.568,
      "step": 357
    },
    {
      "epoch": 28.235294117647058,
      "grad_norm": 14.318737983703613,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.2332,
      "step": 360
    },
    {
      "epoch": 28.941176470588236,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.666705846786499,
      "eval_runtime": 3.9608,
      "eval_samples_per_second": 45.445,
      "eval_steps_per_second": 1.515,
      "step": 369
    },
    {
      "epoch": 29.019607843137255,
      "grad_norm": 19.979778289794922,
      "learning_rate": 4.128086419753087e-05,
      "loss": 0.2071,
      "step": 370
    },
    {
      "epoch": 29.80392156862745,
      "grad_norm": 14.415696144104004,
      "learning_rate": 4.089506172839506e-05,
      "loss": 0.2012,
      "step": 380
    },
    {
      "epoch": 29.96078431372549,
      "eval_accuracy": 0.7944444444444444,
      "eval_loss": 0.7399319410324097,
      "eval_runtime": 3.8411,
      "eval_samples_per_second": 46.861,
      "eval_steps_per_second": 1.562,
      "step": 382
    },
    {
      "epoch": 30.58823529411765,
      "grad_norm": 10.09865665435791,
      "learning_rate": 4.0509259259259265e-05,
      "loss": 0.1866,
      "step": 390
    },
    {
      "epoch": 30.980392156862745,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.7311467528343201,
      "eval_runtime": 4.2025,
      "eval_samples_per_second": 42.831,
      "eval_steps_per_second": 1.428,
      "step": 395
    },
    {
      "epoch": 31.372549019607842,
      "grad_norm": 31.518268585205078,
      "learning_rate": 4.012345679012346e-05,
      "loss": 0.2031,
      "step": 400
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.7944444444444444,
      "eval_loss": 0.7076573967933655,
      "eval_runtime": 3.8553,
      "eval_samples_per_second": 46.689,
      "eval_steps_per_second": 1.556,
      "step": 408
    },
    {
      "epoch": 32.15686274509804,
      "grad_norm": 18.065155029296875,
      "learning_rate": 3.973765432098765e-05,
      "loss": 0.1969,
      "step": 410
    },
    {
      "epoch": 32.94117647058823,
      "grad_norm": 19.082073211669922,
      "learning_rate": 3.935185185185186e-05,
      "loss": 0.1969,
      "step": 420
    },
    {
      "epoch": 32.94117647058823,
      "eval_accuracy": 0.7666666666666667,
      "eval_loss": 0.7769466638565063,
      "eval_runtime": 3.9404,
      "eval_samples_per_second": 45.681,
      "eval_steps_per_second": 1.523,
      "step": 420
    },
    {
      "epoch": 33.72549019607843,
      "grad_norm": 23.67195701599121,
      "learning_rate": 3.8966049382716055e-05,
      "loss": 0.1968,
      "step": 430
    },
    {
      "epoch": 33.96078431372549,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.7666174173355103,
      "eval_runtime": 3.7732,
      "eval_samples_per_second": 47.705,
      "eval_steps_per_second": 1.59,
      "step": 433
    },
    {
      "epoch": 34.509803921568626,
      "grad_norm": 22.488903045654297,
      "learning_rate": 3.8580246913580246e-05,
      "loss": 0.1712,
      "step": 440
    },
    {
      "epoch": 34.98039215686274,
      "eval_accuracy": 0.8,
      "eval_loss": 0.6795583367347717,
      "eval_runtime": 3.7971,
      "eval_samples_per_second": 47.404,
      "eval_steps_per_second": 1.58,
      "step": 446
    },
    {
      "epoch": 35.294117647058826,
      "grad_norm": 20.480487823486328,
      "learning_rate": 3.8194444444444444e-05,
      "loss": 0.1813,
      "step": 450
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.8111111111111111,
      "eval_loss": 0.6653857827186584,
      "eval_runtime": 3.8151,
      "eval_samples_per_second": 47.181,
      "eval_steps_per_second": 1.573,
      "step": 459
    },
    {
      "epoch": 36.07843137254902,
      "grad_norm": 15.032731056213379,
      "learning_rate": 3.780864197530865e-05,
      "loss": 0.1625,
      "step": 460
    },
    {
      "epoch": 36.86274509803921,
      "grad_norm": 12.346388816833496,
      "learning_rate": 3.742283950617284e-05,
      "loss": 0.1678,
      "step": 470
    },
    {
      "epoch": 36.94117647058823,
      "eval_accuracy": 0.7888888888888889,
      "eval_loss": 0.6851311326026917,
      "eval_runtime": 3.8058,
      "eval_samples_per_second": 47.296,
      "eval_steps_per_second": 1.577,
      "step": 471
    },
    {
      "epoch": 37.64705882352941,
      "grad_norm": 16.994136810302734,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.1461,
      "step": 480
    },
    {
      "epoch": 37.96078431372549,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.7054334878921509,
      "eval_runtime": 3.847,
      "eval_samples_per_second": 46.79,
      "eval_steps_per_second": 1.56,
      "step": 484
    },
    {
      "epoch": 38.431372549019606,
      "grad_norm": 9.07701587677002,
      "learning_rate": 3.665123456790124e-05,
      "loss": 0.1244,
      "step": 490
    },
    {
      "epoch": 38.98039215686274,
      "eval_accuracy": 0.8055555555555556,
      "eval_loss": 0.7013460993766785,
      "eval_runtime": 3.8047,
