{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.743034055727556,
  "eval_steps": 500,
  "global_step": 6700,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.030959752321981424,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001,
      "loss": 2.397,
      "step": 10
    },
    {
      "epoch": 0.06191950464396285,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001,
      "loss": 2.1975,
      "step": 20
    },
    {
      "epoch": 0.09287925696594428,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001,
      "loss": 2.1624,
      "step": 30
    },
    {
      "epoch": 0.1238390092879257,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001,
      "loss": 2.0724,
      "step": 40
    },
    {
      "epoch": 0.15479876160990713,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0001,
      "loss": 2.0248,
      "step": 50
    },
    {
      "epoch": 0.18575851393188855,
      "grad_norm": 0.421875,
      "learning_rate": 0.0001,
      "loss": 2.2899,
      "step": 60
    },
    {
      "epoch": 0.21671826625386997,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001,
      "loss": 2.1192,
      "step": 70
    },
    {
      "epoch": 0.2476780185758514,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001,
      "loss": 2.0941,
      "step": 80
    },
    {
      "epoch": 0.2786377708978328,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001,
      "loss": 2.0773,
      "step": 90
    },
    {
      "epoch": 0.30959752321981426,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001,
      "loss": 1.9767,
      "step": 100
    },
    {
      "epoch": 0.34055727554179566,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001,
      "loss": 2.1887,
      "step": 110
    },
    {
      "epoch": 0.3715170278637771,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001,
      "loss": 2.0403,
      "step": 120
    },
    {
      "epoch": 0.4024767801857585,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001,
      "loss": 2.0129,
      "step": 130
    },
    {
      "epoch": 0.43343653250773995,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001,
      "loss": 2.0155,
      "step": 140
    },
    {
      "epoch": 0.46439628482972134,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0001,
      "loss": 2.0159,
      "step": 150
    },
    {
      "epoch": 0.4953560371517028,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001,
      "loss": 2.1815,
      "step": 160
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001,
      "loss": 2.0064,
      "step": 170
    },
    {
      "epoch": 0.5572755417956656,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001,
      "loss": 2.03,
      "step": 180
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001,
      "loss": 2.0138,
      "step": 190
    },
    {
      "epoch": 0.6191950464396285,
      "grad_norm": 1.53125,
      "learning_rate": 0.0001,
      "loss": 1.9765,
      "step": 200
    },
    {
      "epoch": 0.6501547987616099,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001,
      "loss": 2.2076,
      "step": 210
    },
    {
      "epoch": 0.6811145510835913,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001,
      "loss": 2.0659,
      "step": 220
    },
    {
      "epoch": 0.7120743034055728,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001,
      "loss": 2.0689,
      "step": 230
    },
    {
      "epoch": 0.7430340557275542,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001,
      "loss": 1.8855,
      "step": 240
    },
    {
      "epoch": 0.7739938080495357,
      "grad_norm": 1.3125,
      "learning_rate": 0.0001,
      "loss": 1.812,
      "step": 250
    },
    {
      "epoch": 0.804953560371517,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001,
      "loss": 2.1327,
      "step": 260
    },
    {
      "epoch": 0.8359133126934984,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001,
      "loss": 2.0221,
      "step": 270
    },
    {
      "epoch": 0.8668730650154799,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001,
      "loss": 1.9692,
      "step": 280
    },
    {
      "epoch": 0.8978328173374613,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001,
      "loss": 1.8887,
      "step": 290
    },
    {
      "epoch": 0.9287925696594427,
      "grad_norm": 1.453125,
      "learning_rate": 0.0001,
      "loss": 1.8405,
      "step": 300
    },
    {
      "epoch": 0.9597523219814241,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001,
      "loss": 2.0955,
      "step": 310
    },
    {
      "epoch": 0.9907120743034056,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001,
      "loss": 1.9367,
      "step": 320
    },
    {
      "epoch": 1.021671826625387,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001,
      "loss": 2.0857,
      "step": 330
    },
    {
      "epoch": 1.0526315789473684,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001,
      "loss": 1.8077,
      "step": 340
    },
    {
      "epoch": 1.08359133126935,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001,
      "loss": 1.5881,
      "step": 350
    },
    {
      "epoch": 1.1145510835913313,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001,
      "loss": 1.5215,
      "step": 360
    },
    {
      "epoch": 1.1455108359133126,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0001,
      "loss": 1.3684,
      "step": 370
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001,
      "loss": 1.7801,
      "step": 380
    },
    {
      "epoch": 1.2074303405572755,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001,
      "loss": 1.7372,
      "step": 390
    },
    {
      "epoch": 1.238390092879257,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001,
      "loss": 1.6767,
      "step": 400
    },
    {
      "epoch": 1.2693498452012384,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001,
      "loss": 1.4923,
      "step": 410
    },
    {
      "epoch": 1.3003095975232197,
      "grad_norm": 1.34375,
      "learning_rate": 0.0001,
      "loss": 1.2514,
      "step": 420
    },
    {
      "epoch": 1.3312693498452013,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001,
      "loss": 1.7803,
      "step": 430
    },
    {
      "epoch": 1.3622291021671826,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001,
      "loss": 1.7424,
      "step": 440
    },
    {
      "epoch": 1.3931888544891642,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001,
      "loss": 1.6923,
      "step": 450
    },
    {
      "epoch": 1.4241486068111455,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001,
      "loss": 1.4639,
      "step": 460
    },
    {
      "epoch": 1.4551083591331269,
      "grad_norm": 1.5234375,
      "learning_rate": 0.0001,
      "loss": 1.4187,
      "step": 470
    },
    {
      "epoch": 1.4860681114551084,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001,
      "loss": 1.7615,
      "step": 480
    },
    {
      "epoch": 1.5170278637770898,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001,
      "loss": 1.8067,
      "step": 490
    },
    {
      "epoch": 1.5479876160990713,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001,
      "loss": 1.6603,
      "step": 500
    },
    {
      "epoch": 1.5789473684210527,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001,
      "loss": 1.4714,
      "step": 510
    },
    {
      "epoch": 1.609907120743034,
      "grad_norm": 1.53125,
      "learning_rate": 0.0001,
      "loss": 1.3003,
      "step": 520
    },
    {
      "epoch": 1.6408668730650153,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001,
      "loss": 1.783,
      "step": 530
    },
    {
      "epoch": 1.671826625386997,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001,
      "loss": 1.8179,
      "step": 540
    },
    {
      "epoch": 1.7027863777089784,
      "grad_norm": 1.0,
      "learning_rate": 0.0001,
      "loss": 1.6297,
      "step": 550
    },
    {
      "epoch": 1.7337461300309598,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001,
      "loss": 1.5301,
      "step": 560
    },
    {
      "epoch": 1.7647058823529411,
      "grad_norm": 1.5625,
      "learning_rate": 0.0001,
      "loss": 1.3402,
      "step": 570
    },
    {
      "epoch": 1.7956656346749225,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001,
      "loss": 1.83,
      "step": 580
    },
    {
      "epoch": 1.826625386996904,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001,
      "loss": 1.774,
      "step": 590
    },
    {
      "epoch": 1.8575851393188856,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001,
      "loss": 1.6684,
      "step": 600
    },
    {
      "epoch": 1.888544891640867,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001,
      "loss": 1.5126,
      "step": 610
    },
    {
      "epoch": 1.9195046439628483,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001,
      "loss": 1.3526,
      "step": 620
    },
    {
      "epoch": 1.9504643962848296,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001,
      "loss": 1.6972,
      "step": 630
    },
    {
      "epoch": 1.9814241486068112,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001,
      "loss": 1.5535,
      "step": 640
    },
    {
      "epoch": 2.0123839009287927,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001,
      "loss": 1.5312,
      "step": 650
    },
    {
      "epoch": 2.043343653250774,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001,
      "loss": 1.4688,
      "step": 660
    },
    {
      "epoch": 2.0743034055727554,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001,
      "loss": 1.1771,
      "step": 670
    },
    {
      "epoch": 2.1052631578947367,
      "grad_norm": 1.828125,
      "learning_rate": 0.0001,
      "loss": 0.9073,
      "step": 680
    },
    {
      "epoch": 2.136222910216718,
      "grad_norm": 1.875,
      "learning_rate": 0.0001,
      "loss": 0.7219,
      "step": 690
    },
    {
      "epoch": 2.1671826625387,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001,
      "loss": 1.0551,
      "step": 700
    },
    {
      "epoch": 2.198142414860681,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001,
      "loss": 1.5082,
      "step": 710
    },
    {
      "epoch": 2.2291021671826625,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001,
      "loss": 1.1655,
      "step": 720
    },
    {
      "epoch": 2.260061919504644,
      "grad_norm": 1.578125,
      "learning_rate": 0.0001,
      "loss": 0.9166,
      "step": 730
    },
    {
      "epoch": 2.291021671826625,
      "grad_norm": 1.96875,
      "learning_rate": 0.0001,
      "loss": 0.711,
      "step": 740
    },
    {
      "epoch": 2.321981424148607,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001,
      "loss": 1.0771,
      "step": 750
    },
    {
      "epoch": 2.3529411764705883,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001,
      "loss": 1.4724,
      "step": 760
    },
    {
      "epoch": 2.3839009287925697,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001,
      "loss": 1.1995,
      "step": 770
    },
    {
      "epoch": 2.414860681114551,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001,
      "loss": 0.886,
      "step": 780
    },
    {
      "epoch": 2.4458204334365323,
      "grad_norm": 1.8984375,
      "learning_rate": 0.0001,
      "loss": 0.6432,
      "step": 790
    },
    {
      "epoch": 2.476780185758514,
      "grad_norm": 1.125,
      "learning_rate": 0.0001,
      "loss": 1.0749,
      "step": 800
    },
    {
      "epoch": 2.5077399380804954,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001,
      "loss": 1.4332,
      "step": 810
    },
    {
      "epoch": 2.538699690402477,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001,
      "loss": 1.2049,
      "step": 820
    },
    {
      "epoch": 2.569659442724458,
      "grad_norm": 1.65625,
      "learning_rate": 0.0001,
      "loss": 0.9171,
      "step": 830
    },
    {
      "epoch": 2.6006191950464395,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0001,
      "loss": 0.7159,
      "step": 840
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001,
