{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 168.0,
  "global_step": 672,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.005952380952380952,
      "grad_norm": 180.0,
      "learning_rate": 2.9411764705882356e-07,
      "loss": 2.0328304767608643,
      "step": 1,
      "token_acc": 0.7142857142857143
    },
    {
      "epoch": 0.011904761904761904,
      "grad_norm": 215.0,
      "learning_rate": 5.882352941176471e-07,
      "loss": 2.000833034515381,
      "step": 2,
      "token_acc": 0.7
    },
    {
      "epoch": 0.017857142857142856,
      "grad_norm": 260.0,
      "learning_rate": 8.823529411764707e-07,
      "loss": 2.172398567199707,
      "step": 3,
      "token_acc": 0.6666666666666666
    },
    {
      "epoch": 0.023809523809523808,
      "grad_norm": 180.0,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 2.0073838233947754,
      "step": 4,
      "token_acc": 0.7142857142857143
    },
    {
      "epoch": 0.02976190476190476,
      "grad_norm": 205.0,
      "learning_rate": 1.4705882352941177e-06,
      "loss": 1.9239921569824219,
      "step": 5,
      "token_acc": 0.7222222222222222
    },
    {
      "epoch": 0.03571428571428571,
      "grad_norm": 177.0,
      "learning_rate": 1.7647058823529414e-06,
      "loss": 1.5457267761230469,
      "step": 6,
      "token_acc": 0.7
    },
    {
      "epoch": 0.041666666666666664,
      "grad_norm": 182.0,
      "learning_rate": 2.058823529411765e-06,
      "loss": 1.9132648706436157,
      "step": 7,
      "token_acc": 0.631578947368421
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 183.0,
      "learning_rate": 2.3529411764705885e-06,
      "loss": 1.5335670709609985,
      "step": 8,
      "token_acc": 0.6842105263157895
    },
    {
      "epoch": 0.05357142857142857,
      "grad_norm": 139.0,
      "learning_rate": 2.647058823529412e-06,
      "loss": 0.855933427810669,
      "step": 9,
      "token_acc": 0.8421052631578947
    },
    {
      "epoch": 0.05952380952380952,
      "grad_norm": 146.0,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 1.2274945974349976,
      "step": 10,
      "token_acc": 0.7619047619047619
    },
    {
      "epoch": 0.06547619047619048,
      "grad_norm": 149.0,
      "learning_rate": 3.2352941176470594e-06,
      "loss": 1.2174829244613647,
      "step": 11,
      "token_acc": 0.7777777777777778
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 114.0,
      "learning_rate": 3.529411764705883e-06,
      "loss": 0.48068639636039734,
      "step": 12,
      "token_acc": 0.8333333333333334
    },
    {
      "epoch": 0.07738095238095238,
      "grad_norm": 73.5,
      "learning_rate": 3.8235294117647055e-06,
      "loss": 0.6723679900169373,
      "step": 13,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 78.5,
      "learning_rate": 4.11764705882353e-06,
      "loss": 0.5634061098098755,
      "step": 14,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.08928571428571429,
      "grad_norm": 42.25,
      "learning_rate": 4.411764705882353e-06,
      "loss": 0.1162911206483841,
      "step": 15,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 2.5,
      "learning_rate": 4.705882352941177e-06,
      "loss": 0.0034662913531064987,
      "step": 16,
      "token_acc": 1.0
    },
    {
      "epoch": 0.10119047619047619,
      "grad_norm": 68.0,
      "learning_rate": 5e-06,
      "loss": 0.7013167142868042,
      "step": 17,
      "token_acc": 0.9
    },
    {
      "epoch": 0.10714285714285714,
      "grad_norm": 55.25,
      "learning_rate": 5.294117647058824e-06,
      "loss": 0.31273576617240906,
      "step": 18,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.1130952380952381,
      "grad_norm": 34.75,
      "learning_rate": 5.588235294117647e-06,
      "loss": 0.21721778810024261,
      "step": 19,
      "token_acc": 0.8636363636363636
    },
    {
      "epoch": 0.11904761904761904,
      "grad_norm": 30.375,
      "learning_rate": 5.882352941176471e-06,
      "loss": 0.16250188648700714,
      "step": 20,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.125,
      "grad_norm": 55.0,
      "learning_rate": 6.176470588235295e-06,
      "loss": 0.17018504440784454,
      "step": 21,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.13095238095238096,
      "grad_norm": 19.375,
      "learning_rate": 6.470588235294119e-06,
      "loss": 0.15194657444953918,
      "step": 22,
      "token_acc": 0.95
    },
    {
      "epoch": 0.13690476190476192,
      "grad_norm": 41.5,
      "learning_rate": 6.764705882352942e-06,
      "loss": 0.3741224408149719,
      "step": 23,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 23.25,
      "learning_rate": 7.058823529411766e-06,
      "loss": 0.11306555569171906,
      "step": 24,
      "token_acc": 0.9
    },
    {
      "epoch": 0.1488095238095238,
      "grad_norm": 7.65625,
      "learning_rate": 7.352941176470589e-06,
      "loss": 0.02346169948577881,
      "step": 25,
      "token_acc": 1.0
    },
    {
      "epoch": 0.15476190476190477,
      "grad_norm": 0.494140625,
      "learning_rate": 7.647058823529411e-06,
      "loss": 0.0013536266051232815,
      "step": 26,
      "token_acc": 1.0
    },
    {
      "epoch": 0.16071428571428573,
      "grad_norm": 52.5,
      "learning_rate": 7.941176470588236e-06,
      "loss": 0.5905251502990723,
      "step": 27,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 28.5,
      "learning_rate": 8.23529411764706e-06,
      "loss": 0.20858846604824066,
      "step": 28,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.17261904761904762,
      "grad_norm": 36.75,
      "learning_rate": 8.529411764705883e-06,
      "loss": 0.3382631838321686,
      "step": 29,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 23.125,
      "learning_rate": 8.823529411764707e-06,
      "loss": 0.1937665194272995,
      "step": 30,
      "token_acc": 0.9
    },
    {
      "epoch": 0.18452380952380953,
      "grad_norm": 0.62109375,
      "learning_rate": 9.11764705882353e-06,
      "loss": 0.0014201141893863678,
      "step": 31,
      "token_acc": 1.0
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 16.625,
      "learning_rate": 9.411764705882354e-06,
      "loss": 0.04398690164089203,
      "step": 32,
      "token_acc": 0.95
    },
    {
      "epoch": 0.19642857142857142,
      "grad_norm": 25.75,
      "learning_rate": 9.705882352941177e-06,
      "loss": 0.2503039836883545,
      "step": 33,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.20238095238095238,
      "grad_norm": 11.625,
      "learning_rate": 1e-05,
      "loss": 0.04316756874322891,
      "step": 34,
      "token_acc": 0.95
    },
    {
      "epoch": 0.20833333333333334,
      "grad_norm": 34.5,
      "learning_rate": 9.999939382570075e-06,
      "loss": 0.39119866490364075,
      "step": 35,
      "token_acc": 0.95
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 25.125,
      "learning_rate": 9.999757531750086e-06,
      "loss": 0.2241765409708023,
      "step": 36,
      "token_acc": 0.9090909090909091
    },
    {
      "epoch": 0.22023809523809523,
      "grad_norm": 13.9375,
      "learning_rate": 9.999454451949364e-06,
      "loss": 0.0885719284415245,
      "step": 37,
      "token_acc": 0.95
    },
    {
      "epoch": 0.2261904761904762,
      "grad_norm": 48.0,
      "learning_rate": 9.999030150516681e-06,
      "loss": 0.5089930891990662,
      "step": 38,
      "token_acc": 0.9047619047619048
    },
    {
      "epoch": 0.23214285714285715,
      "grad_norm": 11.5625,
      "learning_rate": 9.998484637740058e-06,
      "loss": 0.09222182631492615,
      "step": 39,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 51.0,
      "learning_rate": 9.997817926846528e-06,
      "loss": 0.4717664122581482,
      "step": 40,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.24404761904761904,
      "grad_norm": 9.9375,
      "learning_rate": 9.997030034001815e-06,
      "loss": 0.07856762409210205,
      "step": 41,
      "token_acc": 0.95
    },
    {
      "epoch": 0.25,
      "grad_norm": 23.0,
      "learning_rate": 9.99612097830993e-06,
      "loss": 0.19967058300971985,
      "step": 42,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 0.25595238095238093,
      "grad_norm": 27.5,
      "learning_rate": 9.995090781812724e-06,
      "loss": 0.23093397915363312,
      "step": 43,
      "token_acc": 0.85
    },
    {
      "epoch": 0.2619047619047619,
      "grad_norm": 25.5,
      "learning_rate": 9.993939469489342e-06,
      "loss": 0.28952309489250183,
      "step": 44,
      "token_acc": 0.8095238095238095
    },
    {
      "epoch": 0.26785714285714285,
      "grad_norm": 26.25,
      "learning_rate": 9.99266706925562e-06,
      "loss": 0.24939370155334473,
      "step": 45,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.27380952380952384,
      "grad_norm": 32.75,
      "learning_rate": 9.991273611963413e-06,
      "loss": 0.4245058298110962,
      "step": 46,
      "token_acc": 0.8421052631578947
    },
    {
      "epoch": 0.27976190476190477,
      "grad_norm": 20.375,
      "learning_rate": 9.98975913139984e-06,
      "loss": 0.15520256757736206,
      "step": 47,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 12.4375,
      "learning_rate": 9.98812366428647e-06,
      "loss": 0.0844191312789917,
      "step": 48,
      "token_acc": 1.0
    },
    {
      "epoch": 0.2916666666666667,
      "grad_norm": 25.375,
      "learning_rate": 9.986367250278423e-06,
      "loss": 0.25481945276260376,
      "step": 49,
      "token_acc": 0.85
    },
    {
      "epoch": 0.2976190476190476,
      "grad_norm": 15.8125,
      "learning_rate": 9.984489931963429e-06,
      "loss": 0.16055771708488464,
      "step": 50,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.30357142857142855,
      "grad_norm": 25.5,
      "learning_rate": 9.982491754860763e-06,
      "loss": 0.31199318170547485,
      "step": 51,
      "token_acc": 0.8421052631578947
    },
    {
      "epoch": 0.30952380952380953,
      "grad_norm": 18.625,
      "learning_rate": 9.980372767420179e-06,
      "loss": 0.27254724502563477,
      "step": 52,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.31547619047619047,
      "grad_norm": 17.0,
      "learning_rate": 9.978133021020697e-06,
      "loss": 0.16340498626232147,
      "step": 53,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.32142857142857145,
      "grad_norm": 19.125,
      "learning_rate": 9.97577256996939e-06,
      "loss": 0.2168363332748413,
      "step": 54,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.3273809523809524,
      "grad_norm": 20.625,
      "learning_rate": 9.97329147150005e-06,
      "loss": 0.3217720687389374,
      "step": 55,
      "token_acc": 0.7894736842105263
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 16.0,
      "learning_rate": 9.970689785771798e-06,
      "loss": 0.16012439131736755,
      "step": 56,
      "token_acc": 0.9
    },
    {
      "epoch": 0.3392857142857143,
      "grad_norm": 15.3125,
      "learning_rate": 9.96796757586764e-06,
      "loss": 0.1560428887605667,
      "step": 57,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.34523809523809523,
      "grad_norm": 15.375,
      "learning_rate": 9.965124907792916e-06,
      "loss": 0.20544831454753876,
      "step": 58,
      "token_acc": 0.95
    },
    {
      "epoch": 0.35119047619047616,
      "grad_norm": 18.0,
      "learning_rate": 9.962161850473723e-06,
      "loss": 0.24054399132728577,
      "step": 59,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 14.0,
      "learning_rate": 9.95907847575523e-06,
      "loss": 0.12745679914951324,
      "step": 60,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.3630952380952381,
      "grad_norm": 15.5,
      "learning_rate": 9.955874858399936e-06,
      "loss": 0.26499828696250916,
      "step": 61,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.36904761904761907,
      "grad_norm": 17.5,
      "learning_rate": 9.952551076085864e-06,
      "loss": 0.2377898097038269,
      "step": 62,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.375,
      "grad_norm": 16.25,
      "learning_rate": 9.949107209404664e-06,
      "loss": 0.24094776809215546,
      "step": 63,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 21.625,
      "learning_rate": 9.945543341859681e-06,
      "loss": 0.31726279854774475,
      "step": 64,
      "token_acc": 0.9
    },
    {
      "epoch": 0.3869047619047619,
      "grad_norm": 9.75,
      "learning_rate": 9.94185955986391e-06,
      "loss": 0.0700460895895958,
      "step": 65,
      "token_acc": 1.0
    },
    {
      "epoch": 0.39285714285714285,
      "grad_norm": 13.1875,
      "learning_rate": 9.938055952737908e-06,
      "loss": 0.1853027641773224,
      "step": 66,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 0.39880952380952384,
      "grad_norm": 15.25,
      "learning_rate": 9.934132612707631e-06,
      "loss": 0.2067536860704422,
      "step": 67,
      "token_acc": 0.9
    },
    {
      "epoch": 0.40476190476190477,
      "grad_norm": 21.625,
      "learning_rate": 9.930089634902197e-06,
      "loss": 0.2864565849304199,
      "step": 68,
      "token_acc": 0.9090909090909091
    },
    {
      "epoch": 0.4107142857142857,
      "grad_norm": 18.375,
      "learning_rate": 9.925927117351573e-06,
      "loss": 0.20295722782611847,
      "step": 69,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 28.875,
      "learning_rate": 9.921645160984205e-06,
      "loss": 0.5202264189720154,
      "step": 70,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.4226190476190476,
      "grad_norm": 12.0,
      "learning_rate": 9.917243869624573e-06,
      "loss": 0.12216765433549881,
      "step": 71,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 13.6875,
      "learning_rate": 9.91272334999066e-06,
      "loss": 0.16414539515972137,
      "step": 72,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 0.43452380952380953,
      "grad_norm": 14.5,
      "learning_rate": 9.908083711691383e-06,
      "loss": 0.18212856352329254,
      "step": 73,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.44047619047619047,
      "grad_norm": 17.25,
      "learning_rate": 9.903325067223918e-06,
      "loss": 0.2009446918964386,
      "step": 74,
      "token_acc": 0.9
    },
    {
      "epoch": 0.44642857142857145,
      "grad_norm": 13.0625,
      "learning_rate": 9.898447531970989e-06,
      "loss": 0.08332519978284836,
      "step": 75,
      "token_acc": 0.95
    },
    {
      "epoch": 0.4523809523809524,
      "grad_norm": 5.9375,
      "learning_rate": 9.893451224198051e-06,
      "loss": 0.10153824090957642,
      "step": 76,
      "token_acc": 0.95
    },
    {
      "epoch": 0.4583333333333333,
      "grad_norm": 9.0625,
      "learning_rate": 9.888336265050443e-06,
      "loss": 0.08780962228775024,
      "step": 77,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.4642857142857143,
      "grad_norm": 12.9375,
      "learning_rate": 9.883102778550434e-06,
      "loss": 0.15168383717536926,
      "step": 78,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.47023809523809523,
      "grad_norm": 14.4375,
      "learning_rate": 9.877750891594224e-06,
      "loss": 0.12056589126586914,
      "step": 79,
      "token_acc": 0.95
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 15.5625,
      "learning_rate": 9.872280733948867e-06,
      "loss": 0.17763763666152954,
      "step": 80,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.48214285714285715,
      "grad_norm": 20.125,
      "learning_rate": 9.866692438249124e-06,
      "loss": 0.2938810884952545,
      "step": 81,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.4880952380952381,
      "grad_norm": 26.0,
      "learning_rate": 9.86098613999424e-06,
      "loss": 0.3427335321903229,
      "step": 82,
      "token_acc": 0.9047619047619048
    },
    {
      "epoch": 0.49404761904761907,
      "grad_norm": 14.1875,
      "learning_rate": 9.855161977544672e-06,
      "loss": 0.20410454273223877,
      "step": 83,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.125,
      "learning_rate": 9.849220092118721e-06,
      "loss": 0.07883279770612717,
      "step": 84,
      "token_acc": 1.0
    },
    {
      "epoch": 0.5059523809523809,
      "grad_norm": 19.375,
      "learning_rate": 9.84316062778912e-06,
      "loss": 0.1371012032032013,
      "step": 85,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.5119047619047619,
      "grad_norm": 21.5,
      "learning_rate": 9.836983731479526e-06,
      "loss": 0.2181922197341919,
      "step": 86,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.5178571428571429,
      "grad_norm": 20.5,
      "learning_rate": 9.830689552960974e-06,
      "loss": 0.19635120034217834,