      "eval_samples_per_second": 47.31,
      "eval_steps_per_second": 1.577,
      "step": 497
    },
    {
      "epoch": 39.21568627450981,
      "grad_norm": 9.217803955078125,
      "learning_rate": 3.626543209876543e-05,
      "loss": 0.1385,
      "step": 500
    },
    {
      "epoch": 40.0,
      "grad_norm": 18.938865661621094,
      "learning_rate": 3.587962962962963e-05,
      "loss": 0.1329,
      "step": 510
    },
    {
      "epoch": 40.0,
      "eval_accuracy": 0.8,
      "eval_loss": 0.6785274147987366,
      "eval_runtime": 3.86,
      "eval_samples_per_second": 46.632,
      "eval_steps_per_second": 1.554,
      "step": 510
    },
    {
      "epoch": 40.78431372549019,
      "grad_norm": 12.516762733459473,
      "learning_rate": 3.5493827160493834e-05,
      "loss": 0.1186,
      "step": 520
    },
    {
      "epoch": 40.94117647058823,
      "eval_accuracy": 0.7777777777777778,
      "eval_loss": 0.7499803900718689,
      "eval_runtime": 3.8115,
      "eval_samples_per_second": 47.225,
      "eval_steps_per_second": 1.574,
      "step": 522
    },
    {
      "epoch": 41.568627450980394,
      "grad_norm": 9.269219398498535,
      "learning_rate": 3.5108024691358025e-05,
      "loss": 0.1397,
      "step": 530
    },
    {
      "epoch": 41.96078431372549,
      "eval_accuracy": 0.8166666666666667,
      "eval_loss": 0.6819199919700623,
      "eval_runtime": 3.8462,
      "eval_samples_per_second": 46.799,
      "eval_steps_per_second": 1.56,
      "step": 535
    },
    {
      "epoch": 42.35294117647059,
      "grad_norm": 23.982585906982422,
      "learning_rate": 3.472222222222222e-05,
      "loss": 0.1324,
      "step": 540
    },
    {
      "epoch": 42.98039215686274,
      "eval_accuracy": 0.8111111111111111,
      "eval_loss": 0.6256746649742126,
      "eval_runtime": 3.8195,
      "eval_samples_per_second": 47.126,
      "eval_steps_per_second": 1.571,
      "step": 548
    },
    {
      "epoch": 43.13725490196079,
      "grad_norm": 11.348409652709961,
      "learning_rate": 3.4336419753086427e-05,
      "loss": 0.1461,
      "step": 550
    },
    {
      "epoch": 43.92156862745098,
      "grad_norm": 14.504171371459961,
      "learning_rate": 3.395061728395062e-05,
      "loss": 0.111,
      "step": 560
    },
    {
      "epoch": 44.0,
      "eval_accuracy": 0.8277777777777777,
      "eval_loss": 0.5938891172409058,
      "eval_runtime": 3.8255,
      "eval_samples_per_second": 47.053,
      "eval_steps_per_second": 1.568,
      "step": 561
    },
    {
      "epoch": 44.705882352941174,
      "grad_norm": 15.88039493560791,
      "learning_rate": 3.3564814814814815e-05,
      "loss": 0.1228,
      "step": 570
    },
    {
      "epoch": 44.94117647058823,
      "eval_accuracy": 0.8222222222222222,
      "eval_loss": 0.6379250288009644,
      "eval_runtime": 3.8691,
      "eval_samples_per_second": 46.522,
      "eval_steps_per_second": 1.551,
      "step": 573
    },
    {
      "epoch": 45.490196078431374,
      "grad_norm": 14.967761993408203,
      "learning_rate": 3.317901234567901e-05,
      "loss": 0.1085,
      "step": 580
    },
    {
      "epoch": 45.96078431372549,
      "eval_accuracy": 0.8222222222222222,
      "eval_loss": 0.6788524389266968,
      "eval_runtime": 3.8236,
      "eval_samples_per_second": 47.077,
      "eval_steps_per_second": 1.569,
      "step": 586
    },
    {
      "epoch": 46.27450980392157,
      "grad_norm": 7.978495121002197,
      "learning_rate": 3.279320987654321e-05,
      "loss": 0.1234,
      "step": 590
    },
    {
      "epoch": 46.98039215686274,
      "eval_accuracy": 0.8277777777777777,
      "eval_loss": 0.624097466468811,
      "eval_runtime": 3.7905,
      "eval_samples_per_second": 47.487,
      "eval_steps_per_second": 1.583,
      "step": 599
    },
    {
      "epoch": 47.05882352941177,
      "grad_norm": 21.228994369506836,
      "learning_rate": 3.240740740740741e-05,
      "loss": 0.1007,
      "step": 600
    },
    {
      "epoch": 47.84313725490196,
      "grad_norm": 16.632568359375,
      "learning_rate": 3.2021604938271605e-05,
      "loss": 0.1129,
      "step": 610
    },
    {
      "epoch": 48.0,
      "eval_accuracy": 0.7888888888888889,
      "eval_loss": 0.750299334526062,
      "eval_runtime": 3.8266,
      "eval_samples_per_second": 47.039,
      "eval_steps_per_second": 1.568,
      "step": 612
    },
    {
      "epoch": 48.627450980392155,
      "grad_norm": 8.629143714904785,
      "learning_rate": 3.16358024691358e-05,
      "loss": 0.1197,
      "step": 620
    },
    {
      "epoch": 48.94117647058823,
      "eval_accuracy": 0.7944444444444444,
      "eval_loss": 0.6861774325370789,
      "eval_runtime": 3.8119,
      "eval_samples_per_second": 47.22,
      "eval_steps_per_second": 1.574,
      "step": 624
    },
    {
      "epoch": 49.411764705882355,
      "grad_norm": 7.733061790466309,
      "learning_rate": 3.125e-05,
      "loss": 0.0898,
      "step": 630
    },
    {
      "epoch": 49.96078431372549,