      "loss": 1.0601,
      "step": 850
    },
    {
      "epoch": 2.6625386996904026,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001,
      "loss": 1.4824,
      "step": 860
    },
    {
      "epoch": 2.693498452012384,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001,
      "loss": 1.2073,
      "step": 870
    },
    {
      "epoch": 2.7244582043343653,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0001,
      "loss": 0.9479,
      "step": 880
    },
    {
      "epoch": 2.7554179566563466,
      "grad_norm": 1.609375,
      "learning_rate": 0.0001,
      "loss": 0.7719,
      "step": 890
    },
    {
      "epoch": 2.7863777089783284,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001,
      "loss": 1.1318,
      "step": 900
    },
    {
      "epoch": 2.8173374613003097,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001,
      "loss": 1.4771,
      "step": 910
    },
    {
      "epoch": 2.848297213622291,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001,
      "loss": 1.1876,
      "step": 920
    },
    {
      "epoch": 2.8792569659442724,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0001,
      "loss": 0.9698,
      "step": 930
    },
    {
      "epoch": 2.9102167182662537,
      "grad_norm": 1.6875,
      "learning_rate": 0.0001,
      "loss": 0.7357,
      "step": 940
    },
    {
      "epoch": 2.9411764705882355,
      "grad_norm": 1.375,
      "learning_rate": 0.0001,
      "loss": 1.0427,
      "step": 950
    },
    {
      "epoch": 2.972136222910217,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001,
      "loss": 1.1243,
      "step": 960
    },
    {
      "epoch": 3.003095975232198,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001,
      "loss": 0.7874,
      "step": 970
    },
    {
      "epoch": 3.0340557275541795,
      "grad_norm": 1.578125,
      "learning_rate": 0.0001,
      "loss": 1.1201,
      "step": 980
    },
    {
      "epoch": 3.065015479876161,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001,
      "loss": 0.6787,
      "step": 990
    },
    {
      "epoch": 3.0959752321981426,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0001,
      "loss": 0.4248,
      "step": 1000
    },
    {
      "epoch": 3.126934984520124,
      "grad_norm": 1.640625,
      "learning_rate": 0.0001,
      "loss": 0.3441,
      "step": 1010
    },
    {
      "epoch": 3.1578947368421053,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001,
      "loss": 0.3431,
      "step": 1020
    },
    {
      "epoch": 3.1888544891640866,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001,
      "loss": 1.0994,
      "step": 1030
    },
    {
      "epoch": 3.219814241486068,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0001,
      "loss": 0.6538,
      "step": 1040
    },
    {
      "epoch": 3.2507739938080498,
      "grad_norm": 1.9140625,
      "learning_rate": 0.0001,
      "loss": 0.4197,
      "step": 1050
    },
    {
      "epoch": 3.281733746130031,
      "grad_norm": 1.984375,
      "learning_rate": 0.0001,
      "loss": 0.322,
      "step": 1060
    },
    {
      "epoch": 3.3126934984520124,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001,
      "loss": 0.382,
      "step": 1070
    },
    {
      "epoch": 3.343653250773994,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001,
      "loss": 1.1316,
      "step": 1080
    },
    {
      "epoch": 3.374613003095975,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0001,
      "loss": 0.7036,
      "step": 1090
    },
    {
      "epoch": 3.405572755417957,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0001,
      "loss": 0.4367,
      "step": 1100
    },
    {
      "epoch": 3.4365325077399382,
      "grad_norm": 2.078125,
      "learning_rate": 0.0001,
      "loss": 0.2773,
      "step": 1110
    },
    {
      "epoch": 3.4674922600619196,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001,
      "loss": 0.3452,
      "step": 1120
    },
    {
      "epoch": 3.498452012383901,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001,
      "loss": 1.0893,
      "step": 1130
    },
    {
      "epoch": 3.5294117647058822,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001,
      "loss": 0.6688,
      "step": 1140
    },
    {
      "epoch": 3.560371517027864,
      "grad_norm": 1.7109375,
      "learning_rate": 0.0001,
      "loss": 0.4231,
      "step": 1150
    },
    {
      "epoch": 3.5913312693498454,
      "grad_norm": 1.890625,
      "learning_rate": 0.0001,
      "loss": 0.3495,
      "step": 1160
    },
    {
      "epoch": 3.6222910216718267,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001,
      "loss": 0.3711,
      "step": 1170
    },
    {
      "epoch": 3.653250773993808,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001,
      "loss": 1.0888,
      "step": 1180
    },
    {
      "epoch": 3.6842105263157894,
      "grad_norm": 1.484375,
      "learning_rate": 0.0001,
      "loss": 0.6578,
      "step": 1190
    },
    {
      "epoch": 3.715170278637771,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001,
      "loss": 0.4095,
      "step": 1200
    },
    {
      "epoch": 3.746130030959752,
      "grad_norm": 1.671875,
      "learning_rate": 0.0001,
      "loss": 0.3025,
      "step": 1210
    },
    {
      "epoch": 3.777089783281734,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001,
      "loss": 0.3888,
      "step": 1220
    },
    {
      "epoch": 3.808049535603715,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0001,
      "loss": 1.1246,
      "step": 1230
    },
    {
      "epoch": 3.8390092879256965,
      "grad_norm": 1.375,
      "learning_rate": 0.0001,
      "loss": 0.7352,
      "step": 1240
    },
    {
      "epoch": 3.8699690402476783,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0001,
      "loss": 0.4738,
      "step": 1250
    },
    {
      "epoch": 3.900928792569659,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0001,
      "loss": 0.3303,
      "step": 1260
    },
    {
      "epoch": 3.931888544891641,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001,
      "loss": 0.3753,
      "step": 1270
    },
    {
      "epoch": 3.9628482972136223,
      "grad_norm": 1.46875,
      "learning_rate": 0.0001,
      "loss": 0.8517,
      "step": 1280
    },
    {
      "epoch": 3.9938080495356036,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0001,
      "loss": 0.3625,
      "step": 1290
    },
    {
      "epoch": 4.024767801857585,
      "grad_norm": 1.5,
      "learning_rate": 0.0001,
      "loss": 0.6243,
      "step": 1300
    },
    {
      "epoch": 4.055727554179566,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001,
      "loss": 0.3312,
      "step": 1310
    },
    {
      "epoch": 4.086687306501548,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001,
      "loss": 0.1841,
      "step": 1320
    },
    {
      "epoch": 4.117647058823529,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0001,
      "loss": 0.1307,
      "step": 1330
    },
    {
      "epoch": 4.148606811145511,
      "grad_norm": 1.84375,
      "learning_rate": 0.0001,
      "loss": 0.1003,
      "step": 1340
    },
    {
      "epoch": 4.179566563467493,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001,
      "loss": 0.6324,
      "step": 1350
    },
    {
      "epoch": 4.2105263157894735,
      "grad_norm": 1.625,
      "learning_rate": 0.0001,
      "loss": 0.3237,
      "step": 1360
    },
    {
      "epoch": 4.241486068111455,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001,
      "loss": 0.1651,
      "step": 1370
    },
    {
      "epoch": 4.272445820433436,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001,
      "loss": 0.1206,
      "step": 1380
    },
    {
      "epoch": 4.303405572755418,
      "grad_norm": 2.171875,
      "learning_rate": 0.0001,
      "loss": 0.1298,
      "step": 1390
    },
    {
      "epoch": 4.3343653250774,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001,
      "loss": 0.5857,
      "step": 1400
    },
    {
      "epoch": 4.365325077399381,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001,
      "loss": 0.3743,
      "step": 1410
    },
    {
      "epoch": 4.396284829721362,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001,
      "loss": 0.2053,
      "step": 1420
    },
    {
      "epoch": 4.427244582043343,
      "grad_norm": 1.703125,
      "learning_rate": 0.0001,
      "loss": 0.1376,
      "step": 1430
    },
    {
      "epoch": 4.458204334365325,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0001,
      "loss": 0.1043,
      "step": 1440
    },
    {
      "epoch": 4.489164086687307,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001,
      "loss": 0.6281,
      "step": 1450
    },
    {
      "epoch": 4.520123839009288,
      "grad_norm": 1.640625,
      "learning_rate": 0.0001,
      "loss": 0.3621,
      "step": 1460
    },
    {
      "epoch": 4.5510835913312695,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001,
      "loss": 0.2105,
      "step": 1470
    },
    {
      "epoch": 4.58204334365325,
      "grad_norm": 1.5625,
      "learning_rate": 0.0001,
      "loss": 0.1891,
      "step": 1480
    },
    {
      "epoch": 4.613003095975232,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001,
      "loss": 0.1215,
      "step": 1490
    },
    {
      "epoch": 4.643962848297214,
      "grad_norm": 1.4375,
      "learning_rate": 0.0001,
      "loss": 0.6574,
      "step": 1500
    },
    {
      "epoch": 4.674922600619195,
      "grad_norm": 1.8515625,
      "learning_rate": 0.0001,
      "loss": 0.3907,
      "step": 1510
    },
    {
      "epoch": 4.705882352941177,
      "grad_norm": 1.4375,
      "learning_rate": 0.0001,
      "loss": 0.2264,
      "step": 1520
    },
    {
      "epoch": 4.7368421052631575,
      "grad_norm": 1.4375,
      "learning_rate": 0.0001,
      "loss": 0.1498,
      "step": 1530
    },
    {
      "epoch": 4.767801857585139,
      "grad_norm": 1.375,
      "learning_rate": 0.0001,
      "loss": 0.1238,
      "step": 1540
    },
    {
      "epoch": 4.798761609907121,
      "grad_norm": 1.359375,
      "learning_rate": 0.0001,
      "loss": 0.65,
      "step": 1550
    },
    {
      "epoch": 4.829721362229102,
      "grad_norm": 1.609375,
      "learning_rate": 0.0001,
      "loss": 0.3445,
      "step": 1560
    },
    {
      "epoch": 4.860681114551084,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001,
      "loss": 0.2066,
      "step": 1570
    },
    {
      "epoch": 4.891640866873065,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001,
      "loss": 0.143,
      "step": 1580
    },
    {
      "epoch": 4.922600619195046,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0001,
      "loss": 0.1308,
      "step": 1590
    },
    {
      "epoch": 4.953560371517028,
      "grad_norm": 1.453125,
      "learning_rate": 0.0001,
      "loss": 0.4828,
      "step": 1600
    },
    {
      "epoch": 4.984520123839009,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0001,
      "loss": 0.1781,
      "step": 1610
    },
    {
      "epoch": 5.015479876160991,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001,
      "loss": 0.3468,
      "step": 1620
    },
    {
      "epoch": 5.046439628482972,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0001,
      "loss": 0.1674,
      "step": 1630
    },
    {
      "epoch": 5.077399380804954,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0001,
      "loss": 0.0983,
      "step": 1640
    },
    {
      "epoch": 5.108359133126935,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001,
      "loss": 0.0648,
      "step": 1650
    },
    {
      "epoch": 5.139318885448916,
      "grad_norm": 1.765625,
      "learning_rate": 0.0001,
      "loss": 0.0799,
      "step": 1660
    },
    {
      "epoch": 5.170278637770898,
      "grad_norm": 1.125,
      "learning_rate": 0.0001,
      "loss": 0.2795,
      "step": 1670
    },
    {
      "epoch": 5.201238390092879,
      "grad_norm": 1.453125,
      "learning_rate": 0.0001,
      "loss": 0.1749,
      "step": 1680
    },
    {