      "step": 87,
      "token_acc": 0.9
    },
    {
      "epoch": 0.5238095238095238,
      "grad_norm": 19.625,
      "learning_rate": 9.824278244848236e-06,
      "loss": 0.1989951729774475,
      "step": 88,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.5297619047619048,
      "grad_norm": 21.625,
      "learning_rate": 9.817749962596115e-06,
      "loss": 0.19838982820510864,
      "step": 89,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.5357142857142857,
      "grad_norm": 16.875,
      "learning_rate": 9.811104864495691e-06,
      "loss": 0.12101458013057709,
      "step": 90,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.5416666666666666,
      "grad_norm": 10.3125,
      "learning_rate": 9.804343111670472e-06,
      "loss": 0.04938970133662224,
      "step": 91,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.5476190476190477,
      "grad_norm": 20.375,
      "learning_rate": 9.797464868072489e-06,
      "loss": 0.3261977732181549,
      "step": 92,
      "token_acc": 0.9
    },
    {
      "epoch": 0.5535714285714286,
      "grad_norm": 23.125,
      "learning_rate": 9.790470300478318e-06,
      "loss": 0.23016251623630524,
      "step": 93,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.5595238095238095,
      "grad_norm": 15.5625,
      "learning_rate": 9.783359578485047e-06,
      "loss": 0.1902378648519516,
      "step": 94,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.5654761904761905,
      "grad_norm": 8.4375,
      "learning_rate": 9.776132874506153e-06,
      "loss": 0.037527915090322495,
      "step": 95,
      "token_acc": 1.0
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 17.125,
      "learning_rate": 9.768790363767321e-06,
      "loss": 0.23314255475997925,
      "step": 96,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.5773809523809523,
      "grad_norm": 8.0625,
      "learning_rate": 9.761332224302209e-06,
      "loss": 0.08382070809602737,
      "step": 97,
      "token_acc": 1.0
    },
    {
      "epoch": 0.5833333333333334,
      "grad_norm": 12.5,
      "learning_rate": 9.753758636948112e-06,
      "loss": 0.16923876106739044,
      "step": 98,
      "token_acc": 0.8571428571428571
    },
    {
      "epoch": 0.5892857142857143,
      "grad_norm": 16.625,
      "learning_rate": 9.74606978534159e-06,
      "loss": 0.1556047648191452,
      "step": 99,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 0.5952380952380952,
      "grad_norm": 16.625,
      "learning_rate": 9.738265855914014e-06,
      "loss": 0.11820109933614731,
      "step": 100,
      "token_acc": 0.9047619047619048
    },
    {
      "epoch": 0.6011904761904762,
      "grad_norm": 16.5,
      "learning_rate": 9.730347037887041e-06,
      "loss": 0.1404218077659607,
      "step": 101,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.6071428571428571,
      "grad_norm": 6.96875,
      "learning_rate": 9.722313523268028e-06,
      "loss": 0.052532874047756195,
      "step": 102,
      "token_acc": 1.0
    },
    {
      "epoch": 0.6130952380952381,
      "grad_norm": 12.1875,
      "learning_rate": 9.714165506845381e-06,
      "loss": 0.08662351220846176,
      "step": 103,
      "token_acc": 0.95
    },
    {
      "epoch": 0.6190476190476191,
      "grad_norm": 16.125,
      "learning_rate": 9.705903186183828e-06,
      "loss": 0.14111609756946564,
      "step": 104,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.625,
      "grad_norm": 15.3125,
      "learning_rate": 9.697526761619621e-06,
      "loss": 0.08410780876874924,
      "step": 105,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.6309523809523809,
      "grad_norm": 25.0,
      "learning_rate": 9.689036436255698e-06,
      "loss": 0.3091953694820404,
      "step": 106,
      "token_acc": 0.9090909090909091
    },
    {
      "epoch": 0.6369047619047619,
      "grad_norm": 1.90625,
      "learning_rate": 9.680432415956736e-06,
      "loss": 0.005599863827228546,
      "step": 107,
      "token_acc": 1.0
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 23.75,
      "learning_rate": 9.671714909344175e-06,
      "loss": 0.1167428269982338,
      "step": 108,
      "token_acc": 0.95
    },
    {
      "epoch": 0.6488095238095238,
      "grad_norm": 34.0,
      "learning_rate": 9.66288412779115e-06,
      "loss": 0.25187408924102783,
      "step": 109,
      "token_acc": 0.9047619047619048
    },
    {
      "epoch": 0.6547619047619048,
      "grad_norm": 13.1875,
      "learning_rate": 9.653940285417381e-06,
      "loss": 0.055955931544303894,
      "step": 110,
      "token_acc": 1.0
    },
    {
      "epoch": 0.6607142857142857,
      "grad_norm": 23.25,
      "learning_rate": 9.644883599083959e-06,
      "loss": 0.19123758375644684,
      "step": 111,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 9.5625,
      "learning_rate": 9.635714288388103e-06,
      "loss": 0.07087983936071396,
      "step": 112,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.6726190476190477,
      "grad_norm": 17.5,
      "learning_rate": 9.626432575657834e-06,
      "loss": 0.13504379987716675,
      "step": 113,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 0.6785714285714286,
      "grad_norm": 27.5,
      "learning_rate": 9.617038685946578e-06,
      "loss": 0.372283399105072,
      "step": 114,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.6845238095238095,
      "grad_norm": 45.0,
      "learning_rate": 9.60753284702772e-06,
      "loss": 0.25675544142723083,
      "step": 115,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.6904761904761905,
      "grad_norm": 21.125,
      "learning_rate": 9.597915289389067e-06,
      "loss": 0.26502567529678345,
      "step": 116,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 0.6964285714285714,
      "grad_norm": 56.25,
      "learning_rate": 9.58818624622727e-06,
      "loss": 0.6296467781066895,
      "step": 117,
      "token_acc": 0.8421052631578947
    },
    {
      "epoch": 0.7023809523809523,
      "grad_norm": 18.5,
      "learning_rate": 9.578345953442163e-06,
      "loss": 0.1636037677526474,
      "step": 118,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.7083333333333334,
      "grad_norm": 20.0,
      "learning_rate": 9.568394649631055e-06,
      "loss": 0.19119992852210999,
      "step": 119,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 13.9375,
      "learning_rate": 9.558332576082925e-06,
      "loss": 0.05579443275928497,
      "step": 120,
      "token_acc": 1.0
    },
    {
      "epoch": 0.7202380952380952,
      "grad_norm": 17.375,
      "learning_rate": 9.548159976772593e-06,
      "loss": 0.10840553790330887,
      "step": 121,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 0.7261904761904762,
      "grad_norm": 26.25,
      "learning_rate": 9.537877098354787e-06,
      "loss": 0.22527143359184265,
      "step": 122,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.7321428571428571,
      "grad_norm": 24.25,
      "learning_rate": 9.527484190158171e-06,
      "loss": 0.21126054227352142,
      "step": 123,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 0.7380952380952381,
      "grad_norm": 10.375,
      "learning_rate": 9.5169815041793e-06,
      "loss": 0.04842006787657738,
      "step": 124,
      "token_acc": 1.0
    },
    {
      "epoch": 0.7440476190476191,
      "grad_norm": 32.25,
      "learning_rate": 9.506369295076505e-06,
      "loss": 0.15709196031093597,
      "step": 125,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.75,
      "grad_norm": 24.125,
      "learning_rate": 9.495647820163725e-06,
      "loss": 0.18596792221069336,
      "step": 126,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 0.7559523809523809,
      "grad_norm": 24.5,
      "learning_rate": 9.484817339404261e-06,
      "loss": 0.20715166628360748,
      "step": 127,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 7.875,
      "learning_rate": 9.473878115404477e-06,
      "loss": 0.04500706493854523,
      "step": 128,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.7678571428571429,
      "grad_norm": 6.34375,
      "learning_rate": 9.462830413407427e-06,
      "loss": 0.023552263155579567,
      "step": 129,
      "token_acc": 1.0
    },
    {
      "epoch": 0.7738095238095238,
      "grad_norm": 14.5,
      "learning_rate": 9.451674501286436e-06,
      "loss": 0.09156066179275513,
      "step": 130,
      "token_acc": 0.9565217391304348
    },
    {
      "epoch": 0.7797619047619048,
      "grad_norm": 13.1875,
      "learning_rate": 9.440410649538592e-06,
      "loss": 0.13985063135623932,
      "step": 131,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 25.125,
      "learning_rate": 9.42903913127819e-06,
      "loss": 0.2945712208747864,
      "step": 132,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.7916666666666666,
      "grad_norm": 21.875,
      "learning_rate": 9.417560222230115e-06,
      "loss": 0.15177886188030243,
      "step": 133,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 0.7976190476190477,
      "grad_norm": 6.375,
      "learning_rate": 9.405974200723156e-06,
      "loss": 0.043519552797079086,
      "step": 134,
      "token_acc": 1.0
    },
    {
      "epoch": 0.8035714285714286,
      "grad_norm": 15.875,
      "learning_rate": 9.394281347683247e-06,
      "loss": 0.09082482755184174,
      "step": 135,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.8095238095238095,
      "grad_norm": 4.25,
      "learning_rate": 9.382481946626673e-06,
      "loss": 0.016082266345620155,
      "step": 136,
      "token_acc": 1.0
    },
    {
      "epoch": 0.8154761904761905,
      "grad_norm": 23.25,
      "learning_rate": 9.370576283653178e-06,
      "loss": 0.25259846448898315,
      "step": 137,
      "token_acc": 0.8636363636363636
    },
    {
      "epoch": 0.8214285714285714,
      "grad_norm": 1.484375,
      "learning_rate": 9.358564647439037e-06,
      "loss": 0.0049808272160589695,
      "step": 138,
      "token_acc": 1.0
    },
    {
      "epoch": 0.8273809523809523,
      "grad_norm": 17.875,
      "learning_rate": 9.34644732923006e-06,
      "loss": 0.17535477876663208,
      "step": 139,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 21.125,
      "learning_rate": 9.33422462283452e-06,
      "loss": 0.20156142115592957,
      "step": 140,
      "token_acc": 0.9
    },
    {
      "epoch": 0.8392857142857143,
      "grad_norm": 18.375,
      "learning_rate": 9.321896824616036e-06,
      "loss": 0.12329314649105072,
      "step": 141,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.8452380952380952,
      "grad_norm": 37.0,
      "learning_rate": 9.309464233486386e-06,
      "loss": 0.25263017416000366,
      "step": 142,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.8511904761904762,
      "grad_norm": 23.625,
      "learning_rate": 9.29692715089826e-06,
      "loss": 0.32131433486938477,
      "step": 143,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 11.9375,
      "learning_rate": 9.284285880837947e-06,
      "loss": 0.09042059630155563,
      "step": 144,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 0.8630952380952381,
      "grad_norm": 14.4375,
      "learning_rate": 9.271540729817969e-06,
      "loss": 0.14959029853343964,
      "step": 145,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.8690476190476191,
      "grad_norm": 20.75,
      "learning_rate": 9.258692006869644e-06,
      "loss": 0.15013572573661804,
      "step": 146,
      "token_acc": 0.95
    },
    {
      "epoch": 0.875,
      "grad_norm": 19.875,
      "learning_rate": 9.245740023535596e-06,
      "loss": 0.10887682437896729,
      "step": 147,
      "token_acc": 0.95
    },
    {
      "epoch": 0.8809523809523809,
      "grad_norm": 11.6875,
      "learning_rate": 9.232685093862206e-06,
      "loss": 0.07650676369667053,
      "step": 148,
      "token_acc": 1.0
    },
    {
      "epoch": 0.8869047619047619,
      "grad_norm": 27.375,
      "learning_rate": 9.219527534391983e-06,
      "loss": 0.13167469203472137,
      "step": 149,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 0.8928571428571429,
      "grad_norm": 8.375,
      "learning_rate": 9.206267664155906e-06,
      "loss": 0.0321711041033268,
      "step": 150,
      "token_acc": 1.0
    },
    {
      "epoch": 0.8988095238095238,
      "grad_norm": 2.375,
      "learning_rate": 9.192905804665677e-06,
      "loss": 0.009442750364542007,
      "step": 151,
      "token_acc": 1.0
    },
    {
      "epoch": 0.9047619047619048,
      "grad_norm": 16.125,
      "learning_rate": 9.179442279905927e-06,
      "loss": 0.08805715292692184,
      "step": 152,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 0.9107142857142857,
      "grad_norm": 21.75,
      "learning_rate": 9.165877416326365e-06,
      "loss": 0.0804082527756691,
      "step": 153,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.9166666666666666,
      "grad_norm": 2.765625,
      "learning_rate": 9.152211542833856e-06,
      "loss": 0.010968723334372044,
      "step": 154,
      "token_acc": 1.0
    },
    {
      "epoch": 0.9226190476190477,
      "grad_norm": 10.75,
      "learning_rate": 9.138444990784455e-06,
      "loss": 0.07799652218818665,
      "step": 155,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 0.9285714285714286,
      "grad_norm": 34.0,
      "learning_rate": 9.124578093975358e-06,
      "loss": 0.20043228566646576,
      "step": 156,
      "token_acc": 0.9090909090909091
    },
    {
      "epoch": 0.9345238095238095,
      "grad_norm": 11.3125,
      "learning_rate": 9.110611188636828e-06,
      "loss": 0.08889784663915634,
      "step": 157,
      "token_acc": 0.9565217391304348
    },
    {
      "epoch": 0.9404761904761905,
      "grad_norm": 26.25,
      "learning_rate": 9.096544613424026e-06,
      "loss": 0.2805810570716858,
      "step": 158,
      "token_acc": 0.9
    },
    {
      "epoch": 0.9464285714285714,
      "grad_norm": 25.25,
      "learning_rate": 9.082378709408805e-06,
      "loss": 0.19485194981098175,
      "step": 159,
      "token_acc": 0.8571428571428571
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 26.0,
      "learning_rate": 9.068113820071447e-06,
      "loss": 0.32104992866516113,
      "step": 160,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.9583333333333334,
      "grad_norm": 27.75,
      "learning_rate": 9.053750291292321e-06,
      "loss": 0.09421157091856003,
      "step": 161,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 0.9642857142857143,
      "grad_norm": 12.75,
      "learning_rate": 9.039288471343505e-06,
      "loss": 0.11460894346237183,
      "step": 162,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 0.9702380952380952,
      "grad_norm": 2.03125,
      "learning_rate": 9.024728710880345e-06,
      "loss": 0.005733321886509657,
      "step": 163,
      "token_acc": 1.0
    },
    {
      "epoch": 0.9761904761904762,
      "grad_norm": 15.3125,
      "learning_rate": 9.010071362932945e-06,
      "loss": 0.08128063380718231,
      "step": 164,
      "token_acc": 0.95
    },
    {
      "epoch": 0.9821428571428571,
      "grad_norm": 11.125,
      "learning_rate": 8.995316782897605e-06,
      "loss": 0.07053440809249878,
      "step": 165,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 0.9880952380952381,
      "grad_norm": 7.71875,
      "learning_rate": 8.98046532852822e-06,
      "loss": 0.020816773176193237,
      "step": 166,
      "token_acc": 1.0
    },
    {
      "epoch": 0.9940476190476191,
      "grad_norm": 20.875,
      "learning_rate": 8.965517359927583e-06,
      "loss": 0.13768719136714935,
      "step": 167,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.359375,
      "learning_rate": 8.950473239538672e-06,
      "loss": 0.010362718254327774,
      "step": 168,
      "token_acc": 1.0
    },
    {
      "epoch": 1.005952380952381,
      "grad_norm": 10.5,
      "learning_rate": 8.935333332135853e-06,
      "loss": 0.06425925344228745,
      "step": 169,
      "token_acc": 1.0
    },
    {
      "epoch": 1.0119047619047619,
      "grad_norm": 10.6875,
      "learning_rate": 8.920098004816035e-06,
      "loss": 0.1227460503578186,
      "step": 170,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.0178571428571428,
      "grad_norm": 16.375,
      "learning_rate": 8.904767626989774e-06,
      "loss": 0.10397586226463318,
      "step": 171,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.0238095238095237,
      "grad_norm": 3.90625,
      "learning_rate": 8.88934257037231e-06,
      "loss": 0.007945025339722633,
      "step": 172,
      "token_acc": 1.0
    },
    {
      "epoch": 1.0297619047619047,
      "grad_norm": 23.5,
      "learning_rate": 8.873823208974557e-06,
      "loss": 0.33320122957229614,