      "eval_accuracy": 0.7888888888888889,
      "eval_loss": 0.6763875484466553,
      "eval_runtime": 3.818,
      "eval_samples_per_second": 47.145,
      "eval_steps_per_second": 1.571,
      "step": 637
    },
    {
      "epoch": 50.19607843137255,
      "grad_norm": 4.7213454246521,
      "learning_rate": 3.08641975308642e-05,
      "loss": 0.1021,
      "step": 640
    },
    {
      "epoch": 50.98039215686274,
      "grad_norm": 7.109160423278809,
      "learning_rate": 3.04783950617284e-05,
      "loss": 0.1057,
      "step": 650
    },
    {
      "epoch": 50.98039215686274,
      "eval_accuracy": 0.8166666666666667,
      "eval_loss": 0.6338934898376465,
      "eval_runtime": 3.8669,
      "eval_samples_per_second": 46.549,
      "eval_steps_per_second": 1.552,
      "step": 650
    },
    {
      "epoch": 51.76470588235294,
      "grad_norm": 16.793262481689453,
      "learning_rate": 3.0092592592592593e-05,
      "loss": 0.0893,
      "step": 660
    },
    {
      "epoch": 52.0,
      "eval_accuracy": 0.85,
      "eval_loss": 0.5828067064285278,
      "eval_runtime": 3.8135,
      "eval_samples_per_second": 47.2,
      "eval_steps_per_second": 1.573,
      "step": 663
    },
    {
      "epoch": 52.549019607843135,
      "grad_norm": 15.305388450622559,
      "learning_rate": 2.970679012345679e-05,
      "loss": 0.0736,
      "step": 670
    },
    {
      "epoch": 52.94117647058823,
      "eval_accuracy": 0.8111111111111111,
      "eval_loss": 0.6572611331939697,
      "eval_runtime": 3.8338,
      "eval_samples_per_second": 46.951,
      "eval_steps_per_second": 1.565,
      "step": 675
    },
    {
      "epoch": 53.333333333333336,
      "grad_norm": 13.537842750549316,
      "learning_rate": 2.9320987654320992e-05,
      "loss": 0.0752,
      "step": 680
    },
    {
      "epoch": 53.96078431372549,
      "eval_accuracy": 0.7944444444444444,
      "eval_loss": 0.6806420087814331,
      "eval_runtime": 3.821,
      "eval_samples_per_second": 47.108,
      "eval_steps_per_second": 1.57,
      "step": 688
    },
    {
      "epoch": 54.11764705882353,
      "grad_norm": 21.195058822631836,
      "learning_rate": 2.8935185185185186e-05,
      "loss": 0.081,
      "step": 690
    },
    {
      "epoch": 54.90196078431372,
      "grad_norm": 4.959319591522217,
      "learning_rate": 2.8549382716049384e-05,
      "loss": 0.1127,
      "step": 700
    },
    {
      "epoch": 54.98039215686274,
      "eval_accuracy": 0.8111111111111111,
      "eval_loss": 0.6222459673881531,
      "eval_runtime": 3.8138,
      "eval_samples_per_second": 47.197,
      "eval_steps_per_second": 1.573,
      "step": 701
    },
    {
      "epoch": 55.68627450980392,
      "grad_norm": 15.941388130187988,
      "learning_rate": 2.8163580246913578e-05,
      "loss": 0.1126,
      "step": 710
    },
    {
      "epoch": 56.0,
      "eval_accuracy": 0.8166666666666667,
      "eval_loss": 0.6305037140846252,
      "eval_runtime": 3.7889,
      "eval_samples_per_second": 47.507,
      "eval_steps_per_second": 1.584,
      "step": 714
    },
    {
      "epoch": 56.470588235294116,
      "grad_norm": 12.527265548706055,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.0874,
      "step": 720
    },
    {
      "epoch": 56.94117647058823,
      "eval_accuracy": 0.8111111111111111,
      "eval_loss": 0.6593422293663025,
      "eval_runtime": 3.7695,
      "eval_samples_per_second": 47.751,
      "eval_steps_per_second": 1.592,
      "step": 726
    },
    {
      "epoch": 57.254901960784316,
      "grad_norm": 7.021444797515869,
      "learning_rate": 2.7391975308641977e-05,
      "loss": 0.0806,
      "step": 730
    },
    {
      "epoch": 57.96078431372549,
      "eval_accuracy": 0.8166666666666667,
      "eval_loss": 0.7005773782730103,
      "eval_runtime": 3.7654,
      "eval_samples_per_second": 47.804,
      "eval_steps_per_second": 1.593,
      "step": 739
    },
    {
      "epoch": 58.03921568627451,
      "grad_norm": 13.148822784423828,
      "learning_rate": 2.700617283950617e-05,
      "loss": 0.0862,
      "step": 740
    },
    {
      "epoch": 58.8235294117647,
      "grad_norm": 21.20357322692871,
      "learning_rate": 2.6620370370370372e-05,
      "loss": 0.0978,
      "step": 750
    },
    {
      "epoch": 58.98039215686274,
      "eval_accuracy": 0.8055555555555556,
      "eval_loss": 0.6680053472518921,
      "eval_runtime": 3.8444,
      "eval_samples_per_second": 46.821,
      "eval_steps_per_second": 1.561,
      "step": 752
    },
    {
      "epoch": 59.6078431372549,
      "grad_norm": 14.901837348937988,
      "learning_rate": 2.623456790123457e-05,
      "loss": 0.0875,
      "step": 760
    },
    {
      "epoch": 60.0,
      "eval_accuracy": 0.8166666666666667,
      "eval_loss": 0.67389976978302,
      "eval_runtime": 3.8169,
      "eval_samples_per_second": 47.158,
      "eval_steps_per_second": 1.572,
      "step": 765
    },
    {
      "epoch": 60.3921568627451,
      "grad_norm": 7.75632905960083,
      "learning_rate": 2.5848765432098764e-05,