      "epoch": 5.232198142414861,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001,
      "loss": 0.1115,
      "step": 1690
    },
    {
      "epoch": 5.2631578947368425,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001,
      "loss": 0.0732,
      "step": 1700
    },
    {
      "epoch": 5.294117647058823,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001,
      "loss": 0.0614,
      "step": 1710
    },
    {
      "epoch": 5.325077399380805,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001,
      "loss": 0.29,
      "step": 1720
    },
    {
      "epoch": 5.356037151702786,
      "grad_norm": 1.59375,
      "learning_rate": 0.0001,
      "loss": 0.1984,
      "step": 1730
    },
    {
      "epoch": 5.386996904024768,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001,
      "loss": 0.0959,
      "step": 1740
    },
    {
      "epoch": 5.41795665634675,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001,
      "loss": 0.075,
      "step": 1750
    },
    {
      "epoch": 5.4489164086687305,
      "grad_norm": 1.65625,
      "learning_rate": 0.0001,
      "loss": 0.0607,
      "step": 1760
    },
    {
      "epoch": 5.479876160990712,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001,
      "loss": 0.2815,
      "step": 1770
    },
    {
      "epoch": 5.510835913312693,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0001,
      "loss": 0.1861,
      "step": 1780
    },
    {
      "epoch": 5.541795665634675,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001,
      "loss": 0.1012,
      "step": 1790
    },
    {
      "epoch": 5.572755417956657,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0001,
      "loss": 0.067,
      "step": 1800
    },
    {
      "epoch": 5.603715170278638,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0001,
      "loss": 0.0749,
      "step": 1810
    },
    {
      "epoch": 5.634674922600619,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001,
      "loss": 0.2726,
      "step": 1820
    },
    {
      "epoch": 5.6656346749226,
      "grad_norm": 2.15625,
      "learning_rate": 0.0001,
      "loss": 0.1799,
      "step": 1830
    },
    {
      "epoch": 5.696594427244582,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001,
      "loss": 0.1021,
      "step": 1840
    },
    {
      "epoch": 5.727554179566564,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001,
      "loss": 0.0691,
      "step": 1850
    },
    {
      "epoch": 5.758513931888545,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001,
      "loss": 0.0645,
      "step": 1860
    },
    {
      "epoch": 5.7894736842105265,
      "grad_norm": 1.3125,
      "learning_rate": 0.0001,
      "loss": 0.3198,
      "step": 1870
    },
    {
      "epoch": 5.820433436532507,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001,
      "loss": 0.218,
      "step": 1880
    },
    {
      "epoch": 5.851393188854489,
      "grad_norm": 1.359375,
      "learning_rate": 0.0001,
      "loss": 0.1078,
      "step": 1890
    },
    {
      "epoch": 5.882352941176471,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001,
      "loss": 0.0751,
      "step": 1900
    },
    {
      "epoch": 5.913312693498452,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001,
      "loss": 0.0749,
      "step": 1910
    },
    {
      "epoch": 5.944272445820434,
      "grad_norm": 1.546875,
      "learning_rate": 0.0001,
      "loss": 0.2301,
      "step": 1920
    },
    {
      "epoch": 5.975232198142415,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0001,
      "loss": 0.1302,
      "step": 1930
    },
    {
      "epoch": 6.006191950464396,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001,
      "loss": 0.1467,
      "step": 1940
    },
    {
      "epoch": 6.037151702786378,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001,
      "loss": 0.1079,
      "step": 1950
    },
    {
      "epoch": 6.068111455108359,
      "grad_norm": 1.9765625,
      "learning_rate": 0.0001,
      "loss": 0.0593,
      "step": 1960
    },
    {
      "epoch": 6.099071207430341,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001,
      "loss": 0.0459,
      "step": 1970
    },
    {
      "epoch": 6.130030959752322,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001,
      "loss": 0.0337,
      "step": 1980
    },
    {
      "epoch": 6.1609907120743035,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001,
      "loss": 0.1309,
      "step": 1990
    },
    {
      "epoch": 6.191950464396285,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001,
      "loss": 0.1197,
      "step": 2000
    },
    {
      "epoch": 6.222910216718266,
      "grad_norm": 1.484375,
      "learning_rate": 0.0001,
      "loss": 0.0668,
      "step": 2010
    },
    {
      "epoch": 6.253869969040248,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001,
      "loss": 0.0569,
      "step": 2020
    },
    {
      "epoch": 6.284829721362229,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0001,
      "loss": 0.0432,
      "step": 2030
    },
    {
      "epoch": 6.315789473684211,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001,
      "loss": 0.1081,
      "step": 2040
    },
    {
      "epoch": 6.346749226006192,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001,
      "loss": 0.1104,
      "step": 2050
    },
    {
      "epoch": 6.377708978328173,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0001,
      "loss": 0.0799,
      "step": 2060
    },
    {
      "epoch": 6.408668730650155,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001,
      "loss": 0.0499,
      "step": 2070
    },
    {
      "epoch": 6.439628482972136,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001,
      "loss": 0.0472,
      "step": 2080
    },
    {
      "epoch": 6.470588235294118,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001,
      "loss": 0.1062,
      "step": 2090
    },
    {
      "epoch": 6.5015479876160995,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001,
      "loss": 0.1386,
      "step": 2100
    },
    {
      "epoch": 6.53250773993808,
      "grad_norm": 1.734375,
      "learning_rate": 0.0001,
      "loss": 0.0756,
      "step": 2110
    },
    {
      "epoch": 6.563467492260062,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001,
      "loss": 0.0602,
      "step": 2120
    },
    {
      "epoch": 6.594427244582043,
      "grad_norm": 1.640625,
      "learning_rate": 0.0001,
      "loss": 0.0422,
      "step": 2130
    },
    {
      "epoch": 6.625386996904025,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001,
      "loss": 0.1138,
      "step": 2140
    },
    {
      "epoch": 6.656346749226007,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001,
      "loss": 0.1462,
      "step": 2150
    },
    {
      "epoch": 6.687306501547988,
      "grad_norm": 1.3125,
      "learning_rate": 0.0001,
      "loss": 0.0758,
      "step": 2160
    },
    {
      "epoch": 6.718266253869969,
      "grad_norm": 1.5625,
      "learning_rate": 0.0001,
      "loss": 0.0621,
      "step": 2170
    },
    {
      "epoch": 6.74922600619195,
      "grad_norm": 2.4375,
      "learning_rate": 0.0001,
      "loss": 0.0544,
      "step": 2180
    },
    {
      "epoch": 6.780185758513932,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001,
      "loss": 0.1573,
      "step": 2190
    },
    {
      "epoch": 6.811145510835914,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001,
      "loss": 0.1575,
      "step": 2200
    },
    {
      "epoch": 6.842105263157895,
      "grad_norm": 1.5625,
      "learning_rate": 0.0001,
      "loss": 0.0724,
      "step": 2210
    },
    {
      "epoch": 6.8730650154798765,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001,
      "loss": 0.0658,
      "step": 2220
    },
    {
      "epoch": 6.904024767801857,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001,
      "loss": 0.0435,
      "step": 2230
    },
    {
      "epoch": 6.934984520123839,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001,
      "loss": 0.0894,
      "step": 2240
    },
    {
      "epoch": 6.965944272445821,
      "grad_norm": 1.34375,
      "learning_rate": 0.0001,
      "loss": 0.1117,
      "step": 2250
    },
    {
      "epoch": 6.996904024767802,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001,
      "loss": 0.0593,
      "step": 2260
    },
    {
      "epoch": 7.027863777089784,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001,
      "loss": 0.0889,
      "step": 2270
    },
    {
      "epoch": 7.0588235294117645,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001,
      "loss": 0.0476,
      "step": 2280
    },
    {
      "epoch": 7.089783281733746,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001,
      "loss": 0.0382,
      "step": 2290
    },
    {
      "epoch": 7.120743034055727,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0001,
      "loss": 0.0366,
      "step": 2300
    },
    {
      "epoch": 7.151702786377709,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001,
      "loss": 0.0277,
      "step": 2310
    },
    {
      "epoch": 7.182662538699691,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001,
      "loss": 0.1059,
      "step": 2320
    },
    {
      "epoch": 7.213622291021672,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001,
      "loss": 0.0564,
      "step": 2330
    },
    {
      "epoch": 7.244582043343653,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001,
      "loss": 0.0435,
      "step": 2340
    },
    {
      "epoch": 7.275541795665634,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001,
      "loss": 0.0381,
      "step": 2350
    },
    {
      "epoch": 7.306501547987616,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001,
      "loss": 0.0567,
      "step": 2360
    },
    {
      "epoch": 7.337461300309598,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001,
      "loss": 0.1014,
      "step": 2370
    },
    {
      "epoch": 7.368421052631579,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001,
      "loss": 0.0596,
      "step": 2380
    },
    {
      "epoch": 7.3993808049535605,
      "grad_norm": 1.625,
      "learning_rate": 0.0001,
      "loss": 0.047,
      "step": 2390
    },
    {
      "epoch": 7.430340557275541,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001,
      "loss": 0.0298,
      "step": 2400
    },
    {
      "epoch": 7.461300309597523,
      "grad_norm": 2.234375,
      "learning_rate": 0.0001,
      "loss": 0.0365,
      "step": 2410
    },
    {
      "epoch": 7.492260061919505,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001,
      "loss": 0.1112,
      "step": 2420
    },
    {
      "epoch": 7.523219814241486,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001,
      "loss": 0.0604,
      "step": 2430
    },
    {
      "epoch": 7.554179566563468,
      "grad_norm": 1.9296875,
      "learning_rate": 0.0001,
      "loss": 0.0465,
      "step": 2440
    },
    {
      "epoch": 7.585139318885449,
      "grad_norm": 1.546875,
      "learning_rate": 0.0001,
      "loss": 0.046,
      "step": 2450
    },
    {
      "epoch": 7.61609907120743,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001,
      "loss": 0.0346,
      "step": 2460
    },
    {
      "epoch": 7.647058823529412,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001,
      "loss": 0.1163,
      "step": 2470
    },
    {
      "epoch": 7.678018575851393,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001,
      "loss": 0.058,
      "step": 2480
    },
    {
      "epoch": 7.708978328173375,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001,
      "loss": 0.0476,
      "step": 2490
    },
    {
      "epoch": 7.739938080495356,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001,
      "loss": 0.0487,
      "step": 2500
    },
    {
      "epoch": 7.7708978328173375,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0001,
      "loss": 0.0327,
      "step": 2510
    },
    {
      "epoch": 7.801857585139319,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001,
      "loss": 0.1199,
      "step": 2520
    },
    {