      "step": 173,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.0357142857142858,
      "grad_norm": 23.375,
      "learning_rate": 8.85820991909404e-06,
      "loss": 0.13462264835834503,
      "step": 174,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 1.0416666666666667,
      "grad_norm": 14.625,
      "learning_rate": 8.842503079305757e-06,
      "loss": 0.06944146752357483,
      "step": 175,
      "token_acc": 1.0
    },
    {
      "epoch": 1.0476190476190477,
      "grad_norm": 11.75,
      "learning_rate": 8.826703070453014e-06,
      "loss": 0.07360465079545975,
      "step": 176,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.0535714285714286,
      "grad_norm": 22.375,
      "learning_rate": 8.810810275638183e-06,
      "loss": 0.19790047407150269,
      "step": 177,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 1.0595238095238095,
      "grad_norm": 3.453125,
      "learning_rate": 8.794825080213415e-06,
      "loss": 0.01580767333507538,
      "step": 178,
      "token_acc": 1.0
    },
    {
      "epoch": 1.0654761904761905,
      "grad_norm": 8.375,
      "learning_rate": 8.778747871771293e-06,
      "loss": 0.02923252433538437,
      "step": 179,
      "token_acc": 1.0
    },
    {
      "epoch": 1.0714285714285714,
      "grad_norm": 7.28125,
      "learning_rate": 8.76257904013544e-06,
      "loss": 0.04556063935160637,
      "step": 180,
      "token_acc": 0.95
    },
    {
      "epoch": 1.0773809523809523,
      "grad_norm": 24.5,
      "learning_rate": 8.746318977351066e-06,
      "loss": 0.17441561818122864,
      "step": 181,
      "token_acc": 0.9
    },
    {
      "epoch": 1.0833333333333333,
      "grad_norm": 7.40625,
      "learning_rate": 8.729968077675454e-06,
      "loss": 0.030688166618347168,
      "step": 182,
      "token_acc": 1.0
    },
    {
      "epoch": 1.0892857142857142,
      "grad_norm": 23.875,
      "learning_rate": 8.713526737568415e-06,
      "loss": 0.17615407705307007,
      "step": 183,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.0952380952380953,
      "grad_norm": 18.75,
      "learning_rate": 8.696995355682656e-06,
      "loss": 0.15877707302570343,
      "step": 184,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 1.1011904761904763,
      "grad_norm": 28.125,
      "learning_rate": 8.680374332854134e-06,
      "loss": 0.16839782893657684,
      "step": 185,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.1071428571428572,
      "grad_norm": 27.875,
      "learning_rate": 8.663664072092324e-06,
      "loss": 0.30390995740890503,
      "step": 186,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.1130952380952381,
      "grad_norm": 13.5625,
      "learning_rate": 8.646864978570445e-06,
      "loss": 0.06602757424116135,
      "step": 187,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.119047619047619,
      "grad_norm": 26.25,
      "learning_rate": 8.629977459615655e-06,
      "loss": 0.17160753905773163,
      "step": 188,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.125,
      "grad_norm": 30.625,
      "learning_rate": 8.613001924699146e-06,
      "loss": 0.18934032320976257,
      "step": 189,
      "token_acc": 0.9
    },
    {
      "epoch": 1.130952380952381,
      "grad_norm": 3.421875,
      "learning_rate": 8.595938785426241e-06,
      "loss": 0.005042365752160549,
      "step": 190,
      "token_acc": 1.0
    },
    {
      "epoch": 1.1369047619047619,
      "grad_norm": 0.059814453125,
      "learning_rate": 8.578788455526398e-06,
      "loss": 0.00020425590628292412,
      "step": 191,
      "token_acc": 1.0
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 7.84375,
      "learning_rate": 8.561551350843185e-06,
      "loss": 0.04840868338942528,
      "step": 192,
      "token_acc": 1.0
    },
    {
      "epoch": 1.1488095238095237,
      "grad_norm": 3.6875,
      "learning_rate": 8.544227889324199e-06,
      "loss": 0.009256002493202686,
      "step": 193,
      "token_acc": 1.0
    },
    {
      "epoch": 1.1547619047619047,
      "grad_norm": 0.031494140625,
      "learning_rate": 8.526818491010922e-06,
      "loss": 0.00011041459219995886,
      "step": 194,
      "token_acc": 1.0
    },
    {
      "epoch": 1.1607142857142858,
      "grad_norm": 0.06298828125,
      "learning_rate": 8.509323578028547e-06,
      "loss": 0.0001653869985602796,
      "step": 195,
      "token_acc": 1.0
    },
    {
      "epoch": 1.1666666666666667,
      "grad_norm": 23.0,
      "learning_rate": 8.491743574575743e-06,
      "loss": 0.29463639855384827,
      "step": 196,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.1726190476190477,
      "grad_norm": 3.296875,
      "learning_rate": 8.474078906914359e-06,
      "loss": 0.009911641478538513,
      "step": 197,
      "token_acc": 1.0
    },
    {
      "epoch": 1.1785714285714286,
      "grad_norm": 1.7890625,
      "learning_rate": 8.456330003359093e-06,
      "loss": 0.005750651005655527,
      "step": 198,
      "token_acc": 1.0
    },
    {
      "epoch": 1.1845238095238095,
      "grad_norm": 17.375,
      "learning_rate": 8.438497294267117e-06,
      "loss": 0.03297002986073494,
      "step": 199,
      "token_acc": 1.0
    },
    {
      "epoch": 1.1904761904761905,
      "grad_norm": 0.765625,
      "learning_rate": 8.420581212027625e-06,
      "loss": 0.0009522189502604306,
      "step": 200,
      "token_acc": 1.0
    },
    {
      "epoch": 1.1964285714285714,
      "grad_norm": 3.5625,
      "learning_rate": 8.402582191051365e-06,
      "loss": 0.007570045534521341,
      "step": 201,
      "token_acc": 1.0
    },
    {
      "epoch": 1.2023809523809523,
      "grad_norm": 31.375,
      "learning_rate": 8.38450066776009e-06,
      "loss": 0.12239433825016022,
      "step": 202,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 1.2083333333333333,
      "grad_norm": 4.6875,
      "learning_rate": 8.36633708057599e-06,
      "loss": 0.012513522058725357,
      "step": 203,
      "token_acc": 1.0
    },
    {
      "epoch": 1.2142857142857142,
      "grad_norm": 20.125,
      "learning_rate": 8.348091869911054e-06,
      "loss": 0.1393841803073883,
      "step": 204,
      "token_acc": 0.9130434782608695
    },
    {
      "epoch": 1.2202380952380953,
      "grad_norm": 1.8046875,
      "learning_rate": 8.329765478156394e-06,
      "loss": 0.003990611527115107,
      "step": 205,
      "token_acc": 1.0
    },
    {
      "epoch": 1.2261904761904763,
      "grad_norm": 26.75,
      "learning_rate": 8.311358349671516e-06,
      "loss": 0.08855653554201126,
      "step": 206,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 1.2321428571428572,
      "grad_norm": 29.0,
      "learning_rate": 8.292870930773551e-06,
      "loss": 0.13506564497947693,
      "step": 207,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 1.2380952380952381,
      "grad_norm": 2.5,
      "learning_rate": 8.274303669726427e-06,
      "loss": 0.0079928208142519,
      "step": 208,
      "token_acc": 1.0
    },
    {
      "epoch": 1.244047619047619,
      "grad_norm": 16.75,
      "learning_rate": 8.255657016729997e-06,
      "loss": 0.04852926731109619,
      "step": 209,
      "token_acc": 1.0
    },
    {
      "epoch": 1.25,
      "grad_norm": 30.125,
      "learning_rate": 8.23693142390914e-06,
      "loss": 0.4011387526988983,
      "step": 210,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 1.255952380952381,
      "grad_norm": 6.6875,
      "learning_rate": 8.218127345302775e-06,
      "loss": 0.02324041537940502,
      "step": 211,
      "token_acc": 1.0
    },
    {
      "epoch": 1.2619047619047619,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.199245236852871e-06,
      "loss": 0.00033977616112679243,
      "step": 212,
      "token_acc": 1.0
    },
    {
      "epoch": 1.2678571428571428,
      "grad_norm": 1.7109375,
      "learning_rate": 8.180285556393384e-06,
      "loss": 0.0018349780002608895,
      "step": 213,
      "token_acc": 1.0
    },
    {
      "epoch": 1.2738095238095237,
      "grad_norm": 86.0,
      "learning_rate": 8.161248763639154e-06,
      "loss": 0.28514519333839417,
      "step": 214,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.2797619047619047,
      "grad_norm": 23.75,
      "learning_rate": 8.142135320174758e-06,
      "loss": 0.03289160877466202,
      "step": 215,
      "token_acc": 1.0
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 39.0,
      "learning_rate": 8.122945689443328e-06,
      "loss": 0.41763025522232056,
      "step": 216,
      "token_acc": 0.9047619047619048
    },
    {
      "epoch": 1.2916666666666667,
      "grad_norm": 31.25,
      "learning_rate": 8.1036803367353e-06,
      "loss": 0.05297483503818512,
      "step": 217,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.2976190476190477,
      "grad_norm": 1.6796875,
      "learning_rate": 8.084339729177142e-06,
      "loss": 0.0020398059859871864,
      "step": 218,
      "token_acc": 1.0
    },
    {
      "epoch": 1.3035714285714286,
      "grad_norm": 2.28125,
      "learning_rate": 8.064924335720023e-06,
      "loss": 0.005319602787494659,
      "step": 219,
      "token_acc": 1.0
    },
    {
      "epoch": 1.3095238095238095,
      "grad_norm": 15.5,
      "learning_rate": 8.045434627128446e-06,
      "loss": 0.04212145507335663,
      "step": 220,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.3154761904761905,
      "grad_norm": 45.75,
      "learning_rate": 8.025871075968828e-06,
      "loss": 0.18461646139621735,
      "step": 221,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 1.3214285714285714,
      "grad_norm": 0.5,
      "learning_rate": 8.006234156598043e-06,
      "loss": 0.0012304981937631965,
      "step": 222,
      "token_acc": 1.0
    },
    {
      "epoch": 1.3273809523809523,
      "grad_norm": 38.0,
      "learning_rate": 7.986524345151924e-06,
      "loss": 0.14975251257419586,
      "step": 223,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.00518798828125,
      "learning_rate": 7.966742119533724e-06,
      "loss": 1.5729372535133734e-05,
      "step": 224,
      "token_acc": 1.0
    },
    {
      "epoch": 1.3392857142857144,
      "grad_norm": 8.125,
      "learning_rate": 7.946887959402504e-06,
      "loss": 0.01573719084262848,
      "step": 225,
      "token_acc": 1.0
    },
    {
      "epoch": 1.3452380952380953,
      "grad_norm": 18.5,
      "learning_rate": 7.926962346161535e-06,
      "loss": 0.08771181851625443,
      "step": 226,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 1.3511904761904763,
      "grad_norm": 12.75,
      "learning_rate": 7.9069657629466e-06,
      "loss": 0.02646712400019169,
      "step": 227,
      "token_acc": 1.0
    },
    {
      "epoch": 1.3571428571428572,
      "grad_norm": 42.0,
      "learning_rate": 7.886898694614292e-06,
      "loss": 0.28221967816352844,
      "step": 228,
      "token_acc": 0.8421052631578947
    },
    {
      "epoch": 1.3630952380952381,
      "grad_norm": 6.8125,
      "learning_rate": 7.866761627730253e-06,
      "loss": 0.00805103313177824,
      "step": 229,
      "token_acc": 1.0
    },
    {
      "epoch": 1.369047619047619,
      "grad_norm": 8.3125,
      "learning_rate": 7.846555050557381e-06,
      "loss": 0.036620743572711945,
      "step": 230,
      "token_acc": 1.0
    },
    {
      "epoch": 1.375,
      "grad_norm": 0.06787109375,
      "learning_rate": 7.826279453043985e-06,
      "loss": 0.0001228916662512347,
      "step": 231,
      "token_acc": 1.0
    },
    {
      "epoch": 1.380952380952381,
      "grad_norm": 35.5,
      "learning_rate": 7.805935326811913e-06,
      "loss": 0.13240763545036316,
      "step": 232,
      "token_acc": 0.95
    },
    {
      "epoch": 1.3869047619047619,
      "grad_norm": 1.1328125,
      "learning_rate": 7.78552316514462e-06,
      "loss": 0.001800128142349422,
      "step": 233,
      "token_acc": 1.0
    },
    {
      "epoch": 1.3928571428571428,
      "grad_norm": 3.484375,
      "learning_rate": 7.765043462975217e-06,
      "loss": 0.0031609057914465666,
      "step": 234,
      "token_acc": 1.0
    },
    {
      "epoch": 1.3988095238095237,
      "grad_norm": 8.9375,
      "learning_rate": 7.744496716874472e-06,
      "loss": 0.0249114278703928,
      "step": 235,
      "token_acc": 1.0
    },
    {
      "epoch": 1.4047619047619047,
      "grad_norm": 60.25,
      "learning_rate": 7.723883425038759e-06,
      "loss": 0.388060063123703,
      "step": 236,
      "token_acc": 0.9090909090909091
    },
    {
      "epoch": 1.4107142857142856,
      "grad_norm": 0.016357421875,
      "learning_rate": 7.703204087277989e-06,
      "loss": 3.872206070809625e-05,
      "step": 237,
      "token_acc": 1.0
    },
    {
      "epoch": 1.4166666666666667,
      "grad_norm": 71.5,
      "learning_rate": 7.682459205003484e-06,
      "loss": 0.04742514342069626,
      "step": 238,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 1.4226190476190477,
      "grad_norm": 25.875,
      "learning_rate": 7.661649281215823e-06,
      "loss": 0.0250425785779953,
      "step": 239,
      "token_acc": 1.0
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 0.0020294189453125,
      "learning_rate": 7.640774820492647e-06,
      "loss": 5.06947117173695e-06,
      "step": 240,
      "token_acc": 1.0
    },
    {
      "epoch": 1.4345238095238095,
      "grad_norm": 15.625,
      "learning_rate": 7.619836328976416e-06,
      "loss": 0.00836940761655569,
      "step": 241,
      "token_acc": 1.0
    },
    {
      "epoch": 1.4404761904761905,
      "grad_norm": 10.25,
      "learning_rate": 7.598834314362151e-06,
      "loss": 0.011928348802030087,
      "step": 242,
      "token_acc": 1.0
    },
    {
      "epoch": 1.4464285714285714,
      "grad_norm": 32.75,
      "learning_rate": 7.57776928588511e-06,
      "loss": 0.1259881556034088,
      "step": 243,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.4523809523809523,
      "grad_norm": 0.07470703125,
      "learning_rate": 7.556641754308447e-06,
      "loss": 9.815546218305826e-05,
      "step": 244,
      "token_acc": 1.0
    },
    {
      "epoch": 1.4583333333333333,
      "grad_norm": 37.75,
      "learning_rate": 7.535452231910829e-06,
      "loss": 0.17217057943344116,
      "step": 245,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 1.4642857142857144,
      "grad_norm": 7.6875,
      "learning_rate": 7.514201232474012e-06,
      "loss": 0.006803174503147602,
      "step": 246,
      "token_acc": 1.0
    },
    {
      "epoch": 1.4702380952380953,
      "grad_norm": 26.375,
      "learning_rate": 7.492889271270382e-06,
      "loss": 0.08546401560306549,
      "step": 247,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 1.4761904761904763,
      "grad_norm": 53.25,
      "learning_rate": 7.471516865050468e-06,
      "loss": 0.01696542091667652,
      "step": 248,
      "token_acc": 1.0
    },
    {
      "epoch": 1.4821428571428572,
      "grad_norm": 0.2470703125,
      "learning_rate": 7.450084532030402e-06,
      "loss": 0.00047926127444952726,
      "step": 249,
      "token_acc": 1.0
    },
    {
      "epoch": 1.4880952380952381,
      "grad_norm": 115.0,
      "learning_rate": 7.428592791879361e-06,
      "loss": 0.15446770191192627,
      "step": 250,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.494047619047619,
      "grad_norm": 1.3359375,
      "learning_rate": 7.407042165706969e-06,
      "loss": 0.002424891572445631,
      "step": 251,
      "token_acc": 1.0
    },
    {
      "epoch": 1.5,
      "grad_norm": 35.0,
      "learning_rate": 7.385433176050654e-06,
      "loss": 0.2921290695667267,
      "step": 252,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.505952380952381,
      "grad_norm": 86.5,
      "learning_rate": 7.36376634686298e-06,
      "loss": 1.3855719566345215,
      "step": 253,
      "token_acc": 0.8888888888888888
    },
    {
      "epoch": 1.5119047619047619,
      "grad_norm": 0.07763671875,
      "learning_rate": 7.342042203498952e-06,
      "loss": 8.286959928227589e-05,
      "step": 254,
      "token_acc": 1.0
    },
    {
      "epoch": 1.5178571428571428,
      "grad_norm": 0.005401611328125,
      "learning_rate": 7.320261272703259e-06,
      "loss": 1.136190530814929e-05,
      "step": 255,
      "token_acc": 1.0
    },
    {
      "epoch": 1.5238095238095237,
      "grad_norm": 3.59375,
      "learning_rate": 7.298424082597526e-06,
      "loss": 0.011372007429599762,
      "step": 256,
      "token_acc": 1.0
    },
    {
      "epoch": 1.5297619047619047,
      "grad_norm": 73.5,
      "learning_rate": 7.276531162667484e-06,
      "loss": 0.6309632658958435,
      "step": 257,
      "token_acc": 0.9565217391304348
    },
    {
      "epoch": 1.5357142857142856,
      "grad_norm": 52.75,
      "learning_rate": 7.254583043750152e-06,