      "loss": 0.0722,
      "step": 770
    },
    {
      "epoch": 60.94117647058823,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.6340806484222412,
      "eval_runtime": 3.804,
      "eval_samples_per_second": 47.318,
      "eval_steps_per_second": 1.577,
      "step": 777
    },
    {
      "epoch": 61.1764705882353,
      "grad_norm": 15.415738105773926,
      "learning_rate": 2.5462962962962965e-05,
      "loss": 0.0901,
      "step": 780
    },
    {
      "epoch": 61.96078431372549,
      "grad_norm": 4.349020004272461,
      "learning_rate": 2.5077160493827162e-05,
      "loss": 0.0942,
      "step": 790
    },
    {
      "epoch": 61.96078431372549,
      "eval_accuracy": 0.8,
      "eval_loss": 0.6428362727165222,
      "eval_runtime": 3.7936,
      "eval_samples_per_second": 47.448,
      "eval_steps_per_second": 1.582,
      "step": 790
    },
    {
      "epoch": 62.745098039215684,
      "grad_norm": 9.049278259277344,
      "learning_rate": 2.4691358024691357e-05,
      "loss": 0.0957,
      "step": 800
    },
    {
      "epoch": 62.98039215686274,
      "eval_accuracy": 0.8,
      "eval_loss": 0.6757560968399048,
      "eval_runtime": 3.8348,
      "eval_samples_per_second": 46.938,
      "eval_steps_per_second": 1.565,
      "step": 803
    },
    {
      "epoch": 63.529411764705884,
      "grad_norm": 11.159144401550293,
      "learning_rate": 2.4305555555555558e-05,
      "loss": 0.0814,
      "step": 810
    },
    {
      "epoch": 64.0,
      "eval_accuracy": 0.8166666666666667,
      "eval_loss": 0.6104480028152466,
      "eval_runtime": 3.8591,
      "eval_samples_per_second": 46.643,
      "eval_steps_per_second": 1.555,
      "step": 816
    },
    {
      "epoch": 64.31372549019608,
      "grad_norm": 11.834258079528809,
      "learning_rate": 2.3919753086419755e-05,
      "loss": 0.077,
      "step": 820
    },
    {
      "epoch": 64.94117647058823,
      "eval_accuracy": 0.8111111111111111,
      "eval_loss": 0.6226403713226318,
      "eval_runtime": 3.8932,
      "eval_samples_per_second": 46.235,
      "eval_steps_per_second": 1.541,
      "step": 828
    },
    {
      "epoch": 65.09803921568627,
      "grad_norm": 11.226044654846191,
      "learning_rate": 2.353395061728395e-05,
      "loss": 0.0862,
      "step": 830
    },
    {
      "epoch": 65.88235294117646,
      "grad_norm": 14.373990058898926,
      "learning_rate": 2.314814814814815e-05,
      "loss": 0.1004,
      "step": 840
    },
    {
      "epoch": 65.96078431372548,
      "eval_accuracy": 0.8055555555555556,
      "eval_loss": 0.6898564696311951,
      "eval_runtime": 3.8229,
      "eval_samples_per_second": 47.085,
      "eval_steps_per_second": 1.569,
      "step": 841
    },
    {
      "epoch": 66.66666666666667,
      "grad_norm": 8.56983757019043,
      "learning_rate": 2.2762345679012348e-05,
      "loss": 0.0697,
      "step": 850
    },
    {
      "epoch": 66.98039215686275,
      "eval_accuracy": 0.8166666666666667,
      "eval_loss": 0.7104570865631104,
      "eval_runtime": 3.7843,
      "eval_samples_per_second": 47.565,
      "eval_steps_per_second": 1.585,
      "step": 854
    },
    {
      "epoch": 67.45098039215686,
      "grad_norm": 7.092602729797363,
      "learning_rate": 2.2376543209876542e-05,
      "loss": 0.0754,
      "step": 860
    },
    {
      "epoch": 68.0,
      "eval_accuracy": 0.8111111111111111,
      "eval_loss": 0.675083339214325,
      "eval_runtime": 3.8377,
      "eval_samples_per_second": 46.903,
      "eval_steps_per_second": 1.563,
      "step": 867
    },
    {
      "epoch": 68.23529411764706,
      "grad_norm": 21.711984634399414,
      "learning_rate": 2.1990740740740743e-05,
      "loss": 0.0842,
      "step": 870
    },
    {
      "epoch": 68.94117647058823,
      "eval_accuracy": 0.7833333333333333,
      "eval_loss": 0.6912497878074646,
      "eval_runtime": 4.1438,
      "eval_samples_per_second": 43.438,
      "eval_steps_per_second": 1.448,
      "step": 879
    },
    {
      "epoch": 69.01960784313725,
      "grad_norm": 17.59021759033203,
      "learning_rate": 2.1604938271604937e-05,
      "loss": 0.0815,
      "step": 880
    },
    {
      "epoch": 69.80392156862744,
      "grad_norm": 6.925079822540283,
      "learning_rate": 2.1219135802469135e-05,
      "loss": 0.0684,
      "step": 890
    },
    {
      "epoch": 69.96078431372548,
      "eval_accuracy": 0.8166666666666667,
      "eval_loss": 0.7235284447669983,
      "eval_runtime": 3.8887,
      "eval_samples_per_second": 46.288,
      "eval_steps_per_second": 1.543,
      "step": 892
    },
    {
      "epoch": 70.58823529411765,
      "grad_norm": 10.211894989013672,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.0684,
      "step": 900
    },
    {
      "epoch": 70.98039215686275,
      "eval_accuracy": 0.8277777777777777,
      "eval_loss": 0.5839894413948059,
      "eval_runtime": 3.8698,
      "eval_samples_per_second": 46.514,
      "eval_steps_per_second": 1.55,
      "step": 905
    },
    {