      "epoch": 7.8328173374613,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001,
      "loss": 0.0639,
      "step": 2530
    },
    {
      "epoch": 7.863777089783282,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001,
      "loss": 0.0535,
      "step": 2540
    },
    {
      "epoch": 7.894736842105263,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001,
      "loss": 0.0386,
      "step": 2550
    },
    {
      "epoch": 7.925696594427245,
      "grad_norm": 2.5,
      "learning_rate": 0.0001,
      "loss": 0.0523,
      "step": 2560
    },
    {
      "epoch": 7.956656346749226,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001,
      "loss": 0.1184,
      "step": 2570
    },
    {
      "epoch": 7.987616099071207,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001,
      "loss": 0.0558,
      "step": 2580
    },
    {
      "epoch": 8.018575851393189,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001,
      "loss": 0.0601,
      "step": 2590
    },
    {
      "epoch": 8.04953560371517,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001,
      "loss": 0.0375,
      "step": 2600
    },
    {
      "epoch": 8.08049535603715,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001,
      "loss": 0.0317,
      "step": 2610
    },
    {
      "epoch": 8.111455108359133,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001,
      "loss": 0.0322,
      "step": 2620
    },
    {
      "epoch": 8.142414860681114,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001,
      "loss": 0.024,
      "step": 2630
    },
    {
      "epoch": 8.173374613003096,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001,
      "loss": 0.061,
      "step": 2640
    },
    {
      "epoch": 8.204334365325078,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001,
      "loss": 0.0417,
      "step": 2650
    },
    {
      "epoch": 8.235294117647058,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001,
      "loss": 0.0312,
      "step": 2660
    },
    {
      "epoch": 8.26625386996904,
      "grad_norm": 1.375,
      "learning_rate": 0.0001,
      "loss": 0.0365,
      "step": 2670
    },
    {
      "epoch": 8.297213622291022,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001,
      "loss": 0.025,
      "step": 2680
    },
    {
      "epoch": 8.328173374613003,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001,
      "loss": 0.0621,
      "step": 2690
    },
    {
      "epoch": 8.359133126934985,
      "grad_norm": 1.0,
      "learning_rate": 0.0001,
      "loss": 0.0552,
      "step": 2700
    },
    {
      "epoch": 8.390092879256965,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0001,
      "loss": 0.0391,
      "step": 2710
    },
    {
      "epoch": 8.421052631578947,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001,
      "loss": 0.0272,
      "step": 2720
    },
    {
      "epoch": 8.452012383900929,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001,
      "loss": 0.0184,
      "step": 2730
    },
    {
      "epoch": 8.48297213622291,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001,
      "loss": 0.071,
      "step": 2740
    },
    {
      "epoch": 8.513931888544892,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001,
      "loss": 0.0484,
      "step": 2750
    },
    {
      "epoch": 8.544891640866872,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001,
      "loss": 0.0353,
      "step": 2760
    },
    {
      "epoch": 8.575851393188854,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001,
      "loss": 0.0363,
      "step": 2770
    },
    {
      "epoch": 8.606811145510836,
      "grad_norm": 2.53125,
      "learning_rate": 0.0001,
      "loss": 0.0356,
      "step": 2780
    },
    {
      "epoch": 8.637770897832818,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001,
      "loss": 0.0819,
      "step": 2790
    },
    {
      "epoch": 8.6687306501548,
      "grad_norm": 1.359375,
      "learning_rate": 0.0001,
      "loss": 0.057,
      "step": 2800
    },
    {
      "epoch": 8.69969040247678,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001,
      "loss": 0.0374,
      "step": 2810
    },
    {
      "epoch": 8.730650154798761,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001,
      "loss": 0.0262,
      "step": 2820
    },
    {
      "epoch": 8.761609907120743,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001,
      "loss": 0.0347,
      "step": 2830
    },
    {
      "epoch": 8.792569659442725,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001,
      "loss": 0.0593,
      "step": 2840
    },
    {
      "epoch": 8.823529411764707,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001,
      "loss": 0.0566,
      "step": 2850
    },
    {
      "epoch": 8.854489164086687,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001,
      "loss": 0.0392,
      "step": 2860
    },
    {
      "epoch": 8.885448916408668,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001,
      "loss": 0.0423,
      "step": 2870
    },
    {
      "epoch": 8.91640866873065,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0001,
      "loss": 0.0414,
      "step": 2880
    },
    {
      "epoch": 8.947368421052632,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001,
      "loss": 0.0644,
      "step": 2890
    },
    {
      "epoch": 8.978328173374614,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001,
      "loss": 0.0445,
      "step": 2900
    },
    {
      "epoch": 9.009287925696594,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001,
      "loss": 0.039,
      "step": 2910
    },
    {
      "epoch": 9.040247678018575,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0001,
      "loss": 0.0326,
      "step": 2920
    },
    {
      "epoch": 9.071207430340557,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001,
      "loss": 0.0283,
      "step": 2930
    },
    {
      "epoch": 9.102167182662539,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001,
      "loss": 0.0239,
      "step": 2940
    },
    {
      "epoch": 9.13312693498452,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001,
      "loss": 0.023,
      "step": 2950
    },
    {
      "epoch": 9.1640866873065,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001,
      "loss": 0.0315,
      "step": 2960
    },
    {
      "epoch": 9.195046439628483,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001,
      "loss": 0.0473,
      "step": 2970
    },
    {
      "epoch": 9.226006191950464,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001,
      "loss": 0.0423,
      "step": 2980
    },
    {
      "epoch": 9.256965944272446,
      "grad_norm": 1.7421875,
      "learning_rate": 0.0001,
      "loss": 0.0269,
      "step": 2990
    },
    {
      "epoch": 9.287925696594428,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001,
      "loss": 0.0239,
      "step": 3000
    },
    {
      "epoch": 9.318885448916408,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001,
      "loss": 0.057,
      "step": 3010
    },
    {
      "epoch": 9.34984520123839,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001,
      "loss": 0.048,
      "step": 3020
    },
    {
      "epoch": 9.380804953560371,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001,
      "loss": 0.0347,
      "step": 3030
    },
    {
      "epoch": 9.411764705882353,
      "grad_norm": 1.5625,
      "learning_rate": 0.0001,
      "loss": 0.0284,
      "step": 3040
    },
    {
      "epoch": 9.442724458204335,
      "grad_norm": 0.625,
      "learning_rate": 0.0001,
      "loss": 0.0276,
      "step": 3050
    },
    {
      "epoch": 9.473684210526315,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001,
      "loss": 0.0434,
      "step": 3060
    },
    {
      "epoch": 9.504643962848297,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001,
      "loss": 0.0446,
      "step": 3070
    },
    {
      "epoch": 9.535603715170279,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001,
      "loss": 0.0377,
      "step": 3080
    },
    {
      "epoch": 9.56656346749226,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001,
      "loss": 0.046,
      "step": 3090
    },
    {
      "epoch": 9.597523219814242,
      "grad_norm": 0.265625,
      "learning_rate": 0.0001,
      "loss": 0.0203,
      "step": 3100
    },
    {
      "epoch": 9.628482972136222,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001,
      "loss": 0.0403,
      "step": 3110
    },
    {
      "epoch": 9.659442724458204,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001,
      "loss": 0.0488,
      "step": 3120
    },
    {
      "epoch": 9.690402476780186,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001,
      "loss": 0.0418,
      "step": 3130
    },
    {
      "epoch": 9.721362229102168,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001,
      "loss": 0.0314,
      "step": 3140
    },
    {
      "epoch": 9.75232198142415,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001,
      "loss": 0.0369,
      "step": 3150
    },
    {
      "epoch": 9.78328173374613,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001,
      "loss": 0.047,
      "step": 3160
    },
    {
      "epoch": 9.814241486068111,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001,
      "loss": 0.053,
      "step": 3170
    },
    {
      "epoch": 9.845201238390093,
      "grad_norm": 1.3125,
      "learning_rate": 0.0001,
      "loss": 0.0413,
      "step": 3180
    },
    {
      "epoch": 9.876160990712075,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001,
      "loss": 0.0332,
      "step": 3190
    },
    {
      "epoch": 9.907120743034056,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001,
      "loss": 0.036,
      "step": 3200
    },
    {
      "epoch": 9.938080495356036,
      "grad_norm": 1.125,
      "learning_rate": 0.0001,
      "loss": 0.0494,
      "step": 3210
    },
    {
      "epoch": 9.969040247678018,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001,
      "loss": 0.0505,
      "step": 3220
    },
    {
      "epoch": 10.0,
      "grad_norm": 7.59375,
      "learning_rate": 0.0001,
      "loss": 0.0234,
      "step": 3230
    },
    {
      "epoch": 10.030959752321982,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001,
      "loss": 0.0444,
      "step": 3240
    },
    {
      "epoch": 10.061919504643964,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001,
      "loss": 0.0252,
      "step": 3250
    },
    {
      "epoch": 10.092879256965944,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001,
      "loss": 0.0302,
      "step": 3260
    },
    {
      "epoch": 10.123839009287925,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001,
      "loss": 0.0198,
      "step": 3270
    },
    {
      "epoch": 10.154798761609907,
      "grad_norm": 1.65625,
      "learning_rate": 0.0001,
      "loss": 0.0334,
      "step": 3280
    },
    {
      "epoch": 10.185758513931889,
      "grad_norm": 0.875,
      "learning_rate": 0.0001,
      "loss": 0.0425,
      "step": 3290
    },
    {
      "epoch": 10.21671826625387,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001,
      "loss": 0.035,
      "step": 3300
    },
    {
      "epoch": 10.24767801857585,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001,
      "loss": 0.0222,
      "step": 3310
    },
    {
      "epoch": 10.278637770897832,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001,
      "loss": 0.0189,
      "step": 3320
    },
    {
      "epoch": 10.309597523219814,
      "grad_norm": 0.059326171875,
      "learning_rate": 0.0001,
      "loss": 0.0338,
      "step": 3330
    },
    {
      "epoch": 10.340557275541796,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001,
      "loss": 0.0481,
      "step": 3340
    },
    {
      "epoch": 10.371517027863778,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001,
      "loss": 0.0405,
      "step": 3350
    },
    {
      "epoch": 10.402476780185758,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001,
      "loss": 0.0288,
      "step": 3360
    },
    {
      "epoch": 10.43343653250774,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001,