      "loss": 0.1435670554637909,
      "step": 258,
      "token_acc": 0.95
    },
    {
      "epoch": 1.5416666666666665,
      "grad_norm": 2.296875,
      "learning_rate": 7.232580258020952e-06,
      "loss": 0.0041695269756019115,
      "step": 259,
      "token_acc": 1.0
    },
    {
      "epoch": 1.5476190476190477,
      "grad_norm": 0.02490234375,
      "learning_rate": 7.210523338980814e-06,
      "loss": 5.4452109907288104e-05,
      "step": 260,
      "token_acc": 1.0
    },
    {
      "epoch": 1.5535714285714286,
      "grad_norm": 25.0,
      "learning_rate": 7.1884128214432366e-06,
      "loss": 0.09156952798366547,
      "step": 261,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.5595238095238095,
      "grad_norm": 7.375,
      "learning_rate": 7.1662492415213194e-06,
      "loss": 0.014834444969892502,
      "step": 262,
      "token_acc": 1.0
    },
    {
      "epoch": 1.5654761904761905,
      "grad_norm": 0.00262451171875,
      "learning_rate": 7.14403313661476e-06,
      "loss": 6.991356258367887e-06,
      "step": 263,
      "token_acc": 1.0
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 22.75,
      "learning_rate": 7.1217650453968335e-06,
      "loss": 0.1602877676486969,
      "step": 264,
      "token_acc": 0.9090909090909091
    },
    {
      "epoch": 1.5773809523809523,
      "grad_norm": 33.0,
      "learning_rate": 7.099445507801324e-06,
      "loss": 0.17254498600959778,
      "step": 265,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.5833333333333335,
      "grad_norm": 0.2021484375,
      "learning_rate": 7.0770750650094335e-06,
      "loss": 0.0003263050748500973,
      "step": 266,
      "token_acc": 1.0
    },
    {
      "epoch": 1.5892857142857144,
      "grad_norm": 3.1875,
      "learning_rate": 7.0546542594366605e-06,
      "loss": 0.006397673394531012,
      "step": 267,
      "token_acc": 1.0
    },
    {
      "epoch": 1.5952380952380953,
      "grad_norm": 0.005279541015625,
      "learning_rate": 7.03218363471965e-06,
      "loss": 9.389983461005613e-06,
      "step": 268,
      "token_acc": 1.0
    },
    {
      "epoch": 1.6011904761904763,
      "grad_norm": 0.010986328125,
      "learning_rate": 7.0096637357030105e-06,
      "loss": 3.080944952671416e-05,
      "step": 269,
      "token_acc": 1.0
    },
    {
      "epoch": 1.6071428571428572,
      "grad_norm": 32.25,
      "learning_rate": 6.987095108426102e-06,
      "loss": 0.15948952734470367,
      "step": 270,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.6130952380952381,
      "grad_norm": 35.5,
      "learning_rate": 6.964478300109796e-06,
      "loss": 0.08392617106437683,
      "step": 271,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 1.619047619047619,
      "grad_norm": 30.375,
      "learning_rate": 6.94181385914321e-06,
      "loss": 0.27560704946517944,
      "step": 272,
      "token_acc": 0.9
    },
    {
      "epoch": 1.625,
      "grad_norm": 57.75,
      "learning_rate": 6.91910233507041e-06,
      "loss": 0.2859566807746887,
      "step": 273,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.630952380952381,
      "grad_norm": 1.328125,
      "learning_rate": 6.896344278577083e-06,
      "loss": 0.001716909813694656,
      "step": 274,
      "token_acc": 1.0
    },
    {
      "epoch": 1.6369047619047619,
      "grad_norm": 6.9375,
      "learning_rate": 6.873540241477189e-06,
      "loss": 0.0037305899895727634,
      "step": 275,
      "token_acc": 1.0
    },
    {
      "epoch": 1.6428571428571428,
      "grad_norm": 0.044189453125,
      "learning_rate": 6.850690776699574e-06,
      "loss": 8.438099757768214e-05,
      "step": 276,
      "token_acc": 1.0
    },
    {
      "epoch": 1.6488095238095237,
      "grad_norm": 0.0673828125,
      "learning_rate": 6.8277964382745675e-06,
      "loss": 8.950386836659163e-05,
      "step": 277,
      "token_acc": 1.0
    },
    {
      "epoch": 1.6547619047619047,
      "grad_norm": 20.5,
      "learning_rate": 6.804857781320558e-06,
      "loss": 0.07852990925312042,
      "step": 278,
      "token_acc": 0.95
    },
    {
      "epoch": 1.6607142857142856,
      "grad_norm": 39.75,
      "learning_rate": 6.781875362030512e-06,
      "loss": 0.382799357175827,
      "step": 279,
      "token_acc": 0.95
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 18.125,
      "learning_rate": 6.758849737658508e-06,
      "loss": 0.019406642764806747,
      "step": 280,
      "token_acc": 1.0
    },
    {
      "epoch": 1.6726190476190477,
      "grad_norm": 35.0,
      "learning_rate": 6.735781466506216e-06,
      "loss": 0.26383548974990845,
      "step": 281,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 1.6785714285714286,
      "grad_norm": 2.375,
      "learning_rate": 6.712671107909359e-06,
      "loss": 0.003952500876039267,
      "step": 282,
      "token_acc": 1.0
    },
    {
      "epoch": 1.6845238095238095,
      "grad_norm": 0.0045166015625,
      "learning_rate": 6.6895192222241534e-06,
      "loss": 1.507397428213153e-05,
      "step": 283,
      "token_acc": 1.0
    },
    {
      "epoch": 1.6904761904761905,
      "grad_norm": 56.5,
      "learning_rate": 6.666326370813722e-06,
      "loss": 0.061668217182159424,
      "step": 284,
      "token_acc": 0.95
    },
    {
      "epoch": 1.6964285714285714,
      "grad_norm": 2.421875,
      "learning_rate": 6.643093116034486e-06,
      "loss": 0.008492996916174889,
      "step": 285,
      "token_acc": 1.0
    },
    {
      "epoch": 1.7023809523809523,
      "grad_norm": 29.5,
      "learning_rate": 6.619820021222518e-06,
      "loss": 0.08563974499702454,
      "step": 286,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.7083333333333335,
      "grad_norm": 2.8125,
      "learning_rate": 6.5965076506799e-06,
      "loss": 0.0031951735727488995,
      "step": 287,
      "token_acc": 1.0
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 7.375,
      "learning_rate": 6.573156569661026e-06,
      "loss": 0.015522127971053123,
      "step": 288,
      "token_acc": 1.0
    },
    {
      "epoch": 1.7202380952380953,
      "grad_norm": 43.0,
      "learning_rate": 6.549767344358903e-06,
      "loss": 0.18240037560462952,
      "step": 289,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.7261904761904763,
      "grad_norm": 7.53125,
      "learning_rate": 6.526340541891418e-06,
      "loss": 0.013299450278282166,
      "step": 290,
      "token_acc": 1.0
    },
    {
      "epoch": 1.7321428571428572,
      "grad_norm": 33.75,
      "learning_rate": 6.5028767302875974e-06,
      "loss": 0.09017157554626465,
      "step": 291,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.7380952380952381,
      "grad_norm": 45.25,
      "learning_rate": 6.479376478473822e-06,
      "loss": 0.2781052589416504,
      "step": 292,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.744047619047619,
      "grad_norm": 10.9375,
      "learning_rate": 6.455840356260041e-06,
      "loss": 0.010326328687369823,
      "step": 293,
      "token_acc": 1.0
    },
    {
      "epoch": 1.75,
      "grad_norm": 12.125,
      "learning_rate": 6.432268934325947e-06,
      "loss": 0.036904361099004745,
      "step": 294,
      "token_acc": 1.0
    },
    {
      "epoch": 1.755952380952381,
      "grad_norm": 66.5,
      "learning_rate": 6.408662784207149e-06,
      "loss": 0.7190050482749939,
      "step": 295,
      "token_acc": 0.9
    },
    {
      "epoch": 1.7619047619047619,
      "grad_norm": 1.7265625,
      "learning_rate": 6.385022478281307e-06,
      "loss": 0.0034032100811600685,
      "step": 296,
      "token_acc": 1.0
    },
    {
      "epoch": 1.7678571428571428,
      "grad_norm": 0.00390625,
      "learning_rate": 6.361348589754255e-06,
      "loss": 9.138930181507021e-06,
      "step": 297,
      "token_acc": 1.0
    },
    {
      "epoch": 1.7738095238095237,
      "grad_norm": 0.002349853515625,
      "learning_rate": 6.337641692646106e-06,
      "loss": 8.07943069958128e-06,
      "step": 298,
      "token_acc": 1.0
    },
    {
      "epoch": 1.7797619047619047,
      "grad_norm": 1.03125,
      "learning_rate": 6.313902361777327e-06,
      "loss": 0.002550265518948436,
      "step": 299,
      "token_acc": 1.0
    },
    {
      "epoch": 1.7857142857142856,
      "grad_norm": 17.375,
      "learning_rate": 6.290131172754811e-06,
      "loss": 0.012529956176877022,
      "step": 300,
      "token_acc": 1.0
    },
    {
      "epoch": 1.7916666666666665,
      "grad_norm": 47.75,
      "learning_rate": 6.266328701957911e-06,
      "loss": 0.24420566856861115,
      "step": 301,
      "token_acc": 0.9565217391304348
    },
    {
      "epoch": 1.7976190476190477,
      "grad_norm": 18.875,
      "learning_rate": 6.24249552652447e-06,
      "loss": 0.15898573398590088,
      "step": 302,
      "token_acc": 0.9565217391304348
    },
    {
      "epoch": 1.8035714285714286,
      "grad_norm": 21.25,
      "learning_rate": 6.2186322243368236e-06,
      "loss": 0.26730480790138245,
      "step": 303,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 1.8095238095238095,
      "grad_norm": 21.5,
      "learning_rate": 6.194739374007792e-06,
      "loss": 0.10243863612413406,
      "step": 304,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.8154761904761905,
      "grad_norm": 21.625,
      "learning_rate": 6.170817554866646e-06,
      "loss": 0.0468592569231987,
      "step": 305,
      "token_acc": 1.0
    },
    {
      "epoch": 1.8214285714285714,
      "grad_norm": 64.0,
      "learning_rate": 6.1468673469450655e-06,
      "loss": 0.15088951587677002,
      "step": 306,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 1.8273809523809523,
      "grad_norm": 12.625,
      "learning_rate": 6.122889330963069e-06,
      "loss": 0.008798900060355663,
      "step": 307,
      "token_acc": 1.0
    },
    {
      "epoch": 1.8333333333333335,
      "grad_norm": 0.2275390625,
      "learning_rate": 6.098884088314938e-06,
      "loss": 0.0005418129730969667,
      "step": 308,
      "token_acc": 1.0
    },
    {
      "epoch": 1.8392857142857144,
      "grad_norm": 0.07470703125,
      "learning_rate": 6.074852201055121e-06,
      "loss": 8.928579336497933e-05,
      "step": 309,
      "token_acc": 1.0
    },
    {
      "epoch": 1.8452380952380953,
      "grad_norm": 0.01263427734375,
      "learning_rate": 6.050794251884112e-06,
      "loss": 2.043678796326276e-05,
      "step": 310,
      "token_acc": 1.0
    },
    {
      "epoch": 1.8511904761904763,
      "grad_norm": 31.625,
      "learning_rate": 6.026710824134331e-06,
      "loss": 0.23552614450454712,
      "step": 311,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 0.01007080078125,
      "learning_rate": 6.002602501755974e-06,
      "loss": 2.663072882569395e-05,
      "step": 312,
      "token_acc": 1.0
    },
    {
      "epoch": 1.8630952380952381,
      "grad_norm": 52.0,
      "learning_rate": 5.978469869302861e-06,
      "loss": 0.1901112198829651,
      "step": 313,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.869047619047619,
      "grad_norm": 2.859375,
      "learning_rate": 5.954313511918252e-06,
      "loss": 0.005672132596373558,
      "step": 314,
      "token_acc": 1.0
    },
    {
      "epoch": 1.875,
      "grad_norm": 19.625,
      "learning_rate": 5.9301340153206685e-06,
      "loss": 0.029972020536661148,
      "step": 315,
      "token_acc": 1.0
    },
    {
      "epoch": 1.880952380952381,
      "grad_norm": 1.390625,
      "learning_rate": 5.905931965789688e-06,
      "loss": 0.002449198393151164,
      "step": 316,
      "token_acc": 1.0
    },
    {
      "epoch": 1.8869047619047619,
      "grad_norm": 22.75,
      "learning_rate": 5.881707950151725e-06,
      "loss": 0.11269201338291168,
      "step": 317,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.8928571428571428,
      "grad_norm": 0.4453125,
      "learning_rate": 5.857462555765809e-06,
      "loss": 0.0008049134630709887,
      "step": 318,
      "token_acc": 1.0
    },
    {
      "epoch": 1.8988095238095237,
      "grad_norm": 17.0,
      "learning_rate": 5.8331963705093375e-06,
      "loss": 0.039410550147295,
      "step": 319,
      "token_acc": 0.95
    },
    {
      "epoch": 1.9047619047619047,
      "grad_norm": 9.6875,
      "learning_rate": 5.808909982763825e-06,
      "loss": 0.008211517706513405,
      "step": 320,
      "token_acc": 1.0
    },
    {
      "epoch": 1.9107142857142856,
      "grad_norm": 43.75,
      "learning_rate": 5.784603981400632e-06,
      "loss": 0.2918968200683594,
      "step": 321,
      "token_acc": 0.9
    },
    {
      "epoch": 1.9166666666666665,
      "grad_norm": 34.0,
      "learning_rate": 5.760278955766695e-06,
      "loss": 0.2164626568555832,
      "step": 322,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 1.9226190476190477,
      "grad_norm": 61.5,
      "learning_rate": 5.735935495670229e-06,
      "loss": 0.25891825556755066,
      "step": 323,
      "token_acc": 0.9047619047619048
    },
    {
      "epoch": 1.9285714285714286,
      "grad_norm": 0.0135498046875,
      "learning_rate": 5.711574191366427e-06,
      "loss": 4.2159819713560864e-05,
      "step": 324,
      "token_acc": 1.0
    },
    {
      "epoch": 1.9345238095238095,
      "grad_norm": 49.0,
      "learning_rate": 5.687195633543151e-06,
      "loss": 0.32974255084991455,
      "step": 325,
      "token_acc": 0.8571428571428571
    },
    {
      "epoch": 1.9404761904761905,
      "grad_norm": 49.25,
      "learning_rate": 5.662800413306611e-06,
      "loss": 0.23124001920223236,
      "step": 326,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 1.9464285714285714,
      "grad_norm": 3.40625,
      "learning_rate": 5.6383891221670275e-06,
      "loss": 0.008325645700097084,
      "step": 327,
      "token_acc": 1.0
    },
    {
      "epoch": 1.9523809523809523,
      "grad_norm": 1.421875,
      "learning_rate": 5.613962352024293e-06,
      "loss": 0.002611794276162982,
      "step": 328,
      "token_acc": 1.0
    },
    {
      "epoch": 1.9583333333333335,
      "grad_norm": 7.21875,
      "learning_rate": 5.589520695153618e-06,
      "loss": 0.013795760460197926,
      "step": 329,
      "token_acc": 1.0
    },
    {
      "epoch": 1.9642857142857144,
      "grad_norm": 39.75,
      "learning_rate": 5.5650647441911706e-06,
      "loss": 0.6616204977035522,
      "step": 330,
      "token_acc": 0.95
    },
    {
      "epoch": 1.9702380952380953,
      "grad_norm": 10.25,
      "learning_rate": 5.540595092119709e-06,
      "loss": 0.04873330891132355,
      "step": 331,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 1.9761904761904763,
      "grad_norm": 15.125,
      "learning_rate": 5.516112332254203e-06,
      "loss": 0.0645870491862297,
      "step": 332,
      "token_acc": 0.95
    },
    {
      "epoch": 1.9821428571428572,
      "grad_norm": 1.921875,
      "learning_rate": 5.491617058227443e-06,
      "loss": 0.003113324521109462,
      "step": 333,
      "token_acc": 1.0
    },
    {
      "epoch": 1.9880952380952381,
      "grad_norm": 37.5,
      "learning_rate": 5.46710986397565e-06,
      "loss": 0.22948329150676727,
      "step": 334,
      "token_acc": 0.95
    },
    {
      "epoch": 1.994047619047619,
      "grad_norm": 0.076171875,
      "learning_rate": 5.442591343724081e-06,
      "loss": 0.00020153426157776266,
      "step": 335,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.625,
      "learning_rate": 5.418062091972604e-06,
      "loss": 0.017698675394058228,
      "step": 336,
      "token_acc": 1.0
    },
    {
      "epoch": 2.005952380952381,
      "grad_norm": 0.02001953125,
      "learning_rate": 5.393522703481303e-06,
      "loss": 4.6138073230395094e-05,
      "step": 337,
      "token_acc": 1.0
    },
    {
      "epoch": 2.011904761904762,
      "grad_norm": 3.765625,
      "learning_rate": 5.36897377325604e-06,
      "loss": 0.007646176964044571,
      "step": 338,
      "token_acc": 1.0
    },
    {
      "epoch": 2.017857142857143,
      "grad_norm": 0.022705078125,
      "learning_rate": 5.344415896534039e-06,
      "loss": 7.36122383386828e-05,
      "step": 339,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0238095238095237,
      "grad_norm": 0.007049560546875,
      "learning_rate": 5.319849668769449e-06,
      "loss": 1.7296106307185255e-05,
      "step": 340,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0297619047619047,
      "grad_norm": 0.5859375,
      "learning_rate": 5.295275685618905e-06,
      "loss": 0.0011170448269695044,
      "step": 341,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0357142857142856,
      "grad_norm": 1.4453125,
      "learning_rate": 5.270694542927089e-06,
      "loss": 0.00369239947758615,
      "step": 342,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0416666666666665,