      "epoch": 71.37254901960785,
      "grad_norm": 7.931591987609863,
      "learning_rate": 2.044753086419753e-05,
      "loss": 0.0705,
      "step": 910
    },
    {
      "epoch": 72.0,
      "eval_accuracy": 0.8222222222222222,
      "eval_loss": 0.6635811924934387,
      "eval_runtime": 3.899,
      "eval_samples_per_second": 46.165,
      "eval_steps_per_second": 1.539,
      "step": 918
    },
    {
      "epoch": 72.15686274509804,
      "grad_norm": 8.439017295837402,
      "learning_rate": 2.006172839506173e-05,
      "loss": 0.0532,
      "step": 920
    },
    {
      "epoch": 72.94117647058823,
      "grad_norm": 20.3321475982666,
      "learning_rate": 1.967592592592593e-05,
      "loss": 0.0681,
      "step": 930
    },
    {
      "epoch": 72.94117647058823,
      "eval_accuracy": 0.8,
      "eval_loss": 0.678679883480072,
      "eval_runtime": 3.8456,
      "eval_samples_per_second": 46.807,
      "eval_steps_per_second": 1.56,
      "step": 930
    },
    {
      "epoch": 73.72549019607843,
      "grad_norm": 16.518983840942383,
      "learning_rate": 1.9290123456790123e-05,
      "loss": 0.0906,
      "step": 940
    },
    {
      "epoch": 73.96078431372548,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.6242751479148865,
      "eval_runtime": 3.8813,
      "eval_samples_per_second": 46.376,
      "eval_steps_per_second": 1.546,
      "step": 943
    },
    {
      "epoch": 74.50980392156863,
      "grad_norm": 10.190762519836426,
      "learning_rate": 1.8904320987654324e-05,
      "loss": 0.0453,
      "step": 950
    },
    {
      "epoch": 74.98039215686275,
      "eval_accuracy": 0.8222222222222222,
      "eval_loss": 0.6786649823188782,
      "eval_runtime": 3.8468,
      "eval_samples_per_second": 46.793,
      "eval_steps_per_second": 1.56,
      "step": 956
    },
    {
      "epoch": 75.29411764705883,
      "grad_norm": 12.627336502075195,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.0874,
      "step": 960
    },
    {
      "epoch": 76.0,
      "eval_accuracy": 0.8277777777777777,
      "eval_loss": 0.6259381771087646,
      "eval_runtime": 3.9491,
      "eval_samples_per_second": 45.58,
      "eval_steps_per_second": 1.519,
      "step": 969
    },
    {
      "epoch": 76.07843137254902,
      "grad_norm": 15.190587997436523,
      "learning_rate": 1.8132716049382716e-05,
      "loss": 0.0668,
      "step": 970
    },
    {
      "epoch": 76.86274509803921,
      "grad_norm": 4.534496307373047,
      "learning_rate": 1.7746913580246917e-05,
      "loss": 0.051,
      "step": 980
    },
    {
      "epoch": 76.94117647058823,
      "eval_accuracy": 0.8277777777777777,
      "eval_loss": 0.6590437293052673,
      "eval_runtime": 3.8024,
      "eval_samples_per_second": 47.339,
      "eval_steps_per_second": 1.578,
      "step": 981
    },
    {
      "epoch": 77.6470588235294,
      "grad_norm": 12.094006538391113,
      "learning_rate": 1.736111111111111e-05,
      "loss": 0.0858,
      "step": 990
    },
    {
      "epoch": 77.96078431372548,
      "eval_accuracy": 0.8277777777777777,
      "eval_loss": 0.6306740045547485,
      "eval_runtime": 4.0063,
      "eval_samples_per_second": 44.929,
      "eval_steps_per_second": 1.498,
      "step": 994
    },
    {
      "epoch": 78.43137254901961,
      "grad_norm": 12.669979095458984,
      "learning_rate": 1.697530864197531e-05,
      "loss": 0.0601,
      "step": 1000
    },
    {
      "epoch": 78.98039215686275,
      "eval_accuracy": 0.8444444444444444,
      "eval_loss": 0.6041626334190369,
      "eval_runtime": 3.9261,
      "eval_samples_per_second": 45.847,
      "eval_steps_per_second": 1.528,
      "step": 1007
    },
    {
      "epoch": 79.2156862745098,
      "grad_norm": 13.32419490814209,
      "learning_rate": 1.6589506172839506e-05,
      "loss": 0.0596,
      "step": 1010
    },
    {
      "epoch": 80.0,
      "grad_norm": 17.491554260253906,
      "learning_rate": 1.6203703703703704e-05,
      "loss": 0.0601,
      "step": 1020
    },
    {
      "epoch": 80.0,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.5874945521354675,
      "eval_runtime": 4.0068,
      "eval_samples_per_second": 44.923,
      "eval_steps_per_second": 1.497,
      "step": 1020
    },
    {
      "epoch": 80.7843137254902,
      "grad_norm": 7.148036956787109,
      "learning_rate": 1.58179012345679e-05,
      "loss": 0.067,
      "step": 1030
    },
    {
      "epoch": 80.94117647058823,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.6078370809555054,
      "eval_runtime": 3.7634,
      "eval_samples_per_second": 47.829,
      "eval_steps_per_second": 1.594,
      "step": 1032
    },
    {
      "epoch": 81.56862745098039,
      "grad_norm": 6.120352745056152,
      "learning_rate": 1.54320987654321e-05,
      "loss": 0.0556,
      "step": 1040
    },
    {
      "epoch": 81.96078431372548,
      "eval_accuracy": 0.8444444444444444,
      "eval_loss": 0.6006819605827332,
      "eval_runtime": 3.8921,
      "eval_samples_per_second": 46.248,