      "loss": 0.0322,
      "step": 3370
    },
    {
      "epoch": 10.464396284829721,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001,
      "loss": 0.0249,
      "step": 3380
    },
    {
      "epoch": 10.495356037151703,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001,
      "loss": 0.0556,
      "step": 3390
    },
    {
      "epoch": 10.526315789473685,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001,
      "loss": 0.038,
      "step": 3400
    },
    {
      "epoch": 10.557275541795665,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001,
      "loss": 0.0356,
      "step": 3410
    },
    {
      "epoch": 10.588235294117647,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001,
      "loss": 0.0347,
      "step": 3420
    },
    {
      "epoch": 10.619195046439629,
      "grad_norm": 2.46875,
      "learning_rate": 0.0001,
      "loss": 0.0297,
      "step": 3430
    },
    {
      "epoch": 10.65015479876161,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001,
      "loss": 0.0537,
      "step": 3440
    },
    {
      "epoch": 10.681114551083592,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001,
      "loss": 0.04,
      "step": 3450
    },
    {
      "epoch": 10.712074303405572,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001,
      "loss": 0.031,
      "step": 3460
    },
    {
      "epoch": 10.743034055727554,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001,
      "loss": 0.0355,
      "step": 3470
    },
    {
      "epoch": 10.773993808049536,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001,
      "loss": 0.0308,
      "step": 3480
    },
    {
      "epoch": 10.804953560371517,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001,
      "loss": 0.051,
      "step": 3490
    },
    {
      "epoch": 10.8359133126935,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0001,
      "loss": 0.0372,
      "step": 3500
    },
    {
      "epoch": 10.86687306501548,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001,
      "loss": 0.0272,
      "step": 3510
    },
    {
      "epoch": 10.897832817337461,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001,
      "loss": 0.0314,
      "step": 3520
    },
    {
      "epoch": 10.928792569659443,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001,
      "loss": 0.0278,
      "step": 3530
    },
    {
      "epoch": 10.959752321981425,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001,
      "loss": 0.0579,
      "step": 3540
    },
    {
      "epoch": 10.990712074303406,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001,
      "loss": 0.0348,
      "step": 3550
    },
    {
      "epoch": 11.021671826625386,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001,
      "loss": 0.0299,
      "step": 3560
    },
    {
      "epoch": 11.052631578947368,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001,
      "loss": 0.0307,
      "step": 3570
    },
    {
      "epoch": 11.08359133126935,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001,
      "loss": 0.0273,
      "step": 3580
    },
    {
      "epoch": 11.114551083591332,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001,
      "loss": 0.0251,
      "step": 3590
    },
    {
      "epoch": 11.145510835913313,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0001,
      "loss": 0.0149,
      "step": 3600
    },
    {
      "epoch": 11.176470588235293,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001,
      "loss": 0.0327,
      "step": 3610
    },
    {
      "epoch": 11.207430340557275,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001,
      "loss": 0.0271,
      "step": 3620
    },
    {
      "epoch": 11.238390092879257,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001,
      "loss": 0.0276,
      "step": 3630
    },
    {
      "epoch": 11.269349845201239,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001,
      "loss": 0.0228,
      "step": 3640
    },
    {
      "epoch": 11.30030959752322,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0001,
      "loss": 0.0226,
      "step": 3650
    },
    {
      "epoch": 11.3312693498452,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001,
      "loss": 0.0352,
      "step": 3660
    },
    {
      "epoch": 11.362229102167182,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001,
      "loss": 0.0277,
      "step": 3670
    },
    {
      "epoch": 11.393188854489164,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001,
      "loss": 0.0296,
      "step": 3680
    },
    {
      "epoch": 11.424148606811146,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001,
      "loss": 0.0221,
      "step": 3690
    },
    {
      "epoch": 11.455108359133128,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001,
      "loss": 0.0241,
      "step": 3700
    },
    {
      "epoch": 11.486068111455108,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001,
      "loss": 0.0459,
      "step": 3710
    },
    {
      "epoch": 11.51702786377709,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001,
      "loss": 0.0358,
      "step": 3720
    },
    {
      "epoch": 11.547987616099071,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001,
      "loss": 0.0294,
      "step": 3730
    },
    {
      "epoch": 11.578947368421053,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0001,
      "loss": 0.0359,
      "step": 3740
    },
    {
      "epoch": 11.609907120743035,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001,
      "loss": 0.0281,
      "step": 3750
    },
    {
      "epoch": 11.640866873065015,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001,
      "loss": 0.0449,
      "step": 3760
    },
    {
      "epoch": 11.671826625386997,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001,
      "loss": 0.0386,
      "step": 3770
    },
    {
      "epoch": 11.702786377708978,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001,
      "loss": 0.042,
      "step": 3780
    },
    {
      "epoch": 11.73374613003096,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001,
      "loss": 0.0182,
      "step": 3790
    },
    {
      "epoch": 11.764705882352942,
      "grad_norm": 1.4375,
      "learning_rate": 0.0001,
      "loss": 0.0254,
      "step": 3800
    },
    {
      "epoch": 11.795665634674922,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001,
      "loss": 0.054,
      "step": 3810
    },
    {
      "epoch": 11.826625386996904,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001,
      "loss": 0.0378,
      "step": 3820
    },
    {
      "epoch": 11.857585139318886,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001,
      "loss": 0.0355,
      "step": 3830
    },
    {
      "epoch": 11.888544891640867,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001,
      "loss": 0.0251,
      "step": 3840
    },
    {
      "epoch": 11.91950464396285,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001,
      "loss": 0.027,
      "step": 3850
    },
    {
      "epoch": 11.95046439628483,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001,
      "loss": 0.0408,
      "step": 3860
    },
    {
      "epoch": 11.981424148606811,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001,
      "loss": 0.0295,
      "step": 3870
    },
    {
      "epoch": 12.012383900928793,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001,
      "loss": 0.0388,
      "step": 3880
    },
    {
      "epoch": 12.043343653250774,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001,
      "loss": 0.026,
      "step": 3890
    },
    {
      "epoch": 12.074303405572756,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001,
      "loss": 0.0246,
      "step": 3900
    },
    {
      "epoch": 12.105263157894736,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0001,
      "loss": 0.0277,
      "step": 3910
    },
    {
      "epoch": 12.136222910216718,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001,
      "loss": 0.0215,
      "step": 3920
    },
    {
      "epoch": 12.1671826625387,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001,
      "loss": 0.039,
      "step": 3930
    },
    {
      "epoch": 12.198142414860682,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001,
      "loss": 0.0321,
      "step": 3940
    },
    {
      "epoch": 12.229102167182663,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001,
      "loss": 0.0201,
      "step": 3950
    },
    {
      "epoch": 12.260061919504643,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001,
      "loss": 0.0222,
      "step": 3960
    },
    {
      "epoch": 12.291021671826625,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001,
      "loss": 0.0205,
      "step": 3970
    },
    {
      "epoch": 12.321981424148607,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001,
      "loss": 0.0223,
      "step": 3980
    },
    {
      "epoch": 12.352941176470589,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001,
      "loss": 0.0308,
      "step": 3990
    },
    {
      "epoch": 12.38390092879257,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001,
      "loss": 0.0247,
      "step": 4000
    },
    {
      "epoch": 12.41486068111455,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001,
      "loss": 0.0173,
      "step": 4010
    },
    {
      "epoch": 12.445820433436532,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001,
      "loss": 0.0118,
      "step": 4020
    },
    {
      "epoch": 12.476780185758514,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001,
      "loss": 0.0264,
      "step": 4030
    },
    {
      "epoch": 12.507739938080496,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001,
      "loss": 0.0324,
      "step": 4040
    },
    {
      "epoch": 12.538699690402478,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001,
      "loss": 0.0231,
      "step": 4050
    },
    {
      "epoch": 12.569659442724458,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001,
      "loss": 0.0173,
      "step": 4060
    },
    {
      "epoch": 12.60061919504644,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001,
      "loss": 0.0322,
      "step": 4070
    },
    {
      "epoch": 12.631578947368421,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001,
      "loss": 0.03,
      "step": 4080
    },
    {
      "epoch": 12.662538699690403,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001,
      "loss": 0.0317,
      "step": 4090
    },
    {
      "epoch": 12.693498452012385,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001,
      "loss": 0.0313,
      "step": 4100
    },
    {
      "epoch": 12.724458204334365,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001,
      "loss": 0.023,
      "step": 4110
    },
    {
      "epoch": 12.755417956656347,
      "grad_norm": 1.125,
      "learning_rate": 0.0001,
      "loss": 0.0227,
      "step": 4120
    },
    {
      "epoch": 12.786377708978328,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001,
      "loss": 0.0307,
      "step": 4130
    },
    {
      "epoch": 12.81733746130031,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001,
      "loss": 0.0354,
      "step": 4140
    },
    {
      "epoch": 12.848297213622292,
      "grad_norm": 1.125,
      "learning_rate": 0.0001,
      "loss": 0.0306,
      "step": 4150
    },
    {
      "epoch": 12.879256965944272,
      "grad_norm": 1.40625,
      "learning_rate": 0.0001,
      "loss": 0.0267,
      "step": 4160
    },
    {
      "epoch": 12.910216718266254,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001,
      "loss": 0.0213,
      "step": 4170
    },
    {
      "epoch": 12.941176470588236,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001,
      "loss": 0.0452,
      "step": 4180
    },
    {
      "epoch": 12.972136222910217,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001,
      "loss": 0.0388,
      "step": 4190
    },
    {
      "epoch": 13.003095975232197,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001,
      "loss": 0.0284,
      "step": 4200
    },
    {