      "grad_norm": 0.26171875,
      "learning_rate": 5.246106836712277e-06,
      "loss": 0.0005579355056397617,
      "step": 343,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0476190476190474,
      "grad_norm": 25.0,
      "learning_rate": 5.2215131631518945e-06,
      "loss": 0.14330650866031647,
      "step": 344,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 2.0535714285714284,
      "grad_norm": 0.0517578125,
      "learning_rate": 5.196914118568054e-06,
      "loss": 9.765980212250724e-05,
      "step": 345,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0595238095238093,
      "grad_norm": 0.1240234375,
      "learning_rate": 5.1723102994130994e-06,
      "loss": 0.0003235893091186881,
      "step": 346,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0654761904761907,
      "grad_norm": 77.5,
      "learning_rate": 5.147702302255143e-06,
      "loss": 0.16627377271652222,
      "step": 347,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 2.0714285714285716,
      "grad_norm": 1.234375,
      "learning_rate": 5.123090723763607e-06,
      "loss": 0.0031113955192267895,
      "step": 348,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0773809523809526,
      "grad_norm": 18.0,
      "learning_rate": 5.098476160694741e-06,
      "loss": 0.0468563586473465,
      "step": 349,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0833333333333335,
      "grad_norm": 1.671875,
      "learning_rate": 5.073859209877167e-06,
      "loss": 0.005610193125903606,
      "step": 350,
      "token_acc": 1.0
    },
    {
      "epoch": 2.0892857142857144,
      "grad_norm": 14.4375,
      "learning_rate": 5.049240468197401e-06,
      "loss": 0.08437106758356094,
      "step": 351,
      "token_acc": 0.9565217391304348
    },
    {
      "epoch": 2.0952380952380953,
      "grad_norm": 0.66015625,
      "learning_rate": 5.0246205325853824e-06,
      "loss": 0.0009546325309202075,
      "step": 352,
      "token_acc": 1.0
    },
    {
      "epoch": 2.1011904761904763,
      "grad_norm": 8.1875,
      "learning_rate": 5e-06,
      "loss": 0.047204405069351196,
      "step": 353,
      "token_acc": 1.0
    },
    {
      "epoch": 2.107142857142857,
      "grad_norm": 0.08349609375,
      "learning_rate": 4.975379467414621e-06,
      "loss": 0.00013370050874073058,
      "step": 354,
      "token_acc": 1.0
    },
    {
      "epoch": 2.113095238095238,
      "grad_norm": 0.040771484375,
      "learning_rate": 4.950759531802602e-06,
      "loss": 9.178226173389703e-05,
      "step": 355,
      "token_acc": 1.0
    },
    {
      "epoch": 2.119047619047619,
      "grad_norm": 0.01519775390625,
      "learning_rate": 4.926140790122835e-06,
      "loss": 3.601023490773514e-05,
      "step": 356,
      "token_acc": 1.0
    },
    {
      "epoch": 2.125,
      "grad_norm": 0.039794921875,
      "learning_rate": 4.90152383930526e-06,
      "loss": 8.667759539093822e-05,
      "step": 357,
      "token_acc": 1.0
    },
    {
      "epoch": 2.130952380952381,
      "grad_norm": 0.240234375,
      "learning_rate": 4.876909276236395e-06,
      "loss": 0.0006661512306891382,
      "step": 358,
      "token_acc": 1.0
    },
    {
      "epoch": 2.136904761904762,
      "grad_norm": 50.0,
      "learning_rate": 4.852297697744857e-06,
      "loss": 0.18797577917575836,
      "step": 359,
      "token_acc": 0.95
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 7.4375,
      "learning_rate": 4.827689700586902e-06,
      "loss": 0.014067665673792362,
      "step": 360,
      "token_acc": 1.0
    },
    {
      "epoch": 2.1488095238095237,
      "grad_norm": 11.0,
      "learning_rate": 4.803085881431949e-06,
      "loss": 0.015956411138176918,
      "step": 361,
      "token_acc": 1.0
    },
    {
      "epoch": 2.1547619047619047,
      "grad_norm": 3.921875,
      "learning_rate": 4.778486836848107e-06,
      "loss": 0.006085686385631561,
      "step": 362,
      "token_acc": 1.0
    },
    {
      "epoch": 2.1607142857142856,
      "grad_norm": 0.458984375,
      "learning_rate": 4.7538931632877254e-06,
      "loss": 0.0007959132199175656,
      "step": 363,
      "token_acc": 1.0
    },
    {
      "epoch": 2.1666666666666665,
      "grad_norm": 0.019775390625,
      "learning_rate": 4.729305457072913e-06,
      "loss": 4.894421363133006e-05,
      "step": 364,
      "token_acc": 1.0
    },
    {
      "epoch": 2.1726190476190474,
      "grad_norm": 0.0242919921875,
      "learning_rate": 4.704724314381097e-06,
      "loss": 7.041867502266541e-05,
      "step": 365,
      "token_acc": 1.0
    },
    {
      "epoch": 2.1785714285714284,
      "grad_norm": 0.45703125,
      "learning_rate": 4.680150331230552e-06,
      "loss": 0.0009508822113275528,
      "step": 366,
      "token_acc": 1.0
    },
    {
      "epoch": 2.1845238095238093,
      "grad_norm": 13.5,
      "learning_rate": 4.6555841034659625e-06,
      "loss": 0.03099062480032444,
      "step": 367,
      "token_acc": 1.0
    },
    {
      "epoch": 2.1904761904761907,
      "grad_norm": 0.28515625,
      "learning_rate": 4.631026226743962e-06,
      "loss": 0.0007869544206187129,
      "step": 368,
      "token_acc": 1.0
    },
    {
      "epoch": 2.1964285714285716,
      "grad_norm": 0.049560546875,
      "learning_rate": 4.606477296518698e-06,
      "loss": 7.773852121317759e-05,
      "step": 369,
      "token_acc": 1.0
    },
    {
      "epoch": 2.2023809523809526,
      "grad_norm": 0.029052734375,
      "learning_rate": 4.581937908027397e-06,
      "loss": 7.748129428364336e-05,
      "step": 370,
      "token_acc": 1.0
    },
    {
      "epoch": 2.2083333333333335,
      "grad_norm": 0.00543212890625,
      "learning_rate": 4.55740865627592e-06,
      "loss": 1.4940102118998766e-05,
      "step": 371,
      "token_acc": 1.0
    },
    {
      "epoch": 2.2142857142857144,
      "grad_norm": 10.125,
      "learning_rate": 4.532890136024351e-06,
      "loss": 0.031819961965084076,
      "step": 372,
      "token_acc": 1.0
    },
    {
      "epoch": 2.2202380952380953,
      "grad_norm": 66.5,
      "learning_rate": 4.508382941772558e-06,
      "loss": 0.2309388816356659,
      "step": 373,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 2.2261904761904763,
      "grad_norm": 7.34375,
      "learning_rate": 4.483887667745798e-06,
      "loss": 0.016479093581438065,
      "step": 374,
      "token_acc": 1.0
    },
    {
      "epoch": 2.232142857142857,
      "grad_norm": 0.267578125,
      "learning_rate": 4.459404907880293e-06,
      "loss": 0.000589508272241801,
      "step": 375,
      "token_acc": 1.0
    },
    {
      "epoch": 2.238095238095238,
      "grad_norm": 20.25,
      "learning_rate": 4.434935255808831e-06,
      "loss": 0.07630139589309692,
      "step": 376,
      "token_acc": 0.95
    },
    {
      "epoch": 2.244047619047619,
      "grad_norm": 0.921875,
      "learning_rate": 4.410479304846385e-06,
      "loss": 0.0009514412959106266,
      "step": 377,
      "token_acc": 1.0
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.045654296875,
      "learning_rate": 4.386037647975708e-06,
      "loss": 0.00010391621617600322,
      "step": 378,
      "token_acc": 1.0
    },
    {
      "epoch": 2.255952380952381,
      "grad_norm": 0.01300048828125,
      "learning_rate": 4.361610877832974e-06,
      "loss": 2.7739344659494236e-05,
      "step": 379,
      "token_acc": 1.0
    },
    {
      "epoch": 2.261904761904762,
      "grad_norm": 2.71875,
      "learning_rate": 4.337199586693389e-06,
      "loss": 0.0053138271905481815,
      "step": 380,
      "token_acc": 1.0
    },
    {
      "epoch": 2.267857142857143,
      "grad_norm": 0.021728515625,
      "learning_rate": 4.312804366456851e-06,
      "loss": 4.372529292595573e-05,
      "step": 381,
      "token_acc": 1.0
    },
    {
      "epoch": 2.2738095238095237,
      "grad_norm": 0.0003719329833984375,
      "learning_rate": 4.2884258086335755e-06,
      "loss": 1.3291809182192083e-06,
      "step": 382,
      "token_acc": 1.0
    },
    {
      "epoch": 2.2797619047619047,
      "grad_norm": 0.0185546875,
      "learning_rate": 4.2640645043297715e-06,
      "loss": 3.360396658536047e-05,
      "step": 383,
      "token_acc": 1.0
    },
    {
      "epoch": 2.2857142857142856,
      "grad_norm": 0.007568359375,
      "learning_rate": 4.239721044233306e-06,
      "loss": 1.3990214029035997e-05,
      "step": 384,
      "token_acc": 1.0
    },
    {
      "epoch": 2.2916666666666665,
      "grad_norm": 0.05615234375,
      "learning_rate": 4.215396018599369e-06,
      "loss": 0.00010430814290884882,
      "step": 385,
      "token_acc": 1.0
    },
    {
      "epoch": 2.2976190476190474,
      "grad_norm": 0.00063323974609375,
      "learning_rate": 4.191090017236177e-06,
      "loss": 1.8344934460401419e-06,
      "step": 386,
      "token_acc": 1.0
    },
    {
      "epoch": 2.3035714285714284,
      "grad_norm": 0.000743865966796875,
      "learning_rate": 4.166803629490664e-06,
      "loss": 2.165623300243169e-06,
      "step": 387,
      "token_acc": 1.0
    },
    {
      "epoch": 2.3095238095238093,
      "grad_norm": 0.0203857421875,
      "learning_rate": 4.142537444234192e-06,
      "loss": 2.287254210386891e-05,
      "step": 388,
      "token_acc": 1.0
    },
    {
      "epoch": 2.3154761904761907,
      "grad_norm": 0.50390625,
      "learning_rate": 4.118292049848277e-06,
      "loss": 0.0009783882414922118,
      "step": 389,
      "token_acc": 1.0
    },
    {
      "epoch": 2.3214285714285716,
      "grad_norm": 0.0322265625,
      "learning_rate": 4.094068034210313e-06,
      "loss": 3.7315956433303654e-05,
      "step": 390,
      "token_acc": 1.0
    },
    {
      "epoch": 2.3273809523809526,
      "grad_norm": 0.046630859375,
      "learning_rate": 4.069865984679332e-06,
      "loss": 6.559052417287603e-05,
      "step": 391,
      "token_acc": 1.0
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 112.0,
      "learning_rate": 4.045686488081748e-06,
      "loss": 0.5002931952476501,
      "step": 392,
      "token_acc": 0.9090909090909091
    },
    {
      "epoch": 2.3392857142857144,
      "grad_norm": 61.75,
      "learning_rate": 4.021530130697141e-06,
      "loss": 0.23847536742687225,
      "step": 393,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 2.3452380952380953,
      "grad_norm": 50.25,
      "learning_rate": 3.997397498244028e-06,
      "loss": 0.05840262398123741,
      "step": 394,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 2.3511904761904763,
      "grad_norm": 0.01019287109375,
      "learning_rate": 3.97328917586567e-06,
      "loss": 1.426756352884695e-05,
      "step": 395,
      "token_acc": 1.0
    },
    {
      "epoch": 2.357142857142857,
      "grad_norm": 2.96875,
      "learning_rate": 3.9492057481158905e-06,
      "loss": 0.003948180004954338,
      "step": 396,
      "token_acc": 1.0
    },
    {
      "epoch": 2.363095238095238,
      "grad_norm": 0.0025634765625,
      "learning_rate": 3.92514779894488e-06,
      "loss": 6.280268280534074e-06,
      "step": 397,
      "token_acc": 1.0
    },
    {
      "epoch": 2.369047619047619,
      "grad_norm": 0.14453125,
      "learning_rate": 3.901115911685063e-06,
      "loss": 0.00035102630499750376,
      "step": 398,
      "token_acc": 1.0
    },
    {
      "epoch": 2.375,
      "grad_norm": 3.5,
      "learning_rate": 3.877110669036932e-06,
      "loss": 0.007151306606829166,
      "step": 399,
      "token_acc": 1.0
    },
    {
      "epoch": 2.380952380952381,
      "grad_norm": 45.75,
      "learning_rate": 3.853132653054936e-06,
      "loss": 0.7128760814666748,
      "step": 400,
      "token_acc": 0.95
    },
    {
      "epoch": 2.386904761904762,
      "grad_norm": 0.10205078125,
      "learning_rate": 3.829182445133356e-06,
      "loss": 0.0002187932113884017,
      "step": 401,
      "token_acc": 1.0
    },
    {
      "epoch": 2.392857142857143,
      "grad_norm": 0.0050048828125,
      "learning_rate": 3.8052606259922097e-06,
      "loss": 1.0006265256379265e-05,
      "step": 402,
      "token_acc": 1.0
    },
    {
      "epoch": 2.3988095238095237,
      "grad_norm": 0.01708984375,
      "learning_rate": 3.7813677756631773e-06,
      "loss": 4.1592324123485014e-05,
      "step": 403,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4047619047619047,
      "grad_norm": 0.357421875,
      "learning_rate": 3.75750447347553e-06,
      "loss": 0.0003019916475750506,
      "step": 404,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4107142857142856,
      "grad_norm": 0.0020751953125,
      "learning_rate": 3.7336712980420897e-06,
      "loss": 4.234958851156989e-06,
      "step": 405,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4166666666666665,
      "grad_norm": 0.1162109375,
      "learning_rate": 3.7098688272451893e-06,
      "loss": 0.00011525607260409743,
      "step": 406,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4226190476190474,
      "grad_norm": 0.02392578125,
      "learning_rate": 3.6860976382226747e-06,
      "loss": 4.062395964865573e-05,
      "step": 407,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4285714285714284,
      "grad_norm": 0.0986328125,
      "learning_rate": 3.662358307353897e-06,
      "loss": 0.0001014654990285635,
      "step": 408,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4345238095238093,
      "grad_norm": 0.40234375,
      "learning_rate": 3.638651410245746e-06,
      "loss": 0.0007898493204265833,
      "step": 409,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4404761904761907,
      "grad_norm": 0.2099609375,
      "learning_rate": 3.6149775217186954e-06,
      "loss": 0.00034261544351466,
      "step": 410,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4464285714285716,
      "grad_norm": 1.0703125,
      "learning_rate": 3.5913372157928515e-06,
      "loss": 0.0014202806632965803,
      "step": 411,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4523809523809526,
      "grad_norm": 2.109375,
      "learning_rate": 3.5677310656740537e-06,
      "loss": 0.0017308567184954882,
      "step": 412,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4583333333333335,
      "grad_norm": 32.0,
      "learning_rate": 3.5441596437399596e-06,
      "loss": 0.20744018256664276,
      "step": 413,
      "token_acc": 0.95
    },
    {
      "epoch": 2.4642857142857144,
      "grad_norm": 0.04443359375,
      "learning_rate": 3.5206235215261785e-06,
      "loss": 7.562364044133574e-05,
      "step": 414,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4702380952380953,
      "grad_norm": 0.0008697509765625,
      "learning_rate": 3.4971232697124046e-06,
      "loss": 2.6169143438892206e-06,
      "step": 415,
      "token_acc": 1.0
    },
    {
      "epoch": 2.4761904761904763,
      "grad_norm": 33.75,
      "learning_rate": 3.4736594581085837e-06,
      "loss": 0.2524469196796417,
      "step": 416,
      "token_acc": 0.9090909090909091
    },
    {
      "epoch": 2.482142857142857,
      "grad_norm": 27.5,
      "learning_rate": 3.4502326556411e-06,
      "loss": 0.040829941630363464,
      "step": 417,
      "token_acc": 1.0
    },
    {
      "epoch": 2.488095238095238,
      "grad_norm": 43.0,
      "learning_rate": 3.4268434303389747e-06,
      "loss": 0.2938973605632782,
      "step": 418,
      "token_acc": 0.95
    },
    {
      "epoch": 2.494047619047619,
      "grad_norm": 1.2421875,
      "learning_rate": 3.403492349320101e-06,
      "loss": 0.0010657752864062786,
      "step": 419,
      "token_acc": 1.0
    },
    {
      "epoch": 2.5,
      "grad_norm": 12.8125,
      "learning_rate": 3.380179978777482e-06,
      "loss": 0.011580419726669788,
      "step": 420,
      "token_acc": 1.0
    },
    {
      "epoch": 2.505952380952381,
      "grad_norm": 1.3984375,
      "learning_rate": 3.356906883965516e-06,
      "loss": 0.0012280117953196168,
      "step": 421,
      "token_acc": 1.0
    },
    {
      "epoch": 2.511904761904762,
      "grad_norm": 56.0,
      "learning_rate": 3.33367362918628e-06,
      "loss": 0.2506561875343323,
      "step": 422,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 2.517857142857143,
      "grad_norm": 3.5625,
      "learning_rate": 3.3104807777758487e-06,
      "loss": 0.0018459546845406294,
      "step": 423,
      "token_acc": 1.0
    },
    {
      "epoch": 2.5238095238095237,
      "grad_norm": 0.138671875,
      "learning_rate": 3.2873288920906436e-06,
      "loss": 0.00016272446373477578,
      "step": 424,
      "token_acc": 1.0
    },
    {
      "epoch": 2.5297619047619047,
      "grad_norm": 33.5,
      "learning_rate": 3.2642185334937853e-06,
      "loss": 0.2919960021972656,
      "step": 425,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 2.5357142857142856,