      "eval_steps_per_second": 1.542,
      "step": 1045
    },
    {
      "epoch": 82.3529411764706,
      "grad_norm": 9.7392578125,
      "learning_rate": 1.5046296296296297e-05,
      "loss": 0.0661,
      "step": 1050
    },
    {
      "epoch": 82.98039215686275,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.6062378287315369,
      "eval_runtime": 3.7924,
      "eval_samples_per_second": 47.464,
      "eval_steps_per_second": 1.582,
      "step": 1058
    },
    {
      "epoch": 83.13725490196079,
      "grad_norm": 7.788672924041748,
      "learning_rate": 1.4660493827160496e-05,
      "loss": 0.0594,
      "step": 1060
    },
    {
      "epoch": 83.92156862745098,
      "grad_norm": 3.5243284702301025,
      "learning_rate": 1.4274691358024692e-05,
      "loss": 0.0651,
      "step": 1070
    },
    {
      "epoch": 84.0,
      "eval_accuracy": 0.8111111111111111,
      "eval_loss": 0.6387273669242859,
      "eval_runtime": 3.9274,
      "eval_samples_per_second": 45.832,
      "eval_steps_per_second": 1.528,
      "step": 1071
    },
    {
      "epoch": 84.70588235294117,
      "grad_norm": 10.037181854248047,
      "learning_rate": 1.388888888888889e-05,
      "loss": 0.0546,
      "step": 1080
    },
    {
      "epoch": 84.94117647058823,
      "eval_accuracy": 0.8166666666666667,
      "eval_loss": 0.6861324906349182,
      "eval_runtime": 3.8238,
      "eval_samples_per_second": 47.074,
      "eval_steps_per_second": 1.569,
      "step": 1083
    },
    {
      "epoch": 85.49019607843137,
      "grad_norm": 10.713313102722168,
      "learning_rate": 1.3503086419753085e-05,
      "loss": 0.0827,
      "step": 1090
    },
    {
      "epoch": 85.96078431372548,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.6072664260864258,
      "eval_runtime": 4.2373,
      "eval_samples_per_second": 42.479,
      "eval_steps_per_second": 1.416,
      "step": 1096
    },
    {
      "epoch": 86.27450980392157,
      "grad_norm": 7.2301201820373535,
      "learning_rate": 1.3117283950617285e-05,
      "loss": 0.052,
      "step": 1100
    },
    {
      "epoch": 86.98039215686275,
      "eval_accuracy": 0.85,
      "eval_loss": 0.593485951423645,
      "eval_runtime": 3.8302,
      "eval_samples_per_second": 46.995,
      "eval_steps_per_second": 1.566,
      "step": 1109
    },
    {
      "epoch": 87.05882352941177,
      "grad_norm": 4.73368501663208,
      "learning_rate": 1.2731481481481482e-05,
      "loss": 0.0442,
      "step": 1110
    },
    {
      "epoch": 87.84313725490196,
      "grad_norm": 18.29523277282715,
      "learning_rate": 1.2345679012345678e-05,
      "loss": 0.0524,
      "step": 1120
    },
    {
      "epoch": 88.0,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.5899335145950317,
      "eval_runtime": 3.8551,
      "eval_samples_per_second": 46.692,
      "eval_steps_per_second": 1.556,
      "step": 1122
    },
    {
      "epoch": 88.62745098039215,
      "grad_norm": 5.7875494956970215,
      "learning_rate": 1.1959876543209878e-05,
      "loss": 0.066,
      "step": 1130
    },
    {
      "epoch": 88.94117647058823,
      "eval_accuracy": 0.8444444444444444,
      "eval_loss": 0.5954256057739258,
      "eval_runtime": 3.8582,
      "eval_samples_per_second": 46.654,
      "eval_steps_per_second": 1.555,
      "step": 1134
    },
    {
      "epoch": 89.41176470588235,
      "grad_norm": 7.960011005401611,
      "learning_rate": 1.1574074074074075e-05,
      "loss": 0.0617,
      "step": 1140
    },
    {
      "epoch": 89.96078431372548,
      "eval_accuracy": 0.8444444444444444,
      "eval_loss": 0.6145300269126892,
      "eval_runtime": 3.8594,
      "eval_samples_per_second": 46.639,
      "eval_steps_per_second": 1.555,
      "step": 1147
    },
    {
      "epoch": 90.19607843137256,
      "grad_norm": 5.9040350914001465,
      "learning_rate": 1.1188271604938271e-05,
      "loss": 0.0373,
      "step": 1150
    },
    {
      "epoch": 90.98039215686275,
      "grad_norm": 7.361179828643799,
      "learning_rate": 1.0802469135802469e-05,
      "loss": 0.0572,
      "step": 1160
    },
    {
      "epoch": 90.98039215686275,
      "eval_accuracy": 0.8444444444444444,
      "eval_loss": 0.6176372766494751,
      "eval_runtime": 3.825,
      "eval_samples_per_second": 47.059,
      "eval_steps_per_second": 1.569,
      "step": 1160
    },
    {
      "epoch": 91.76470588235294,
      "grad_norm": 14.031915664672852,
      "learning_rate": 1.0416666666666668e-05,
      "loss": 0.0719,
      "step": 1170
    },
    {
      "epoch": 92.0,
      "eval_accuracy": 0.8277777777777777,
      "eval_loss": 0.6406115889549255,
      "eval_runtime": 3.7626,
      "eval_samples_per_second": 47.839,
      "eval_steps_per_second": 1.595,
      "step": 1173
    },
    {
      "epoch": 92.54901960784314,
      "grad_norm": 12.152432441711426,
      "learning_rate": 1.0030864197530866e-05,
      "loss": 0.0734,
      "step": 1180
    },
    {
      "epoch": 92.94117647058823,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.6484689712524414,