      "epoch": 13.034055727554179,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001,
      "loss": 0.0331,
      "step": 4210
    },
    {
      "epoch": 13.06501547987616,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001,
      "loss": 0.0238,
      "step": 4220
    },
    {
      "epoch": 13.095975232198143,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001,
      "loss": 0.0177,
      "step": 4230
    },
    {
      "epoch": 13.126934984520124,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001,
      "loss": 0.0205,
      "step": 4240
    },
    {
      "epoch": 13.157894736842104,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001,
      "loss": 0.0352,
      "step": 4250
    },
    {
      "epoch": 13.188854489164086,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001,
      "loss": 0.0358,
      "step": 4260
    },
    {
      "epoch": 13.219814241486068,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001,
      "loss": 0.0285,
      "step": 4270
    },
    {
      "epoch": 13.25077399380805,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001,
      "loss": 0.0207,
      "step": 4280
    },
    {
      "epoch": 13.281733746130032,
      "grad_norm": 1.5625,
      "learning_rate": 0.0001,
      "loss": 0.0222,
      "step": 4290
    },
    {
      "epoch": 13.312693498452012,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001,
      "loss": 0.0234,
      "step": 4300
    },
    {
      "epoch": 13.343653250773993,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001,
      "loss": 0.0337,
      "step": 4310
    },
    {
      "epoch": 13.374613003095975,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0001,
      "loss": 0.0259,
      "step": 4320
    },
    {
      "epoch": 13.405572755417957,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001,
      "loss": 0.0281,
      "step": 4330
    },
    {
      "epoch": 13.436532507739939,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001,
      "loss": 0.0242,
      "step": 4340
    },
    {
      "epoch": 13.467492260061919,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001,
      "loss": 0.0456,
      "step": 4350
    },
    {
      "epoch": 13.4984520123839,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001,
      "loss": 0.0372,
      "step": 4360
    },
    {
      "epoch": 13.529411764705882,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001,
      "loss": 0.027,
      "step": 4370
    },
    {
      "epoch": 13.560371517027864,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001,
      "loss": 0.0266,
      "step": 4380
    },
    {
      "epoch": 13.591331269349846,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001,
      "loss": 0.0203,
      "step": 4390
    },
    {
      "epoch": 13.622291021671826,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001,
      "loss": 0.0204,
      "step": 4400
    },
    {
      "epoch": 13.653250773993808,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001,
      "loss": 0.0458,
      "step": 4410
    },
    {
      "epoch": 13.68421052631579,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0001,
      "loss": 0.0324,
      "step": 4420
    },
    {
      "epoch": 13.715170278637771,
      "grad_norm": 1.0,
      "learning_rate": 0.0001,
      "loss": 0.0279,
      "step": 4430
    },
    {
      "epoch": 13.746130030959753,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001,
      "loss": 0.0281,
      "step": 4440
    },
    {
      "epoch": 13.777089783281733,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001,
      "loss": 0.0189,
      "step": 4450
    },
    {
      "epoch": 13.808049535603715,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001,
      "loss": 0.0461,
      "step": 4460
    },
    {
      "epoch": 13.839009287925697,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0001,
      "loss": 0.0346,
      "step": 4470
    },
    {
      "epoch": 13.869969040247678,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001,
      "loss": 0.0338,
      "step": 4480
    },
    {
      "epoch": 13.90092879256966,
      "grad_norm": 1.4375,
      "learning_rate": 0.0001,
      "loss": 0.0323,
      "step": 4490
    },
    {
      "epoch": 13.93188854489164,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001,
      "loss": 0.0363,
      "step": 4500
    },
    {
      "epoch": 13.962848297213622,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001,
      "loss": 0.0411,
      "step": 4510
    },
    {
      "epoch": 13.993808049535604,
      "grad_norm": 2.09375,
      "learning_rate": 0.0001,
      "loss": 0.0235,
      "step": 4520
    },
    {
      "epoch": 14.024767801857585,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001,
      "loss": 0.028,
      "step": 4530
    },
    {
      "epoch": 14.055727554179567,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001,
      "loss": 0.0288,
      "step": 4540
    },
    {
      "epoch": 14.086687306501547,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001,
      "loss": 0.0286,
      "step": 4550
    },
    {
      "epoch": 14.117647058823529,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001,
      "loss": 0.0253,
      "step": 4560
    },
    {
      "epoch": 14.14860681114551,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001,
      "loss": 0.0244,
      "step": 4570
    },
    {
      "epoch": 14.179566563467493,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001,
      "loss": 0.0405,
      "step": 4580
    },
    {
      "epoch": 14.210526315789474,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001,
      "loss": 0.0255,
      "step": 4590
    },
    {
      "epoch": 14.241486068111454,
      "grad_norm": 1.515625,
      "learning_rate": 0.0001,
      "loss": 0.0242,
      "step": 4600
    },
    {
      "epoch": 14.272445820433436,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001,
      "loss": 0.0304,
      "step": 4610
    },
    {
      "epoch": 14.303405572755418,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001,
      "loss": 0.0194,
      "step": 4620
    },
    {
      "epoch": 14.3343653250774,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001,
      "loss": 0.033,
      "step": 4630
    },
    {
      "epoch": 14.365325077399381,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001,
      "loss": 0.032,
      "step": 4640
    },
    {
      "epoch": 14.396284829721361,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001,
      "loss": 0.0223,
      "step": 4650
    },
    {
      "epoch": 14.427244582043343,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001,
      "loss": 0.0294,
      "step": 4660
    },
    {
      "epoch": 14.458204334365325,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001,
      "loss": 0.0339,
      "step": 4670
    },
    {
      "epoch": 14.489164086687307,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001,
      "loss": 0.0364,
      "step": 4680
    },
    {
      "epoch": 14.520123839009289,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001,
      "loss": 0.0268,
      "step": 4690
    },
    {
      "epoch": 14.551083591331269,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001,
      "loss": 0.0193,
      "step": 4700
    },
    {
      "epoch": 14.58204334365325,
      "grad_norm": 0.15625,
      "learning_rate": 0.0001,
      "loss": 0.0213,
      "step": 4710
    },
    {
      "epoch": 14.613003095975232,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001,
      "loss": 0.0141,
      "step": 4720
    },
    {
      "epoch": 14.643962848297214,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001,
      "loss": 0.037,
      "step": 4730
    },
    {
      "epoch": 14.674922600619196,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001,
      "loss": 0.033,
      "step": 4740
    },
    {
      "epoch": 14.705882352941176,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001,
      "loss": 0.0237,
      "step": 4750
    },
    {
      "epoch": 14.736842105263158,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001,
      "loss": 0.0277,
      "step": 4760
    },
    {
      "epoch": 14.76780185758514,
      "grad_norm": 1.59375,
      "learning_rate": 0.0001,
      "loss": 0.024,
      "step": 4770
    },
    {
      "epoch": 14.798761609907121,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001,
      "loss": 0.0478,
      "step": 4780
    },
    {
      "epoch": 14.829721362229103,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001,
      "loss": 0.0321,
      "step": 4790
    },
    {
      "epoch": 14.860681114551083,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001,
      "loss": 0.0309,
      "step": 4800
    },
    {
      "epoch": 14.891640866873065,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001,
      "loss": 0.026,
      "step": 4810
    },
    {
      "epoch": 14.922600619195046,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001,
      "loss": 0.0256,
      "step": 4820
    },
    {
      "epoch": 14.953560371517028,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001,
      "loss": 0.0351,
      "step": 4830
    },
    {
      "epoch": 14.98452012383901,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001,
      "loss": 0.0329,
      "step": 4840
    },
    {
      "epoch": 15.01547987616099,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001,
      "loss": 0.0286,
      "step": 4850
    },
    {
      "epoch": 15.046439628482972,
      "grad_norm": 1.34375,
      "learning_rate": 0.0001,
      "loss": 0.0301,
      "step": 4860
    },
    {
      "epoch": 15.077399380804954,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001,
      "loss": 0.0247,
      "step": 4870
    },
    {
      "epoch": 15.108359133126935,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001,
      "loss": 0.0237,
      "step": 4880
    },
    {
      "epoch": 15.139318885448917,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001,
      "loss": 0.0248,
      "step": 4890
    },
    {
      "epoch": 15.170278637770897,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0001,
      "loss": 0.0246,
      "step": 4900
    },
    {
      "epoch": 15.201238390092879,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001,
      "loss": 0.0234,
      "step": 4910
    },
    {
      "epoch": 15.23219814241486,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0001,
      "loss": 0.0181,
      "step": 4920
    },
    {
      "epoch": 15.263157894736842,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001,
      "loss": 0.0288,
      "step": 4930
    },
    {
      "epoch": 15.294117647058824,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001,
      "loss": 0.0234,
      "step": 4940
    },
    {
      "epoch": 15.325077399380804,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001,
      "loss": 0.0323,
      "step": 4950
    },
    {
      "epoch": 15.356037151702786,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001,
      "loss": 0.0296,
      "step": 4960
    },
    {
      "epoch": 15.386996904024768,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001,
      "loss": 0.0267,
      "step": 4970
    },
    {
      "epoch": 15.41795665634675,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001,
      "loss": 0.0256,
      "step": 4980
    },
    {
      "epoch": 15.448916408668731,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001,
      "loss": 0.0307,
      "step": 4990
    },
    {
      "epoch": 15.479876160990711,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001,
      "loss": 0.0381,
      "step": 5000
    },
    {
      "epoch": 15.510835913312693,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0001,
      "loss": 0.022,
      "step": 5010
    },
    {
      "epoch": 15.541795665634675,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001,
      "loss": 0.0214,
      "step": 5020
    },
    {
      "epoch": 15.572755417956657,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001,
      "loss": 0.023,
      "step": 5030
    },
    {
      "epoch": 15.603715170278639,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0001,
      "loss": 0.0278,
      "step": 5040