      "grad_norm": 0.8984375,
      "learning_rate": 3.2411502623414925e-06,
      "loss": 0.0006140850018709898,
      "step": 426,
      "token_acc": 1.0
    },
    {
      "epoch": 2.5416666666666665,
      "grad_norm": 0.05078125,
      "learning_rate": 3.2181246379694886e-06,
      "loss": 8.062724373303354e-05,
      "step": 427,
      "token_acc": 1.0
    },
    {
      "epoch": 2.5476190476190474,
      "grad_norm": 0.275390625,
      "learning_rate": 3.1951422186794447e-06,
      "loss": 0.00043984033982269466,
      "step": 428,
      "token_acc": 1.0
    },
    {
      "epoch": 2.553571428571429,
      "grad_norm": 0.0272216796875,
      "learning_rate": 3.1722035617254333e-06,
      "loss": 3.694374026963487e-05,
      "step": 429,
      "token_acc": 1.0
    },
    {
      "epoch": 2.5595238095238093,
      "grad_norm": 9.375,
      "learning_rate": 3.149309223300428e-06,
      "loss": 0.01070562843233347,
      "step": 430,
      "token_acc": 1.0
    },
    {
      "epoch": 2.5654761904761907,
      "grad_norm": 13.25,
      "learning_rate": 3.126459758522813e-06,
      "loss": 0.026648035272955894,
      "step": 431,
      "token_acc": 1.0
    },
    {
      "epoch": 2.571428571428571,
      "grad_norm": 0.01361083984375,
      "learning_rate": 3.103655721422917e-06,
      "loss": 1.8180486222263426e-05,
      "step": 432,
      "token_acc": 1.0
    },
    {
      "epoch": 2.5773809523809526,
      "grad_norm": 0.034912109375,
      "learning_rate": 3.080897664929592e-06,
      "loss": 4.7559056838508695e-05,
      "step": 433,
      "token_acc": 1.0
    },
    {
      "epoch": 2.5833333333333335,
      "grad_norm": 33.25,
      "learning_rate": 3.0581861408567907e-06,
      "loss": 0.22851479053497314,
      "step": 434,
      "token_acc": 0.9
    },
    {
      "epoch": 2.5892857142857144,
      "grad_norm": 14.9375,
      "learning_rate": 3.035521699890206e-06,
      "loss": 0.020124109461903572,
      "step": 435,
      "token_acc": 1.0
    },
    {
      "epoch": 2.5952380952380953,
      "grad_norm": 21.5,
      "learning_rate": 3.0129048915739013e-06,
      "loss": 0.020518623292446136,
      "step": 436,
      "token_acc": 1.0
    },
    {
      "epoch": 2.6011904761904763,
      "grad_norm": 0.004486083984375,
      "learning_rate": 2.9903362642969903e-06,
      "loss": 8.588300261180848e-06,
      "step": 437,
      "token_acc": 1.0
    },
    {
      "epoch": 2.607142857142857,
      "grad_norm": 78.5,
      "learning_rate": 2.967816365280351e-06,
      "loss": 0.19017250835895538,
      "step": 438,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 2.613095238095238,
      "grad_norm": 29.5,
      "learning_rate": 2.94534574056334e-06,
      "loss": 0.06718239188194275,
      "step": 439,
      "token_acc": 0.95
    },
    {
      "epoch": 2.619047619047619,
      "grad_norm": 0.0125732421875,
      "learning_rate": 2.9229249349905686e-06,
      "loss": 2.5916346203302965e-05,
      "step": 440,
      "token_acc": 1.0
    },
    {
      "epoch": 2.625,
      "grad_norm": 66.0,
      "learning_rate": 2.9005544921986774e-06,
      "loss": 0.07408752292394638,
      "step": 441,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 2.630952380952381,
      "grad_norm": 51.75,
      "learning_rate": 2.8782349546031673e-06,
      "loss": 0.25111842155456543,
      "step": 442,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 2.636904761904762,
      "grad_norm": 0.0026702880859375,
      "learning_rate": 2.8559668633852433e-06,
      "loss": 7.390777682303451e-06,
      "step": 443,
      "token_acc": 1.0
    },
    {
      "epoch": 2.642857142857143,
      "grad_norm": 0.0023956298828125,
      "learning_rate": 2.8337507584786826e-06,
      "loss": 6.236305125639774e-06,
      "step": 444,
      "token_acc": 1.0
    },
    {
      "epoch": 2.6488095238095237,
      "grad_norm": 0.09814453125,
      "learning_rate": 2.811587178556764e-06,
      "loss": 0.00012843837612308562,
      "step": 445,
      "token_acc": 1.0
    },
    {
      "epoch": 2.6547619047619047,
      "grad_norm": 0.0181884765625,
      "learning_rate": 2.789476661019186e-06,
      "loss": 3.704077607835643e-05,
      "step": 446,
      "token_acc": 1.0
    },
    {
      "epoch": 2.6607142857142856,
      "grad_norm": 25.75,
      "learning_rate": 2.7674197419790493e-06,
      "loss": 0.06991499662399292,
      "step": 447,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.41796875,
      "learning_rate": 2.7454169562498503e-06,
      "loss": 0.0005995816900394857,
      "step": 448,
      "token_acc": 1.0
    },
    {
      "epoch": 2.6726190476190474,
      "grad_norm": 0.00074005126953125,
      "learning_rate": 2.723468837332517e-06,
      "loss": 1.4869754068058683e-06,
      "step": 449,
      "token_acc": 1.0
    },
    {
      "epoch": 2.678571428571429,
      "grad_norm": 0.003662109375,
      "learning_rate": 2.7015759174024756e-06,
      "loss": 8.923187124310061e-06,
      "step": 450,
      "token_acc": 1.0
    },
    {
      "epoch": 2.6845238095238093,
      "grad_norm": 28.75,
      "learning_rate": 2.6797387272967414e-06,
      "loss": 0.059650927782058716,
      "step": 451,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 2.6904761904761907,
      "grad_norm": 0.00494384765625,
      "learning_rate": 2.65795779650105e-06,
      "loss": 9.241196494258475e-06,
      "step": 452,
      "token_acc": 1.0
    },
    {
      "epoch": 2.696428571428571,
      "grad_norm": 33.25,
      "learning_rate": 2.63623365313702e-06,
      "loss": 0.23944468796253204,
      "step": 453,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 2.7023809523809526,
      "grad_norm": 0.01043701171875,
      "learning_rate": 2.614566823949348e-06,
      "loss": 1.1867048669955693e-05,
      "step": 454,
      "token_acc": 1.0
    },
    {
      "epoch": 2.7083333333333335,
      "grad_norm": 38.25,
      "learning_rate": 2.592957834293033e-06,
      "loss": 0.07086221128702164,
      "step": 455,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 2.7142857142857144,
      "grad_norm": 0.0024566650390625,
      "learning_rate": 2.5714072081206407e-06,
      "loss": 6.029250926076202e-06,
      "step": 456,
      "token_acc": 1.0
    },
    {
      "epoch": 2.7202380952380953,
      "grad_norm": 0.039794921875,
      "learning_rate": 2.5499154679696014e-06,
      "loss": 3.6115852708462626e-05,
      "step": 457,
      "token_acc": 1.0
    },
    {
      "epoch": 2.7261904761904763,
      "grad_norm": 3.65625,
      "learning_rate": 2.528483134949535e-06,
      "loss": 0.005144556052982807,
      "step": 458,
      "token_acc": 1.0
    },
    {
      "epoch": 2.732142857142857,
      "grad_norm": 17.875,
      "learning_rate": 2.50711072872962e-06,
      "loss": 0.04441129416227341,
      "step": 459,
      "token_acc": 0.95
    },
    {
      "epoch": 2.738095238095238,
      "grad_norm": 1.703125,
      "learning_rate": 2.4857987675259887e-06,
      "loss": 0.0015799179673194885,
      "step": 460,
      "token_acc": 1.0
    },
    {
      "epoch": 2.744047619047619,
      "grad_norm": 0.12890625,
      "learning_rate": 2.4645477680891734e-06,
      "loss": 0.00016846107610035688,
      "step": 461,
      "token_acc": 1.0
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.00109100341796875,
      "learning_rate": 2.4433582456915556e-06,
      "loss": 3.1258944090950536e-06,
      "step": 462,
      "token_acc": 1.0
    },
    {
      "epoch": 2.755952380952381,
      "grad_norm": 78.5,
      "learning_rate": 2.422230714114891e-06,
      "loss": 0.20750439167022705,
      "step": 463,
      "token_acc": 0.95
    },
    {
      "epoch": 2.761904761904762,
      "grad_norm": 0.0279541015625,
      "learning_rate": 2.4011656856378513e-06,
      "loss": 5.329581108526327e-05,
      "step": 464,
      "token_acc": 1.0
    },
    {
      "epoch": 2.767857142857143,
      "grad_norm": 20.625,
      "learning_rate": 2.3801636710235836e-06,
      "loss": 0.014012366533279419,
      "step": 465,
      "token_acc": 1.0
    },
    {
      "epoch": 2.7738095238095237,
      "grad_norm": 7.15625,
      "learning_rate": 2.3592251795073564e-06,
      "loss": 0.016505159437656403,
      "step": 466,
      "token_acc": 1.0
    },
    {
      "epoch": 2.7797619047619047,
      "grad_norm": 0.004364013671875,
      "learning_rate": 2.338350718784177e-06,
      "loss": 8.457236617687158e-06,
      "step": 467,
      "token_acc": 1.0
    },
    {
      "epoch": 2.7857142857142856,
      "grad_norm": 13.3125,
      "learning_rate": 2.3175407949965167e-06,
      "loss": 0.03953416272997856,
      "step": 468,
      "token_acc": 1.0
    },
    {
      "epoch": 2.7916666666666665,
      "grad_norm": 0.0233154296875,
      "learning_rate": 2.296795912722014e-06,
      "loss": 4.995938797947019e-05,
      "step": 469,
      "token_acc": 1.0
    },
    {
      "epoch": 2.7976190476190474,
      "grad_norm": 3.515625,
      "learning_rate": 2.2761165749612417e-06,
      "loss": 0.0021178722381591797,
      "step": 470,
      "token_acc": 1.0
    },
    {
      "epoch": 2.803571428571429,
      "grad_norm": 0.259765625,
      "learning_rate": 2.25550328312553e-06,
      "loss": 0.00025379922590218484,
      "step": 471,
      "token_acc": 1.0
    },
    {
      "epoch": 2.8095238095238093,
      "grad_norm": 14.4375,
      "learning_rate": 2.2349565370247837e-06,
      "loss": 0.01060462649911642,
      "step": 472,
      "token_acc": 1.0
    },
    {
      "epoch": 2.8154761904761907,
      "grad_norm": 0.146484375,
      "learning_rate": 2.214476834855382e-06,
      "loss": 0.0002505412558093667,
      "step": 473,
      "token_acc": 1.0
    },
    {
      "epoch": 2.821428571428571,
      "grad_norm": 0.50390625,
      "learning_rate": 2.1940646731880887e-06,
      "loss": 0.0007228928152471781,
      "step": 474,
      "token_acc": 1.0
    },
    {
      "epoch": 2.8273809523809526,
      "grad_norm": 14.6875,
      "learning_rate": 2.173720546956015e-06,
      "loss": 0.014972873963415623,
      "step": 475,
      "token_acc": 1.0
    },
    {
      "epoch": 2.8333333333333335,
      "grad_norm": 35.5,
      "learning_rate": 2.1534449494426203e-06,
      "loss": 0.02709815837442875,
      "step": 476,
      "token_acc": 1.0
    },
    {
      "epoch": 2.8392857142857144,
      "grad_norm": 0.98828125,
      "learning_rate": 2.1332383722697483e-06,
      "loss": 0.0016251765191555023,
      "step": 477,
      "token_acc": 1.0
    },
    {
      "epoch": 2.8452380952380953,
      "grad_norm": 41.25,
      "learning_rate": 2.1131013053857097e-06,
      "loss": 0.22743842005729675,
      "step": 478,
      "token_acc": 0.95
    },
    {
      "epoch": 2.8511904761904763,
      "grad_norm": 9.125,
      "learning_rate": 2.0930342370534013e-06,
      "loss": 0.026223070919513702,
      "step": 479,
      "token_acc": 1.0
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 0.46875,
      "learning_rate": 2.073037653838466e-06,
      "loss": 0.0007355016423389316,
      "step": 480,
      "token_acc": 1.0
    },
    {
      "epoch": 2.863095238095238,
      "grad_norm": 55.5,
      "learning_rate": 2.053112040597495e-06,
      "loss": 0.36355313658714294,
      "step": 481,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 2.869047619047619,
      "grad_norm": 0.06298828125,
      "learning_rate": 2.0332578804662783e-06,
      "loss": 0.00011641244782367721,
      "step": 482,
      "token_acc": 1.0
    },
    {
      "epoch": 2.875,
      "grad_norm": 22.875,
      "learning_rate": 2.013475654848076e-06,
      "loss": 0.04043431207537651,
      "step": 483,
      "token_acc": 1.0
    },
    {
      "epoch": 2.880952380952381,
      "grad_norm": 2.265625,
      "learning_rate": 1.99376584340196e-06,
      "loss": 0.004154494497925043,
      "step": 484,
      "token_acc": 1.0
    },
    {
      "epoch": 2.886904761904762,
      "grad_norm": 0.051025390625,
      "learning_rate": 1.9741289240311757e-06,
      "loss": 7.224738510558382e-05,
      "step": 485,
      "token_acc": 1.0
    },
    {
      "epoch": 2.892857142857143,
      "grad_norm": 0.00091552734375,
      "learning_rate": 1.954565372871554e-06,
      "loss": 2.402041445748182e-06,
      "step": 486,
      "token_acc": 1.0
    },
    {
      "epoch": 2.8988095238095237,
      "grad_norm": 81.5,
      "learning_rate": 1.935075664279978e-06,
      "loss": 0.4486009180545807,
      "step": 487,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 2.9047619047619047,
      "grad_norm": 0.41796875,
      "learning_rate": 1.9156602708228584e-06,
      "loss": 0.0003363674331922084,
      "step": 488,
      "token_acc": 1.0
    },
    {
      "epoch": 2.9107142857142856,
      "grad_norm": 17.0,
      "learning_rate": 1.8963196632647008e-06,
      "loss": 0.023882651701569557,
      "step": 489,
      "token_acc": 1.0
    },
    {
      "epoch": 2.9166666666666665,
      "grad_norm": 0.00323486328125,
      "learning_rate": 1.8770543105566752e-06,
      "loss": 8.132353286782745e-06,
      "step": 490,
      "token_acc": 1.0
    },
    {
      "epoch": 2.9226190476190474,
      "grad_norm": 2.0,
      "learning_rate": 1.8578646798252432e-06,
      "loss": 0.0016411024844273925,
      "step": 491,
      "token_acc": 1.0
    },
    {
      "epoch": 2.928571428571429,
      "grad_norm": 25.5,
      "learning_rate": 1.8387512363608496e-06,
      "loss": 0.07898645102977753,
      "step": 492,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 2.9345238095238093,
      "grad_norm": 0.0703125,
      "learning_rate": 1.8197144436066167e-06,
      "loss": 0.00010011708945967257,
      "step": 493,
      "token_acc": 1.0
    },
    {
      "epoch": 2.9404761904761907,
      "grad_norm": 5.21875,
      "learning_rate": 1.8007547631471289e-06,
      "loss": 0.006450686603784561,
      "step": 494,
      "token_acc": 1.0
    },
    {
      "epoch": 2.946428571428571,
      "grad_norm": 0.17578125,
      "learning_rate": 1.781872654697226e-06,
      "loss": 0.0001997375366045162,
      "step": 495,
      "token_acc": 1.0
    },
    {
      "epoch": 2.9523809523809526,
      "grad_norm": 41.25,
      "learning_rate": 1.7630685760908623e-06,
      "loss": 0.7631595730781555,
      "step": 496,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 2.9583333333333335,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.7443429832700038e-06,
      "loss": 1.54236640810268e-05,
      "step": 497,
      "token_acc": 1.0
    },
    {
      "epoch": 2.9642857142857144,
      "grad_norm": 17.125,
      "learning_rate": 1.7256963302735752e-06,
      "loss": 0.014609376899898052,
      "step": 498,
      "token_acc": 1.0
    },
    {
      "epoch": 2.9702380952380953,
      "grad_norm": 16.75,
      "learning_rate": 1.7071290692264492e-06,
      "loss": 0.02625737339258194,
      "step": 499,
      "token_acc": 1.0
    },
    {
      "epoch": 2.9761904761904763,
      "grad_norm": 0.06396484375,
      "learning_rate": 1.6886416503284835e-06,
      "loss": 0.00010523501259740442,
      "step": 500,
      "token_acc": 1.0
    },
    {
      "epoch": 2.982142857142857,
      "grad_norm": 54.25,
      "learning_rate": 1.6702345218436066e-06,
      "loss": 0.21214470267295837,
      "step": 501,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 2.988095238095238,
      "grad_norm": 0.10693359375,
      "learning_rate": 1.6519081300889472e-06,
      "loss": 0.0002360567741561681,
      "step": 502,
      "token_acc": 1.0
    },
    {
      "epoch": 2.994047619047619,
      "grad_norm": 12.6875,
      "learning_rate": 1.6336629194240118e-06,
      "loss": 0.006979122757911682,
      "step": 503,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3984375,
      "learning_rate": 1.6154993322399114e-06,
      "loss": 0.0014935710933059454,
      "step": 504,
      "token_acc": 1.0
    },
    {
      "epoch": 3.005952380952381,
      "grad_norm": 44.0,
      "learning_rate": 1.5974178089486364e-06,
      "loss": 0.061158593744039536,
      "step": 505,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 3.011904761904762,
      "grad_norm": 3.453125,
      "learning_rate": 1.5794187879723755e-06,
      "loss": 0.0066911522299051285,
      "step": 506,
      "token_acc": 1.0
    },
    {
      "epoch": 3.017857142857143,
      "grad_norm": 0.040771484375,
      "learning_rate": 1.561502705732883e-06,
      "loss": 7.653488137293607e-05,
      "step": 507,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0238095238095237,
      "grad_norm": 0.0478515625,
      "learning_rate": 1.543669996640908e-06,
      "loss": 8.507210441166535e-05,
      "step": 508,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0297619047619047,
      "grad_norm": 0.625,
      "learning_rate": 1.5259210930856423e-06,