      "eval_runtime": 3.8146,
      "eval_samples_per_second": 47.187,
      "eval_steps_per_second": 1.573,
      "step": 1185
    },
    {
      "epoch": 93.33333333333333,
      "grad_norm": 7.38240385055542,
      "learning_rate": 9.645061728395062e-06,
      "loss": 0.0616,
      "step": 1190
    },
    {
      "epoch": 93.96078431372548,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.619816243648529,
      "eval_runtime": 3.8069,
      "eval_samples_per_second": 47.282,
      "eval_steps_per_second": 1.576,
      "step": 1198
    },
    {
      "epoch": 94.11764705882354,
      "grad_norm": 15.435483932495117,
      "learning_rate": 9.259259259259259e-06,
      "loss": 0.047,
      "step": 1200
    },
    {
      "epoch": 94.90196078431373,
      "grad_norm": 3.8711276054382324,
      "learning_rate": 8.873456790123458e-06,
      "loss": 0.0557,
      "step": 1210
    },
    {
      "epoch": 94.98039215686275,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.6167161464691162,
      "eval_runtime": 3.7712,
      "eval_samples_per_second": 47.73,
      "eval_steps_per_second": 1.591,
      "step": 1211
    },
    {
      "epoch": 95.68627450980392,
      "grad_norm": 6.233323574066162,
      "learning_rate": 8.487654320987654e-06,
      "loss": 0.0494,
      "step": 1220
    },
    {
      "epoch": 96.0,
      "eval_accuracy": 0.8444444444444444,
      "eval_loss": 0.6479634642601013,
      "eval_runtime": 3.8139,
      "eval_samples_per_second": 47.196,
      "eval_steps_per_second": 1.573,
      "step": 1224
    },
    {
      "epoch": 96.47058823529412,
      "grad_norm": 2.76275897026062,
      "learning_rate": 8.101851851851852e-06,
      "loss": 0.0587,
      "step": 1230
    },
    {
      "epoch": 96.94117647058823,
      "eval_accuracy": 0.85,
      "eval_loss": 0.6075512170791626,
      "eval_runtime": 3.8781,
      "eval_samples_per_second": 46.414,
      "eval_steps_per_second": 1.547,
      "step": 1236
    },
    {
      "epoch": 97.25490196078431,
      "grad_norm": 6.258754730224609,
      "learning_rate": 7.71604938271605e-06,
      "loss": 0.052,
      "step": 1240
    },
    {
      "epoch": 97.96078431372548,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.6511959433555603,
      "eval_runtime": 3.7767,
      "eval_samples_per_second": 47.661,
      "eval_steps_per_second": 1.589,
      "step": 1249
    },
    {
      "epoch": 98.03921568627452,
      "grad_norm": 2.6569933891296387,
      "learning_rate": 7.330246913580248e-06,
      "loss": 0.0511,
      "step": 1250
    },
    {
      "epoch": 98.82352941176471,
      "grad_norm": 6.5559306144714355,
      "learning_rate": 6.944444444444445e-06,
      "loss": 0.0383,
      "step": 1260
    },
    {
      "epoch": 98.98039215686275,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.6781744360923767,
      "eval_runtime": 3.7667,
      "eval_samples_per_second": 47.787,
      "eval_steps_per_second": 1.593,
      "step": 1262
    },
    {
      "epoch": 99.6078431372549,
      "grad_norm": 2.813370704650879,
      "learning_rate": 6.558641975308642e-06,
      "loss": 0.0499,
      "step": 1270
    },
    {
      "epoch": 100.0,
      "eval_accuracy": 0.8277777777777777,
      "eval_loss": 0.6542291641235352,
      "eval_runtime": 3.8026,
      "eval_samples_per_second": 47.337,
      "eval_steps_per_second": 1.578,
      "step": 1275
    },
    {
      "epoch": 100.3921568627451,
      "grad_norm": 16.95922088623047,
      "learning_rate": 6.172839506172839e-06,
      "loss": 0.0511,
      "step": 1280
    },
    {
      "epoch": 100.94117647058823,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.6794776320457458,
      "eval_runtime": 3.8468,
      "eval_samples_per_second": 46.793,
      "eval_steps_per_second": 1.56,
      "step": 1287
    },
    {
      "epoch": 101.17647058823529,
      "grad_norm": 15.969926834106445,
      "learning_rate": 5.787037037037038e-06,
      "loss": 0.0527,
      "step": 1290
    },
    {
      "epoch": 101.96078431372548,
      "grad_norm": 15.459400177001953,
      "learning_rate": 5.401234567901234e-06,
      "loss": 0.0452,
      "step": 1300
    },
    {
      "epoch": 101.96078431372548,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.6739789247512817,
      "eval_runtime": 3.7395,
      "eval_samples_per_second": 48.135,
      "eval_steps_per_second": 1.605,
      "step": 1300
    },
    {
      "epoch": 102.74509803921569,
      "grad_norm": 5.437152862548828,
      "learning_rate": 5.015432098765433e-06,
      "loss": 0.0475,
      "step": 1310
    },
    {
      "epoch": 102.98039215686275,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.6615740656852722,
      "eval_runtime": 3.7883,
      "eval_samples_per_second": 47.515,
      "eval_steps_per_second": 1.584,
      "step": 1313
    },
    {
      "epoch": 103.52941176470588,
      "grad_norm": 12.730355262756348,
      "learning_rate": 4.6296296296296296e-06,
      "loss": 0.0455,
      "step": 1320
    },
    {