    },
    {
      "epoch": 15.634674922600619,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001,
      "loss": 0.0283,
      "step": 5050
    },
    {
      "epoch": 15.6656346749226,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001,
      "loss": 0.0256,
      "step": 5060
    },
    {
      "epoch": 15.696594427244582,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001,
      "loss": 0.031,
      "step": 5070
    },
    {
      "epoch": 15.727554179566564,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001,
      "loss": 0.0159,
      "step": 5080
    },
    {
      "epoch": 15.758513931888546,
      "grad_norm": 1.359375,
      "learning_rate": 0.0001,
      "loss": 0.0241,
      "step": 5090
    },
    {
      "epoch": 15.789473684210526,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001,
      "loss": 0.0371,
      "step": 5100
    },
    {
      "epoch": 15.820433436532507,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001,
      "loss": 0.0257,
      "step": 5110
    },
    {
      "epoch": 15.85139318885449,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001,
      "loss": 0.0325,
      "step": 5120
    },
    {
      "epoch": 15.882352941176471,
      "grad_norm": 1.25,
      "learning_rate": 0.0001,
      "loss": 0.0235,
      "step": 5130
    },
    {
      "epoch": 15.913312693498453,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001,
      "loss": 0.0241,
      "step": 5140
    },
    {
      "epoch": 15.944272445820433,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001,
      "loss": 0.0271,
      "step": 5150
    },
    {
      "epoch": 15.975232198142415,
      "grad_norm": 0.875,
      "learning_rate": 0.0001,
      "loss": 0.0311,
      "step": 5160
    },
    {
      "epoch": 16.006191950464395,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001,
      "loss": 0.0263,
      "step": 5170
    },
    {
      "epoch": 16.037151702786378,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001,
      "loss": 0.0198,
      "step": 5180
    },
    {
      "epoch": 16.068111455108358,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001,
      "loss": 0.0188,
      "step": 5190
    },
    {
      "epoch": 16.09907120743034,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001,
      "loss": 0.0174,
      "step": 5200
    },
    {
      "epoch": 16.13003095975232,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001,
      "loss": 0.022,
      "step": 5210
    },
    {
      "epoch": 16.1609907120743,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001,
      "loss": 0.0216,
      "step": 5220
    },
    {
      "epoch": 16.191950464396285,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001,
      "loss": 0.0271,
      "step": 5230
    },
    {
      "epoch": 16.222910216718265,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001,
      "loss": 0.0227,
      "step": 5240
    },
    {
      "epoch": 16.25386996904025,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0001,
      "loss": 0.0143,
      "step": 5250
    },
    {
      "epoch": 16.28482972136223,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001,
      "loss": 0.0261,
      "step": 5260
    },
    {
      "epoch": 16.31578947368421,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001,
      "loss": 0.0168,
      "step": 5270
    },
    {
      "epoch": 16.346749226006192,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001,
      "loss": 0.0272,
      "step": 5280
    },
    {
      "epoch": 16.377708978328172,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001,
      "loss": 0.0272,
      "step": 5290
    },
    {
      "epoch": 16.408668730650156,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001,
      "loss": 0.0264,
      "step": 5300
    },
    {
      "epoch": 16.439628482972136,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0001,
      "loss": 0.022,
      "step": 5310
    },
    {
      "epoch": 16.470588235294116,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001,
      "loss": 0.0292,
      "step": 5320
    },
    {
      "epoch": 16.5015479876161,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001,
      "loss": 0.0317,
      "step": 5330
    },
    {
      "epoch": 16.53250773993808,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001,
      "loss": 0.025,
      "step": 5340
    },
    {
      "epoch": 16.563467492260063,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001,
      "loss": 0.0223,
      "step": 5350
    },
    {
      "epoch": 16.594427244582043,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001,
      "loss": 0.0171,
      "step": 5360
    },
    {
      "epoch": 16.625386996904023,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001,
      "loss": 0.028,
      "step": 5370
    },
    {
      "epoch": 16.656346749226007,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001,
      "loss": 0.0324,
      "step": 5380
    },
    {
      "epoch": 16.687306501547987,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001,
      "loss": 0.0209,
      "step": 5390
    },
    {
      "epoch": 16.71826625386997,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001,
      "loss": 0.0192,
      "step": 5400
    },
    {
      "epoch": 16.74922600619195,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001,
      "loss": 0.0289,
      "step": 5410
    },
    {
      "epoch": 16.78018575851393,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001,
      "loss": 0.0263,
      "step": 5420
    },
    {
      "epoch": 16.811145510835914,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001,
      "loss": 0.0348,
      "step": 5430
    },
    {
      "epoch": 16.842105263157894,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001,
      "loss": 0.023,
      "step": 5440
    },
    {
      "epoch": 16.873065015479877,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001,
      "loss": 0.0192,
      "step": 5450
    },
    {
      "epoch": 16.904024767801857,
      "grad_norm": 1.9921875,
      "learning_rate": 0.0001,
      "loss": 0.0272,
      "step": 5460
    },
    {
      "epoch": 16.934984520123837,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001,
      "loss": 0.0254,
      "step": 5470
    },
    {
      "epoch": 16.96594427244582,
      "grad_norm": 1.4375,
      "learning_rate": 0.0001,
      "loss": 0.0399,
      "step": 5480
    },
    {
      "epoch": 16.9969040247678,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0001,
      "loss": 0.0345,
      "step": 5490
    },
    {
      "epoch": 17.027863777089784,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001,
      "loss": 0.0233,
      "step": 5500
    },
    {
      "epoch": 17.058823529411764,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001,
      "loss": 0.0205,
      "step": 5510
    },
    {
      "epoch": 17.089783281733745,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0001,
      "loss": 0.0282,
      "step": 5520
    },
    {
      "epoch": 17.120743034055728,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0001,
      "loss": 0.0279,
      "step": 5530
    },
    {
      "epoch": 17.151702786377708,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001,
      "loss": 0.0274,
      "step": 5540
    },
    {
      "epoch": 17.18266253869969,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001,
      "loss": 0.0289,
      "step": 5550
    },
    {
      "epoch": 17.21362229102167,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001,
      "loss": 0.0235,
      "step": 5560
    },
    {
      "epoch": 17.24458204334365,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001,
      "loss": 0.0236,
      "step": 5570
    },
    {
      "epoch": 17.275541795665635,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0001,
      "loss": 0.0229,
      "step": 5580
    },
    {
      "epoch": 17.306501547987615,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0001,
      "loss": 0.027,
      "step": 5590
    },
    {
      "epoch": 17.3374613003096,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001,
      "loss": 0.0254,
      "step": 5600
    },
    {
      "epoch": 17.36842105263158,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001,
      "loss": 0.0232,
      "step": 5610
    },
    {
      "epoch": 17.39938080495356,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001,
      "loss": 0.0189,
      "step": 5620
    },
    {
      "epoch": 17.430340557275542,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001,
      "loss": 0.0262,
      "step": 5630
    },
    {
      "epoch": 17.461300309597522,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001,
      "loss": 0.0114,
      "step": 5640
    },
    {
      "epoch": 17.492260061919506,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001,
      "loss": 0.0345,
      "step": 5650
    },
    {
      "epoch": 17.523219814241486,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001,
      "loss": 0.0289,
      "step": 5660
    },
    {
      "epoch": 17.554179566563466,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001,
      "loss": 0.0186,
      "step": 5670
    },
    {
      "epoch": 17.58513931888545,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001,
      "loss": 0.0317,
      "step": 5680
    },
    {
      "epoch": 17.61609907120743,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0001,
      "loss": 0.0181,
      "step": 5690
    },
    {
      "epoch": 17.647058823529413,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001,
      "loss": 0.0364,
      "step": 5700
    },
    {
      "epoch": 17.678018575851393,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001,
      "loss": 0.0288,
      "step": 5710
    },
    {
      "epoch": 17.708978328173373,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001,
      "loss": 0.0215,
      "step": 5720
    },
    {
      "epoch": 17.739938080495357,
      "grad_norm": 2.078125,
      "learning_rate": 0.0001,
      "loss": 0.0232,
      "step": 5730
    },
    {
      "epoch": 17.770897832817337,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0001,
      "loss": 0.0192,
      "step": 5740
    },
    {
      "epoch": 17.80185758513932,
      "grad_norm": 0.875,
      "learning_rate": 0.0001,
      "loss": 0.0388,
      "step": 5750
    },
    {
      "epoch": 17.8328173374613,
      "grad_norm": 1.125,
      "learning_rate": 0.0001,
      "loss": 0.0225,
      "step": 5760
    },
    {
      "epoch": 17.86377708978328,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001,
      "loss": 0.0256,
      "step": 5770
    },
    {
      "epoch": 17.894736842105264,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001,
      "loss": 0.0192,
      "step": 5780
    },
    {
      "epoch": 17.925696594427244,
      "grad_norm": 1.375,
      "learning_rate": 0.0001,
      "loss": 0.0354,
      "step": 5790
    },
    {
      "epoch": 17.956656346749227,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001,
      "loss": 0.0361,
      "step": 5800
    },
    {
      "epoch": 17.987616099071207,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001,
      "loss": 0.0357,
      "step": 5810
    },
    {
      "epoch": 18.018575851393187,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001,
      "loss": 0.0196,
      "step": 5820
    },
    {
      "epoch": 18.04953560371517,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001,
      "loss": 0.0231,
      "step": 5830
    },
    {
      "epoch": 18.08049535603715,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001,
      "loss": 0.0255,
      "step": 5840
    },
    {
      "epoch": 18.111455108359134,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001,
      "loss": 0.0205,
      "step": 5850
    },
    {
      "epoch": 18.142414860681114,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0001,
      "loss": 0.0204,
      "step": 5860
    },
    {
      "epoch": 18.173374613003094,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001,
      "loss": 0.0271,
      "step": 5870
    },
    {
      "epoch": 18.204334365325078,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001,