      "loss": 0.0007482210057787597,
      "step": 509,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0357142857142856,
      "grad_norm": 0.56640625,
      "learning_rate": 1.5082564254242583e-06,
      "loss": 0.0010129186557605863,
      "step": 510,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0416666666666665,
      "grad_norm": 85.5,
      "learning_rate": 1.4906764219714537e-06,
      "loss": 0.41510215401649475,
      "step": 511,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 3.0476190476190474,
      "grad_norm": 0.04931640625,
      "learning_rate": 1.4731815089890795e-06,
      "loss": 6.346028385451064e-05,
      "step": 512,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0535714285714284,
      "grad_norm": 0.00106048583984375,
      "learning_rate": 1.455772110675804e-06,
      "loss": 2.3716040686849738e-06,
      "step": 513,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0595238095238093,
      "grad_norm": 28.75,
      "learning_rate": 1.438448649156815e-06,
      "loss": 0.0810873806476593,
      "step": 514,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 3.0654761904761907,
      "grad_norm": 0.01458740234375,
      "learning_rate": 1.4212115444736024e-06,
      "loss": 1.3644992577610537e-05,
      "step": 515,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0714285714285716,
      "grad_norm": 0.2109375,
      "learning_rate": 1.4040612145737608e-06,
      "loss": 0.0002722729986999184,
      "step": 516,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0773809523809526,
      "grad_norm": 8.0625,
      "learning_rate": 1.3869980753008537e-06,
      "loss": 0.010679463855922222,
      "step": 517,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0833333333333335,
      "grad_norm": 0.1416015625,
      "learning_rate": 1.370022540384347e-06,
      "loss": 0.00021240914065856487,
      "step": 518,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0892857142857144,
      "grad_norm": 0.41015625,
      "learning_rate": 1.353135021429554e-06,
      "loss": 0.00039127242052927613,
      "step": 519,
      "token_acc": 1.0
    },
    {
      "epoch": 3.0952380952380953,
      "grad_norm": 0.000965118408203125,
      "learning_rate": 1.3363359279076776e-06,
      "loss": 2.159003088308964e-06,
      "step": 520,
      "token_acc": 1.0
    },
    {
      "epoch": 3.1011904761904763,
      "grad_norm": 11.0,
      "learning_rate": 1.3196256671458663e-06,
      "loss": 0.02277286723256111,
      "step": 521,
      "token_acc": 1.0
    },
    {
      "epoch": 3.107142857142857,
      "grad_norm": 0.00150299072265625,
      "learning_rate": 1.3030046443173445e-06,
      "loss": 4.0861759771360084e-06,
      "step": 522,
      "token_acc": 1.0
    },
    {
      "epoch": 3.113095238095238,
      "grad_norm": 0.0028533935546875,
      "learning_rate": 1.2864732624315867e-06,
      "loss": 7.80802838562522e-06,
      "step": 523,
      "token_acc": 1.0
    },
    {
      "epoch": 3.119047619047619,
      "grad_norm": 13.3125,
      "learning_rate": 1.270031922324546e-06,
      "loss": 0.011324808932840824,
      "step": 524,
      "token_acc": 1.0
    },
    {
      "epoch": 3.125,
      "grad_norm": 0.01043701171875,
      "learning_rate": 1.2536810226489354e-06,
      "loss": 1.4981829735916108e-05,
      "step": 525,
      "token_acc": 1.0
    },
    {
      "epoch": 3.130952380952381,
      "grad_norm": 0.002532958984375,
      "learning_rate": 1.237420959864561e-06,
      "loss": 4.350994913693285e-06,
      "step": 526,
      "token_acc": 1.0
    },
    {
      "epoch": 3.136904761904762,
      "grad_norm": 64.5,
      "learning_rate": 1.2212521282287093e-06,
      "loss": 0.15771399438381195,
      "step": 527,
      "token_acc": 0.9565217391304348
    },
    {
      "epoch": 3.142857142857143,
      "grad_norm": 2.171875,
      "learning_rate": 1.2051749197865875e-06,
      "loss": 0.0021707299165427685,
      "step": 528,
      "token_acc": 1.0
    },
    {
      "epoch": 3.1488095238095237,
      "grad_norm": 0.0712890625,
      "learning_rate": 1.1891897243618184e-06,
      "loss": 0.0001262338919332251,
      "step": 529,
      "token_acc": 1.0
    },
    {
      "epoch": 3.1547619047619047,
      "grad_norm": 29.5,
      "learning_rate": 1.173296929546987e-06,
      "loss": 0.13383927941322327,
      "step": 530,
      "token_acc": 0.9523809523809523
    },
    {
      "epoch": 3.1607142857142856,
      "grad_norm": 0.00372314453125,
      "learning_rate": 1.1574969206942443e-06,
      "loss": 6.22129300609231e-06,
      "step": 531,
      "token_acc": 1.0
    },
    {
      "epoch": 3.1666666666666665,
      "grad_norm": 0.494140625,
      "learning_rate": 1.1417900809059623e-06,
      "loss": 0.00043546216329559684,
      "step": 532,
      "token_acc": 1.0
    },
    {
      "epoch": 3.1726190476190474,
      "grad_norm": 0.40625,
      "learning_rate": 1.1261767910254422e-06,
      "loss": 0.000545055023394525,
      "step": 533,
      "token_acc": 1.0
    },
    {
      "epoch": 3.1785714285714284,
      "grad_norm": 0.09130859375,
      "learning_rate": 1.1106574296276923e-06,
      "loss": 0.00018246758554596454,
      "step": 534,
      "token_acc": 1.0
    },
    {
      "epoch": 3.1845238095238093,
      "grad_norm": 0.0038299560546875,
      "learning_rate": 1.095232373010226e-06,
      "loss": 6.838080025772797e-06,
      "step": 535,
      "token_acc": 1.0
    },
    {
      "epoch": 3.1904761904761907,
      "grad_norm": 0.2001953125,
      "learning_rate": 1.0799019951839656e-06,
      "loss": 0.0002770803985185921,
      "step": 536,
      "token_acc": 1.0
    },
    {
      "epoch": 3.1964285714285716,
      "grad_norm": 6.15625,
      "learning_rate": 1.0646666678641477e-06,
      "loss": 0.011164679192006588,
      "step": 537,
      "token_acc": 1.0
    },
    {
      "epoch": 3.2023809523809526,
      "grad_norm": 28.25,
      "learning_rate": 1.0495267604613273e-06,
      "loss": 0.2657778859138489,
      "step": 538,
      "token_acc": 0.95
    },
    {
      "epoch": 3.2083333333333335,
      "grad_norm": 7.0625,
      "learning_rate": 1.0344826400724185e-06,
      "loss": 0.004388697445392609,
      "step": 539,
      "token_acc": 1.0
    },
    {
      "epoch": 3.2142857142857144,
      "grad_norm": 0.0031585693359375,
      "learning_rate": 1.0195346714717813e-06,
      "loss": 5.495462573890109e-06,
      "step": 540,
      "token_acc": 1.0
    },
    {
      "epoch": 3.2202380952380953,
      "grad_norm": 0.01458740234375,
      "learning_rate": 1.0046832171023952e-06,
      "loss": 1.3641641089634504e-05,
      "step": 541,
      "token_acc": 1.0
    },
    {
      "epoch": 3.2261904761904763,
      "grad_norm": 0.003814697265625,
      "learning_rate": 9.899286370670575e-07,
      "loss": 7.39668166716001e-06,
      "step": 542,
      "token_acc": 1.0
    },
    {
      "epoch": 3.232142857142857,
      "grad_norm": 0.00189208984375,
      "learning_rate": 9.752712891196558e-07,
      "loss": 4.3569834815571085e-06,
      "step": 543,
      "token_acc": 1.0
    },
    {
      "epoch": 3.238095238095238,
      "grad_norm": 0.330078125,
      "learning_rate": 9.607115286564972e-07,
      "loss": 0.00024462962755933404,
      "step": 544,
      "token_acc": 1.0
    },
    {
      "epoch": 3.244047619047619,
      "grad_norm": 0.00194549560546875,
      "learning_rate": 9.46249708707681e-07,
      "loss": 4.463622644834686e-06,
      "step": 545,
      "token_acc": 1.0
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.00115966796875,
      "learning_rate": 9.318861799285539e-07,
      "loss": 3.350360202603042e-06,
      "step": 546,
      "token_acc": 1.0
    },
    {
      "epoch": 3.255952380952381,
      "grad_norm": 3.359375,
      "learning_rate": 9.176212905911946e-07,
      "loss": 0.00415303697809577,
      "step": 547,
      "token_acc": 1.0
    },
    {
      "epoch": 3.261904761904762,
      "grad_norm": 24.0,
      "learning_rate": 9.034553865759754e-07,
      "loss": 0.1297740340232849,
      "step": 548,
      "token_acc": 0.9565217391304348
    },
    {
      "epoch": 3.267857142857143,
      "grad_norm": 16.125,
      "learning_rate": 8.893888113631732e-07,
      "loss": 0.027608206495642662,
      "step": 549,
      "token_acc": 1.0
    },
    {
      "epoch": 3.2738095238095237,
      "grad_norm": 2.671875,
      "learning_rate": 8.754219060246432e-07,
      "loss": 0.0026375434827059507,
      "step": 550,
      "token_acc": 1.0
    },
    {
      "epoch": 3.2797619047619047,
      "grad_norm": 0.30859375,
      "learning_rate": 8.615550092155478e-07,
      "loss": 0.00042662909254431725,
      "step": 551,
      "token_acc": 1.0
    },
    {
      "epoch": 3.2857142857142856,
      "grad_norm": 0.0634765625,
      "learning_rate": 8.477884571661449e-07,
      "loss": 9.890252113109455e-05,
      "step": 552,
      "token_acc": 1.0
    },
    {
      "epoch": 3.2916666666666665,
      "grad_norm": 10.8125,
      "learning_rate": 8.341225836736367e-07,
      "loss": 0.008015697821974754,
      "step": 553,
      "token_acc": 1.0
    },
    {
      "epoch": 3.2976190476190474,
      "grad_norm": 0.1552734375,
      "learning_rate": 8.20557720094074e-07,
      "loss": 0.0002810191654134542,
      "step": 554,
      "token_acc": 1.0
    },
    {
      "epoch": 3.3035714285714284,
      "grad_norm": 0.087890625,
      "learning_rate": 8.070941953343242e-07,
      "loss": 0.00015164667274802923,
      "step": 555,
      "token_acc": 1.0
    },
    {
      "epoch": 3.3095238095238093,
      "grad_norm": 0.034912109375,
      "learning_rate": 7.937323358440935e-07,
      "loss": 6.951785326236859e-05,
      "step": 556,
      "token_acc": 1.0
    },
    {
      "epoch": 3.3154761904761907,
      "grad_norm": 37.5,
      "learning_rate": 7.804724656080182e-07,
      "loss": 0.13474984467029572,
      "step": 557,
      "token_acc": 0.8947368421052632
    },
    {
      "epoch": 3.3214285714285716,
      "grad_norm": 0.73046875,
      "learning_rate": 7.673149061377966e-07,
      "loss": 0.0004250766651239246,
      "step": 558,
      "token_acc": 1.0
    },
    {
      "epoch": 3.3273809523809526,
      "grad_norm": 17.5,
      "learning_rate": 7.542599764644049e-07,
      "loss": 0.06693873554468155,
      "step": 559,
      "token_acc": 0.9583333333333334
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 0.5234375,
      "learning_rate": 7.413079931303591e-07,
      "loss": 0.0004485200042836368,
      "step": 560,
      "token_acc": 1.0
    },
    {
      "epoch": 3.3392857142857144,
      "grad_norm": 0.00250244140625,
      "learning_rate": 7.284592701820325e-07,
      "loss": 5.935148237767862e-06,
      "step": 561,
      "token_acc": 1.0
    },
    {
      "epoch": 3.3452380952380953,
      "grad_norm": 0.02392578125,
      "learning_rate": 7.157141191620548e-07,
      "loss": 4.5421067625284195e-05,
      "step": 562,
      "token_acc": 1.0
    },
    {
      "epoch": 3.3511904761904763,
      "grad_norm": 0.000675201416015625,
      "learning_rate": 7.030728491017408e-07,
      "loss": 2.5598519641789608e-06,
      "step": 563,
      "token_acc": 1.0
    },
    {
      "epoch": 3.357142857142857,
      "grad_norm": 58.0,
      "learning_rate": 6.905357665136142e-07,
      "loss": 0.18578100204467773,
      "step": 564,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 3.363095238095238,
      "grad_norm": 0.0869140625,
      "learning_rate": 6.781031753839662e-07,
      "loss": 0.0001655336091062054,
      "step": 565,
      "token_acc": 1.0
    },
    {
      "epoch": 3.369047619047619,
      "grad_norm": 61.5,
      "learning_rate": 6.657753771654812e-07,
      "loss": 0.33632856607437134,
      "step": 566,
      "token_acc": 0.95
    },
    {
      "epoch": 3.375,
      "grad_norm": 1.9375,
      "learning_rate": 6.535526707699408e-07,
      "loss": 0.004421261139214039,
      "step": 567,
      "token_acc": 1.0
    },
    {
      "epoch": 3.380952380952381,
      "grad_norm": 0.671875,
      "learning_rate": 6.414353525609628e-07,
      "loss": 0.0006151712732389569,
      "step": 568,
      "token_acc": 1.0
    },
    {
      "epoch": 3.386904761904762,
      "grad_norm": 18.25,
      "learning_rate": 6.294237163468231e-07,
      "loss": 0.03200333192944527,
      "step": 569,
      "token_acc": 1.0
    },
    {
      "epoch": 3.392857142857143,
      "grad_norm": 2.9375,
      "learning_rate": 6.175180533733277e-07,
      "loss": 0.001050041988492012,
      "step": 570,
      "token_acc": 1.0
    },
    {
      "epoch": 3.3988095238095237,
      "grad_norm": 0.035400390625,
      "learning_rate": 6.057186523167529e-07,
      "loss": 6.42839222564362e-05,
      "step": 571,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4047619047619047,
      "grad_norm": 0.0011138916015625,
      "learning_rate": 5.940257992768456e-07,
      "loss": 3.536516487656627e-06,
      "step": 572,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4107142857142856,
      "grad_norm": 0.0155029296875,
      "learning_rate": 5.824397777698859e-07,
      "loss": 2.117331314366311e-05,
      "step": 573,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4166666666666665,
      "grad_norm": 7.46875,
      "learning_rate": 5.709608687218116e-07,
      "loss": 0.02661375142633915,
      "step": 574,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4226190476190474,
      "grad_norm": 0.00182342529296875,
      "learning_rate": 5.595893504614097e-07,
      "loss": 4.962756065651774e-06,
      "step": 575,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4285714285714284,
      "grad_norm": 0.291015625,
      "learning_rate": 5.483254987135644e-07,
      "loss": 0.0002800719521474093,
      "step": 576,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4345238095238093,
      "grad_norm": 0.302734375,
      "learning_rate": 5.371695865925736e-07,
      "loss": 0.0003275743220001459,
      "step": 577,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4404761904761907,
      "grad_norm": 0.205078125,
      "learning_rate": 5.261218845955246e-07,
      "loss": 0.0003141165361739695,
      "step": 578,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4464285714285716,
      "grad_norm": 0.0003223419189453125,
      "learning_rate": 5.151826605957394e-07,
      "loss": 1.0854301990548265e-06,
      "step": 579,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4523809523809526,
      "grad_norm": 0.150390625,
      "learning_rate": 5.043521798362755e-07,
      "loss": 0.0001499994395999238,
      "step": 580,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4583333333333335,
      "grad_norm": 0.01904296875,
      "learning_rate": 4.936307049234956e-07,
      "loss": 4.018716572318226e-05,
      "step": 581,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4642857142857144,
      "grad_norm": 0.1318359375,
      "learning_rate": 4.830184958207007e-07,
      "loss": 0.00010588918667053804,
      "step": 582,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4702380952380953,
      "grad_norm": 0.0036163330078125,
      "learning_rate": 4.725158098418309e-07,
      "loss": 4.79530854136101e-06,
      "step": 583,
      "token_acc": 1.0
    },
    {
      "epoch": 3.4761904761904763,
      "grad_norm": 0.004974365234375,
      "learning_rate": 4.6212290164521554e-07,
      "loss": 1.0226433914795052e-05,
      "step": 584,
      "token_acc": 1.0
    },
    {
      "epoch": 3.482142857142857,
      "grad_norm": 35.25,
      "learning_rate": 4.5184002322740784e-07,
      "loss": 0.1580720692873001,
      "step": 585,
      "token_acc": 0.9
    },
    {
      "epoch": 3.488095238095238,
      "grad_norm": 0.0390625,
      "learning_rate": 4.4166742391707593e-07,
      "loss": 6.687987479381263e-05,
      "step": 586,
      "token_acc": 1.0
    },
    {
      "epoch": 3.494047619047619,
      "grad_norm": 0.00848388671875,
      "learning_rate": 4.316053503689466e-07,
      "loss": 1.2687462913163472e-05,
      "step": 587,
      "token_acc": 1.0
    },
    {
      "epoch": 3.5,
      "grad_norm": 23.875,
      "learning_rate": 4.2165404655783836e-07,
      "loss": 0.028276776894927025,
      "step": 588,
      "token_acc": 1.0
    },
    {
      "epoch": 3.505952380952381,
      "grad_norm": 0.02099609375,
      "learning_rate": 4.1181375377273237e-07,
      "loss": 2.3627069822396152e-05,
      "step": 589,
      "token_acc": 1.0
    },
    {
      "epoch": 3.511904761904762,
      "grad_norm": 0.07373046875,
      "learning_rate": 4.020847106109349e-07,
      "loss": 0.00011388575512683019,
      "step": 590,
      "token_acc": 1.0
    },
    {
      "epoch": 3.517857142857143,
      "grad_norm": 1.71875,
      "learning_rate": 3.9246715297228176e-07,
      "loss": 0.0009027900523506105,