      "epoch": 104.0,
      "eval_accuracy": 0.8277777777777777,
      "eval_loss": 0.6490476131439209,
      "eval_runtime": 3.8157,
      "eval_samples_per_second": 47.173,
      "eval_steps_per_second": 1.572,
      "step": 1326
    },
    {
      "epoch": 104.31372549019608,
      "grad_norm": 10.348357200622559,
      "learning_rate": 4.243827160493827e-06,
      "loss": 0.0486,
      "step": 1330
    },
    {
      "epoch": 104.94117647058823,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.6331196427345276,
      "eval_runtime": 3.7919,
      "eval_samples_per_second": 47.469,
      "eval_steps_per_second": 1.582,
      "step": 1338
    },
    {
      "epoch": 105.09803921568627,
      "grad_norm": 16.444259643554688,
      "learning_rate": 3.858024691358025e-06,
      "loss": 0.0442,
      "step": 1340
    },
    {
      "epoch": 105.88235294117646,
      "grad_norm": 7.596277236938477,
      "learning_rate": 3.4722222222222224e-06,
      "loss": 0.0585,
      "step": 1350
    },
    {
      "epoch": 105.96078431372548,
      "eval_accuracy": 0.8333333333333334,
      "eval_loss": 0.6298839449882507,
      "eval_runtime": 3.7758,
      "eval_samples_per_second": 47.672,
      "eval_steps_per_second": 1.589,
      "step": 1351
    },
    {
      "epoch": 106.66666666666667,
      "grad_norm": 13.853099822998047,
      "learning_rate": 3.0864197530864196e-06,
      "loss": 0.0549,
      "step": 1360
    },
    {
      "epoch": 106.98039215686275,
      "eval_accuracy": 0.8277777777777777,
      "eval_loss": 0.6397578120231628,
      "eval_runtime": 3.7676,
      "eval_samples_per_second": 47.776,
      "eval_steps_per_second": 1.593,
      "step": 1364
    },
    {
      "epoch": 107.45098039215686,
      "grad_norm": 10.11983871459961,
      "learning_rate": 2.700617283950617e-06,
      "loss": 0.0436,
      "step": 1370
    },
    {
      "epoch": 108.0,
      "eval_accuracy": 0.8444444444444444,
      "eval_loss": 0.6338447332382202,
      "eval_runtime": 3.8768,
      "eval_samples_per_second": 46.43,
      "eval_steps_per_second": 1.548,
      "step": 1377
    },
    {
      "epoch": 108.23529411764706,
      "grad_norm": 7.749964237213135,
      "learning_rate": 2.3148148148148148e-06,
      "loss": 0.0429,
      "step": 1380
    },
    {
      "epoch": 108.94117647058823,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.6458715796470642,
      "eval_runtime": 3.7687,
      "eval_samples_per_second": 47.762,
      "eval_steps_per_second": 1.592,
      "step": 1389
    },
    {
      "epoch": 109.01960784313725,
      "grad_norm": 5.384810447692871,
      "learning_rate": 1.9290123456790124e-06,
      "loss": 0.047,
      "step": 1390
    },
    {
      "epoch": 109.80392156862744,
      "grad_norm": 14.666231155395508,
      "learning_rate": 1.5432098765432098e-06,
      "loss": 0.0449,
      "step": 1400
    },
    {
      "epoch": 109.96078431372548,
      "eval_accuracy": 0.8444444444444444,
      "eval_loss": 0.6469634175300598,
      "eval_runtime": 3.785,
      "eval_samples_per_second": 47.556,
      "eval_steps_per_second": 1.585,
      "step": 1402
    },
    {
      "epoch": 110.58823529411765,
      "grad_norm": 11.94315242767334,
      "learning_rate": 1.1574074074074074e-06,
      "loss": 0.0559,
      "step": 1410
    },
    {
      "epoch": 110.98039215686275,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.646262526512146,
      "eval_runtime": 3.8809,
      "eval_samples_per_second": 46.381,
      "eval_steps_per_second": 1.546,
      "step": 1415
    },
    {
      "epoch": 111.37254901960785,
      "grad_norm": 8.830354690551758,
      "learning_rate": 7.716049382716049e-07,
      "loss": 0.0378,
      "step": 1420
    },
    {
      "epoch": 112.0,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.6480041742324829,
      "eval_runtime": 3.8373,
      "eval_samples_per_second": 46.908,
      "eval_steps_per_second": 1.564,
      "step": 1428
    },
    {
      "epoch": 112.15686274509804,
      "grad_norm": 16.867820739746094,
      "learning_rate": 3.8580246913580245e-07,
      "loss": 0.0509,
      "step": 1430
    },
    {
      "epoch": 112.94117647058823,
      "grad_norm": 13.072942733764648,
      "learning_rate": 0.0,
      "loss": 0.0476,
      "step": 1440
    },
    {
      "epoch": 112.94117647058823,
      "eval_accuracy": 0.8388888888888889,
      "eval_loss": 0.6477780342102051,
      "eval_runtime": 3.8103,
      "eval_samples_per_second": 47.24,
      "eval_steps_per_second": 1.575,
      "step": 1440
    },
    {
      "epoch": 112.94117647058823,
      "step": 1440,
      "total_flos": 4.607069541812011e+18,
      "train_loss": 0.33674598841203585,
      "train_runtime": 5395.5485,
      "train_samples_per_second": 36.03,
      "train_steps_per_second": 0.267
    }
  ],
  "logging_steps": 10,
  "max_steps": 1440,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 120,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.607069541812011e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}