      "loss": 0.0197,
      "step": 5880
    },
    {
      "epoch": 18.235294117647058,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001,
      "loss": 0.0242,
      "step": 5890
    },
    {
      "epoch": 18.26625386996904,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001,
      "loss": 0.0176,
      "step": 5900
    },
    {
      "epoch": 18.29721362229102,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001,
      "loss": 0.0287,
      "step": 5910
    },
    {
      "epoch": 18.328173374613,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001,
      "loss": 0.0243,
      "step": 5920
    },
    {
      "epoch": 18.359133126934985,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001,
      "loss": 0.0255,
      "step": 5930
    },
    {
      "epoch": 18.390092879256965,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001,
      "loss": 0.0256,
      "step": 5940
    },
    {
      "epoch": 18.42105263157895,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001,
      "loss": 0.0248,
      "step": 5950
    },
    {
      "epoch": 18.45201238390093,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0001,
      "loss": 0.0321,
      "step": 5960
    },
    {
      "epoch": 18.48297213622291,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001,
      "loss": 0.0274,
      "step": 5970
    },
    {
      "epoch": 18.513931888544892,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001,
      "loss": 0.0307,
      "step": 5980
    },
    {
      "epoch": 18.544891640866872,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001,
      "loss": 0.025,
      "step": 5990
    },
    {
      "epoch": 18.575851393188856,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0001,
      "loss": 0.0172,
      "step": 6000
    },
    {
      "epoch": 18.606811145510836,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001,
      "loss": 0.0103,
      "step": 6010
    },
    {
      "epoch": 18.637770897832816,
      "grad_norm": 0.875,
      "learning_rate": 0.0001,
      "loss": 0.0398,
      "step": 6020
    },
    {
      "epoch": 18.6687306501548,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001,
      "loss": 0.026,
      "step": 6030
    },
    {
      "epoch": 18.69969040247678,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001,
      "loss": 0.0227,
      "step": 6040
    },
    {
      "epoch": 18.730650154798763,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0001,
      "loss": 0.0217,
      "step": 6050
    },
    {
      "epoch": 18.761609907120743,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001,
      "loss": 0.0141,
      "step": 6060
    },
    {
      "epoch": 18.792569659442723,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001,
      "loss": 0.0314,
      "step": 6070
    },
    {
      "epoch": 18.823529411764707,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0001,
      "loss": 0.0227,
      "step": 6080
    },
    {
      "epoch": 18.854489164086687,
      "grad_norm": 0.875,
      "learning_rate": 0.0001,
      "loss": 0.0313,
      "step": 6090
    },
    {
      "epoch": 18.88544891640867,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001,
      "loss": 0.0243,
      "step": 6100
    },
    {
      "epoch": 18.91640866873065,
      "grad_norm": 0.06982421875,
      "learning_rate": 0.0001,
      "loss": 0.0211,
      "step": 6110
    },
    {
      "epoch": 18.94736842105263,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001,
      "loss": 0.0342,
      "step": 6120
    },
    {
      "epoch": 18.978328173374614,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001,
      "loss": 0.0287,
      "step": 6130
    },
    {
      "epoch": 19.009287925696594,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001,
      "loss": 0.0293,
      "step": 6140
    },
    {
      "epoch": 19.040247678018577,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001,
      "loss": 0.0207,
      "step": 6150
    },
    {
      "epoch": 19.071207430340557,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001,
      "loss": 0.0137,
      "step": 6160
    },
    {
      "epoch": 19.102167182662537,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001,
      "loss": 0.0286,
      "step": 6170
    },
    {
      "epoch": 19.13312693498452,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0001,
      "loss": 0.0153,
      "step": 6180
    },
    {
      "epoch": 19.1640866873065,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001,
      "loss": 0.0256,
      "step": 6190
    },
    {
      "epoch": 19.195046439628484,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001,
      "loss": 0.0273,
      "step": 6200
    },
    {
      "epoch": 19.226006191950464,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001,
      "loss": 0.0226,
      "step": 6210
    },
    {
      "epoch": 19.256965944272444,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001,
      "loss": 0.014,
      "step": 6220
    },
    {
      "epoch": 19.287925696594428,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001,
      "loss": 0.026,
      "step": 6230
    },
    {
      "epoch": 19.318885448916408,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001,
      "loss": 0.019,
      "step": 6240
    },
    {
      "epoch": 19.34984520123839,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001,
      "loss": 0.0279,
      "step": 6250
    },
    {
      "epoch": 19.38080495356037,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001,
      "loss": 0.0209,
      "step": 6260
    },
    {
      "epoch": 19.41176470588235,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001,
      "loss": 0.0243,
      "step": 6270
    },
    {
      "epoch": 19.442724458204335,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001,
      "loss": 0.0199,
      "step": 6280
    },
    {
      "epoch": 19.473684210526315,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001,
      "loss": 0.0384,
      "step": 6290
    },
    {
      "epoch": 19.5046439628483,
      "grad_norm": 1.0,
      "learning_rate": 0.0001,
      "loss": 0.026,
      "step": 6300
    },
    {
      "epoch": 19.53560371517028,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001,
      "loss": 0.0199,
      "step": 6310
    },
    {
      "epoch": 19.56656346749226,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0001,
      "loss": 0.0317,
      "step": 6320
    },
    {
      "epoch": 19.597523219814242,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001,
      "loss": 0.0292,
      "step": 6330
    },
    {
      "epoch": 19.628482972136222,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001,
      "loss": 0.0298,
      "step": 6340
    },
    {
      "epoch": 19.659442724458206,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001,
      "loss": 0.0335,
      "step": 6350
    },
    {
      "epoch": 19.690402476780186,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0001,
      "loss": 0.0287,
      "step": 6360
    },
    {
      "epoch": 19.721362229102166,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001,
      "loss": 0.0213,
      "step": 6370
    },
    {
      "epoch": 19.75232198142415,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001,
      "loss": 0.0155,
      "step": 6380
    },
    {
      "epoch": 19.78328173374613,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001,
      "loss": 0.0211,
      "step": 6390
    },
    {
      "epoch": 19.814241486068113,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001,
      "loss": 0.0323,
      "step": 6400
    },
    {
      "epoch": 19.845201238390093,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001,
      "loss": 0.0206,
      "step": 6410
    },
    {
      "epoch": 19.876160990712073,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001,
      "loss": 0.0151,
      "step": 6420
    },
    {
      "epoch": 19.907120743034056,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001,
      "loss": 0.0178,
      "step": 6430
    },
    {
      "epoch": 19.938080495356036,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001,
      "loss": 0.0226,
      "step": 6440
    },
    {
      "epoch": 19.96904024767802,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001,
      "loss": 0.0291,
      "step": 6450
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.06396484375,
      "learning_rate": 0.0001,
      "loss": 0.0164,
      "step": 6460
    },
    {
      "epoch": 20.03095975232198,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001,
      "loss": 0.02,
      "step": 6470
    },
    {
      "epoch": 20.061919504643964,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001,
      "loss": 0.0126,
      "step": 6480
    },
    {
      "epoch": 20.092879256965944,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001,
      "loss": 0.0153,
      "step": 6490
    },
    {
      "epoch": 20.123839009287927,
      "grad_norm": 1.671875,
      "learning_rate": 0.0001,
      "loss": 0.0186,
      "step": 6500
    },
    {
      "epoch": 20.154798761609907,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001,
      "loss": 0.0212,
      "step": 6510
    },
    {
      "epoch": 20.185758513931887,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001,
      "loss": 0.0213,
      "step": 6520
    },
    {
      "epoch": 20.21671826625387,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001,
      "loss": 0.0243,
      "step": 6530
    },
    {
      "epoch": 20.24767801857585,
      "grad_norm": 1.609375,
      "learning_rate": 0.0001,
      "loss": 0.0153,
      "step": 6540
    },
    {
      "epoch": 20.278637770897834,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001,
      "loss": 0.0196,
      "step": 6550
    },
    {
      "epoch": 20.309597523219814,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001,
      "loss": 0.0068,
      "step": 6560
    },
    {
      "epoch": 20.340557275541794,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001,
      "loss": 0.0212,
      "step": 6570
    },
    {
      "epoch": 20.371517027863778,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001,
      "loss": 0.013,
      "step": 6580
    },
    {
      "epoch": 20.402476780185758,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001,
      "loss": 0.016,
      "step": 6590
    },
    {
      "epoch": 20.43343653250774,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001,
      "loss": 0.0228,
      "step": 6600
    },
    {
      "epoch": 20.46439628482972,
      "grad_norm": 0.3125,
      "learning_rate": 0.0001,
      "loss": 0.0068,
      "step": 6610
    },
    {
      "epoch": 20.4953560371517,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001,
      "loss": 0.0254,
      "step": 6620
    },
    {
      "epoch": 20.526315789473685,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001,
      "loss": 0.0173,
      "step": 6630
    },
    {
      "epoch": 20.557275541795665,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001,
      "loss": 0.0164,
      "step": 6640
    },
    {
      "epoch": 20.58823529411765,
      "grad_norm": 1.34375,
      "learning_rate": 0.0001,
      "loss": 0.0245,
      "step": 6650
    },
    {
      "epoch": 20.61919504643963,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0001,
      "loss": 0.0175,
      "step": 6660
    },
    {
      "epoch": 20.65015479876161,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001,
      "loss": 0.0274,
      "step": 6670
    },
    {
      "epoch": 20.681114551083592,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001,
      "loss": 0.0218,
      "step": 6680
    },
    {
      "epoch": 20.712074303405572,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001,
      "loss": 0.0172,
      "step": 6690
    },
    {
      "epoch": 20.743034055727556,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001,
      "loss": 0.0156,
      "step": 6700
    }
  ],
  "logging_steps": 10,
  "max_steps": 64600,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 200,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.077655366024663e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}