      "step": 591,
      "token_acc": 1.0
    },
    {
      "epoch": 3.5238095238095237,
      "grad_norm": 0.000301361083984375,
      "learning_rate": 3.829613140534222e-07,
      "loss": 1.0530136478337226e-06,
      "step": 592,
      "token_acc": 1.0
    },
    {
      "epoch": 3.5297619047619047,
      "grad_norm": 0.328125,
      "learning_rate": 3.7356742434216775e-07,
      "loss": 0.00048546958714723587,
      "step": 593,
      "token_acc": 1.0
    },
    {
      "epoch": 3.5357142857142856,
      "grad_norm": 44.75,
      "learning_rate": 3.642857116118986e-07,
      "loss": 0.22204943001270294,
      "step": 594,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 3.5416666666666665,
      "grad_norm": 4.59375,
      "learning_rate": 3.5511640091604293e-07,
      "loss": 0.005602268036454916,
      "step": 595,
      "token_acc": 1.0
    },
    {
      "epoch": 3.5476190476190474,
      "grad_norm": 85.5,
      "learning_rate": 3.4605971458262e-07,
      "loss": 0.8018062710762024,
      "step": 596,
      "token_acc": 0.9
    },
    {
      "epoch": 3.553571428571429,
      "grad_norm": 0.006011962890625,
      "learning_rate": 3.371158722088497e-07,
      "loss": 1.0138217476196587e-05,
      "step": 597,
      "token_acc": 1.0
    },
    {
      "epoch": 3.5595238095238093,
      "grad_norm": 7.9375,
      "learning_rate": 3.2828509065582713e-07,
      "loss": 0.020824378356337547,
      "step": 598,
      "token_acc": 1.0
    },
    {
      "epoch": 3.5654761904761907,
      "grad_norm": 0.61328125,
      "learning_rate": 3.195675840432655e-07,
      "loss": 0.0005649444065056741,
      "step": 599,
      "token_acc": 1.0
    },
    {
      "epoch": 3.571428571428571,
      "grad_norm": 0.00250244140625,
      "learning_rate": 3.109635637443026e-07,
      "loss": 5.5516561587865e-06,
      "step": 600,
      "token_acc": 1.0
    },
    {
      "epoch": 3.5773809523809526,
      "grad_norm": 0.033203125,
      "learning_rate": 3.02473238380378e-07,
      "loss": 6.558089808095247e-05,
      "step": 601,
      "token_acc": 1.0
    },
    {
      "epoch": 3.5833333333333335,
      "grad_norm": 0.0194091796875,
      "learning_rate": 2.9409681381617315e-07,
      "loss": 3.2875057513592765e-05,
      "step": 602,
      "token_acc": 1.0
    },
    {
      "epoch": 3.5892857142857144,
      "grad_norm": 32.25,
      "learning_rate": 2.858344931546181e-07,
      "loss": 0.10170060396194458,
      "step": 603,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 3.5952380952380953,
      "grad_norm": 7.65625,
      "learning_rate": 2.776864767319731e-07,
      "loss": 0.017617840319871902,
      "step": 604,
      "token_acc": 1.0
    },
    {
      "epoch": 3.6011904761904763,
      "grad_norm": 0.150390625,
      "learning_rate": 2.696529621129618e-07,
      "loss": 0.00021752170869149268,
      "step": 605,
      "token_acc": 1.0
    },
    {
      "epoch": 3.607142857142857,
      "grad_norm": 0.00250244140625,
      "learning_rate": 2.617341440859883e-07,
      "loss": 2.8477381874836283e-06,
      "step": 606,
      "token_acc": 1.0
    },
    {
      "epoch": 3.613095238095238,
      "grad_norm": 0.0162353515625,
      "learning_rate": 2.539302146584116e-07,
      "loss": 2.9020276997471228e-05,
      "step": 607,
      "token_acc": 1.0
    },
    {
      "epoch": 3.619047619047619,
      "grad_norm": 19.625,
      "learning_rate": 2.4624136305188895e-07,
      "loss": 0.03796076029539108,
      "step": 608,
      "token_acc": 0.9545454545454546
    },
    {
      "epoch": 3.625,
      "grad_norm": 0.07177734375,
      "learning_rate": 2.3866777569779234e-07,
      "loss": 0.00011617827112786472,
      "step": 609,
      "token_acc": 1.0
    },
    {
      "epoch": 3.630952380952381,
      "grad_norm": 0.24609375,
      "learning_rate": 2.3120963623267822e-07,
      "loss": 0.00046131532872095704,
      "step": 610,
      "token_acc": 1.0
    },
    {
      "epoch": 3.636904761904762,
      "grad_norm": 9.75,
      "learning_rate": 2.2386712549384848e-07,
      "loss": 0.008443932048976421,
      "step": 611,
      "token_acc": 1.0
    },
    {
      "epoch": 3.642857142857143,
      "grad_norm": 0.0015716552734375,
      "learning_rate": 2.1664042151495424e-07,
      "loss": 3.883628778567072e-06,
      "step": 612,
      "token_acc": 1.0
    },
    {
      "epoch": 3.6488095238095237,
      "grad_norm": 0.0289306640625,
      "learning_rate": 2.095296995216828e-07,
      "loss": 5.444216731120832e-05,
      "step": 613,
      "token_acc": 1.0
    },
    {
      "epoch": 3.6547619047619047,
      "grad_norm": 3.515625,
      "learning_rate": 2.0253513192751374e-07,
      "loss": 0.0025166566483676434,
      "step": 614,
      "token_acc": 1.0
    },
    {
      "epoch": 3.6607142857142856,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9565688832952846e-07,
      "loss": 0.001579255796968937,
      "step": 615,
      "token_acc": 1.0
    },
    {
      "epoch": 3.6666666666666665,
      "grad_norm": 65.0,
      "learning_rate": 1.8889513550430892e-07,
      "loss": 0.15635859966278076,
      "step": 616,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 3.6726190476190474,
      "grad_norm": 0.228515625,
      "learning_rate": 1.8225003740388546e-07,
      "loss": 0.00023400859208777547,
      "step": 617,
      "token_acc": 1.0
    },
    {
      "epoch": 3.678571428571429,
      "grad_norm": 0.18359375,
      "learning_rate": 1.7572175515176538e-07,
      "loss": 0.00017701646720524877,
      "step": 618,
      "token_acc": 1.0
    },
    {
      "epoch": 3.6845238095238093,
      "grad_norm": 0.00872802734375,
      "learning_rate": 1.693104470390261e-07,
      "loss": 2.0973218852304853e-05,
      "step": 619,
      "token_acc": 1.0
    },
    {
      "epoch": 3.6904761904761907,
      "grad_norm": 32.5,
      "learning_rate": 1.6301626852047504e-07,
      "loss": 0.048330798745155334,
      "step": 620,
      "token_acc": 1.0
    },
    {
      "epoch": 3.696428571428571,
      "grad_norm": 4.84375,
      "learning_rate": 1.5683937221088242e-07,
      "loss": 0.010857430286705494,
      "step": 621,
      "token_acc": 1.0
    },
    {
      "epoch": 3.7023809523809526,
      "grad_norm": 0.002288818359375,
      "learning_rate": 1.5077990788127993e-07,
      "loss": 3.814644514932297e-06,
      "step": 622,
      "token_acc": 1.0
    },
    {
      "epoch": 3.7083333333333335,
      "grad_norm": 0.1240234375,
      "learning_rate": 1.448380224553303e-07,
      "loss": 0.00022894788708072156,
      "step": 623,
      "token_acc": 1.0
    },
    {
      "epoch": 3.7142857142857144,
      "grad_norm": 0.08544921875,
      "learning_rate": 1.3901386000576112e-07,
      "loss": 6.557910091942176e-05,
      "step": 624,
      "token_acc": 1.0
    },
    {
      "epoch": 3.7202380952380953,
      "grad_norm": 0.0159912109375,
      "learning_rate": 1.3330756175087778e-07,
      "loss": 2.8393571483320557e-05,
      "step": 625,
      "token_acc": 1.0
    },
    {
      "epoch": 3.7261904761904763,
      "grad_norm": 0.03515625,
      "learning_rate": 1.2771926605113283e-07,
      "loss": 4.7466954129049554e-05,
      "step": 626,
      "token_acc": 1.0
    },
    {
      "epoch": 3.732142857142857,
      "grad_norm": 0.00738525390625,
      "learning_rate": 1.2224910840577642e-07,
      "loss": 1.7250074961339124e-05,
      "step": 627,
      "token_acc": 1.0
    },
    {
      "epoch": 3.738095238095238,
      "grad_norm": 3.21875,
      "learning_rate": 1.1689722144956672e-07,
      "loss": 0.0025603759568184614,
      "step": 628,
      "token_acc": 1.0
    },
    {
      "epoch": 3.744047619047619,
      "grad_norm": 0.5625,
      "learning_rate": 1.1166373494955696e-07,
      "loss": 0.0008462554542347789,
      "step": 629,
      "token_acc": 1.0
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.0064697265625,
      "learning_rate": 1.06548775801949e-07,
      "loss": 8.85063400346553e-06,
      "step": 630,
      "token_acc": 1.0
    },
    {
      "epoch": 3.755952380952381,
      "grad_norm": 5.90625,
      "learning_rate": 1.0155246802901198e-07,
      "loss": 0.0034541762433946133,
      "step": 631,
      "token_acc": 1.0
    },
    {
      "epoch": 3.761904761904762,
      "grad_norm": 0.0159912109375,
      "learning_rate": 9.667493277608187e-08,
      "loss": 2.0649893485824578e-05,
      "step": 632,
      "token_acc": 1.0
    },
    {
      "epoch": 3.767857142857143,
      "grad_norm": 3.078125,
      "learning_rate": 9.191628830861832e-08,
      "loss": 0.0014245338970795274,
      "step": 633,
      "token_acc": 1.0
    },
    {
      "epoch": 3.7738095238095237,
      "grad_norm": 0.11962890625,
      "learning_rate": 8.727665000934027e-08,
      "loss": 0.00012316815264057368,
      "step": 634,
      "token_acc": 1.0
    },
    {
      "epoch": 3.7797619047619047,
      "grad_norm": 0.51953125,
      "learning_rate": 8.275613037542873e-08,
      "loss": 0.0006893071113154292,
      "step": 635,
      "token_acc": 1.0
    },
    {
      "epoch": 3.7857142857142856,
      "grad_norm": 0.1025390625,
      "learning_rate": 7.835483901579454e-08,
      "loss": 0.0002213457046309486,
      "step": 636,
      "token_acc": 1.0
    },
    {
      "epoch": 3.7916666666666665,
      "grad_norm": 0.00213623046875,
      "learning_rate": 7.407288264842772e-08,
      "loss": 5.276495357975364e-06,
      "step": 637,
      "token_acc": 1.0
    },
    {
      "epoch": 3.7976190476190474,
      "grad_norm": 0.224609375,
      "learning_rate": 6.991036509780391e-08,
      "loss": 0.0003499812737572938,
      "step": 638,
      "token_acc": 1.0
    },
    {
      "epoch": 3.803571428571429,
      "grad_norm": 25.625,
      "learning_rate": 6.58673872923693e-08,
      "loss": 0.04538915678858757,
      "step": 639,
      "token_acc": 0.9473684210526315
    },
    {
      "epoch": 3.8095238095238093,
      "grad_norm": 0.05078125,
      "learning_rate": 6.194404726209358e-08,
      "loss": 0.00010473098518559709,
      "step": 640,
      "token_acc": 1.0
    },
    {
      "epoch": 3.8154761904761907,
      "grad_norm": 0.30078125,
      "learning_rate": 5.8140440136091326e-08,
      "loss": 0.0005553648807108402,
      "step": 641,
      "token_acc": 1.0
    },
    {
      "epoch": 3.821428571428571,
      "grad_norm": 0.2392578125,
      "learning_rate": 5.445665814031942e-08,
      "loss": 0.00028235369245521724,
      "step": 642,
      "token_acc": 1.0
    },
    {
      "epoch": 3.8273809523809526,
      "grad_norm": 0.01275634765625,
      "learning_rate": 5.089279059533658e-08,
      "loss": 1.9687107851495966e-05,
      "step": 643,
      "token_acc": 1.0
    },
    {
      "epoch": 3.8333333333333335,
      "grad_norm": 0.03125,
      "learning_rate": 4.744892391413791e-08,
      "loss": 7.130828453227878e-05,
      "step": 644,
      "token_acc": 1.0
    },
    {
      "epoch": 3.8392857142857144,
      "grad_norm": 0.00970458984375,
      "learning_rate": 4.412514160006376e-08,
      "loss": 1.6902287825359963e-05,
      "step": 645,
      "token_acc": 1.0
    },
    {
      "epoch": 3.8452380952380953,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.092152424477025e-08,
      "loss": 0.00017856384511105716,
      "step": 646,
      "token_acc": 1.0
    },
    {
      "epoch": 3.8511904761904763,
      "grad_norm": 0.55859375,
      "learning_rate": 3.7838149526277514e-08,
      "loss": 0.0006260017398744822,
      "step": 647,
      "token_acc": 1.0
    },
    {
      "epoch": 3.857142857142857,
      "grad_norm": 0.0024261474609375,
      "learning_rate": 3.487509220708563e-08,
      "loss": 5.502282874658704e-06,
      "step": 648,
      "token_acc": 1.0
    },
    {
      "epoch": 3.863095238095238,
      "grad_norm": 0.000911712646484375,
      "learning_rate": 3.2032424132362736e-08,
      "loss": 1.8536920833867043e-06,
      "step": 649,
      "token_acc": 1.0
    },
    {
      "epoch": 3.869047619047619,
      "grad_norm": 37.0,
      "learning_rate": 2.9310214228202016e-08,
      "loss": 0.18283970654010773,
      "step": 650,
      "token_acc": 0.95
    },
    {
      "epoch": 3.875,
      "grad_norm": 55.75,
      "learning_rate": 2.6708528499950758e-08,
      "loss": 0.6082554459571838,
      "step": 651,
      "token_acc": 0.95
    },
    {
      "epoch": 3.880952380952381,
      "grad_norm": 7.0,
      "learning_rate": 2.4227430030609455e-08,
      "loss": 0.004444661550223827,
      "step": 652,
      "token_acc": 1.0
    },
    {
      "epoch": 3.886904761904762,
      "grad_norm": 7.25,
      "learning_rate": 2.1866978979303567e-08,
      "loss": 0.020752394571900368,
      "step": 653,
      "token_acc": 1.0
    },
    {
      "epoch": 3.892857142857143,
      "grad_norm": 52.75,
      "learning_rate": 1.962723257982302e-08,
      "loss": 0.0632629245519638,
      "step": 654,
      "token_acc": 0.9444444444444444
    },
    {
      "epoch": 3.8988095238095237,
      "grad_norm": 0.005462646484375,
      "learning_rate": 1.7508245139236658e-08,
      "loss": 6.178759122121846e-06,
      "step": 655,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9047619047619047,
      "grad_norm": 18.625,
      "learning_rate": 1.5510068036573288e-08,
      "loss": 0.010605025105178356,
      "step": 656,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9107142857142856,
      "grad_norm": 0.00183868408203125,
      "learning_rate": 1.3632749721577132e-08,
      "loss": 3.834502422250807e-06,
      "step": 657,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9166666666666665,
      "grad_norm": 0.0037841796875,
      "learning_rate": 1.1876335713532638e-08,
      "loss": 4.960263140674215e-06,
      "step": 658,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9226190476190474,
      "grad_norm": 0.1279296875,
      "learning_rate": 1.024086860016149e-08,
      "loss": 0.00021628677495755255,
      "step": 659,
      "token_acc": 1.0
    },
    {
      "epoch": 3.928571428571429,
      "grad_norm": 0.353515625,
      "learning_rate": 8.726388036587874e-09,
      "loss": 0.0006422073347494006,
      "step": 660,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9345238095238093,
      "grad_norm": 0.015869140625,
      "learning_rate": 7.332930744380906e-09,
      "loss": 3.2687334169168025e-05,
      "step": 661,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9404761904761907,
      "grad_norm": 0.0218505859375,
      "learning_rate": 6.060530510659246e-09,
      "loss": 2.0428480638656765e-05,
      "step": 662,
      "token_acc": 1.0
    },
    {
      "epoch": 3.946428571428571,
      "grad_norm": 0.0004367828369140625,
      "learning_rate": 4.909218187276743e-09,
      "loss": 1.4007055142428726e-06,
      "step": 663,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9523809523809526,
      "grad_norm": 0.00726318359375,
      "learning_rate": 3.8790216900702615e-09,
      "loss": 8.14547820482403e-06,
      "step": 664,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9583333333333335,
      "grad_norm": 0.24609375,
      "learning_rate": 2.9699659981863306e-09,
      "loss": 0.00034077290911227465,
      "step": 665,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9642857142857144,
      "grad_norm": 0.0015716552734375,
      "learning_rate": 2.182073153471631e-09,
      "loss": 3.6596625250240322e-06,
      "step": 666,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9702380952380953,
      "grad_norm": 15.0625,
      "learning_rate": 1.5153622599428652e-09,
      "loss": 0.02068321965634823,
      "step": 667,
      "token_acc": 1.0
    },
    {
      "epoch": 3.9761904761904763,
      "grad_norm": 0.01556396484375,
      "learning_rate": 9.698494833199068e-10,
      "loss": 2.165958721889183e-05,
      "step": 668,
      "token_acc": 1.0
    },
    {
      "epoch": 3.982142857142857,
      "grad_norm": 0.01416015625,
      "learning_rate": 5.455480506355582e-10,
      "loss": 1.3090951142658014e-05,
      "step": 669,
      "token_acc": 1.0
    },
    {
      "epoch": 3.988095238095238,
      "grad_norm": 16.25,
      "learning_rate": 2.4246824991525085e-10,
      "loss": 0.01774178445339203,
      "step": 670,
      "token_acc": 1.0
    },
    {
      "epoch": 3.994047619047619,
      "grad_norm": 2.328125,
      "learning_rate": 6.061742992613529e-11,
      "loss": 0.0012929077493026853,
      "step": 671,
      "token_acc": 1.0
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.05810546875,
      "learning_rate": 0.0,
      "loss": 9.619742922950536e-05,
      "step": 672,
      "token_acc": 1.0
    }
  ],
  "logging_steps": 1,
  "max_steps": 672,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 168,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.5052516831985664e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}