{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5,
  "eval_steps": 250,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001,
      "grad_norm": 7.1875,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 1.8199,
      "loss/crossentropy": 3.5243165493011475,
      "loss/fcd": 1.1015625,
      "loss/logits": 0.7183534502983093,
      "step": 1
    },
    {
      "epoch": 0.002,
      "grad_norm": 7.03125,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.7544,
      "loss/crossentropy": 3.38218891620636,
      "loss/fcd": 1.06640625,
      "loss/logits": 0.6879583299160004,
      "step": 2
    },
    {
      "epoch": 0.003,
      "grad_norm": 6.90625,
      "learning_rate": 3e-06,
      "loss": 1.7466,
      "loss/crossentropy": 3.4039193391799927,
      "loss/fcd": 1.06640625,
      "loss/logits": 0.6802149713039398,
      "step": 3
    },
    {
      "epoch": 0.004,
      "grad_norm": 8.0,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.7976,
      "loss/crossentropy": 3.5470337867736816,
      "loss/fcd": 1.08984375,
      "loss/logits": 0.7077842652797699,
      "step": 4
    },
    {
      "epoch": 0.005,
      "grad_norm": 7.0,
      "learning_rate": 5e-06,
      "loss": 1.7934,
      "loss/crossentropy": 3.393998146057129,
      "loss/fcd": 1.08203125,
      "loss/logits": 0.7113845348358154,
      "step": 5
    },
    {
      "epoch": 0.006,
      "grad_norm": 6.25,
      "learning_rate": 6e-06,
      "loss": 1.7157,
      "loss/crossentropy": 3.348253607749939,
      "loss/fcd": 1.03125,
      "loss/logits": 0.684424489736557,
      "step": 6
    },
    {
      "epoch": 0.007,
      "grad_norm": 6.90625,
      "learning_rate": 7.000000000000001e-06,
      "loss": 1.7533,
      "loss/crossentropy": 3.370575785636902,
      "loss/fcd": 1.06640625,
      "loss/logits": 0.686925858259201,
      "step": 7
    },
    {
      "epoch": 0.008,
      "grad_norm": 5.65625,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.7326,
      "loss/crossentropy": 3.325868844985962,
      "loss/fcd": 1.06640625,
      "loss/logits": 0.6662170886993408,
      "step": 8
    },
    {
      "epoch": 0.009,
      "grad_norm": 6.0,
      "learning_rate": 9e-06,
      "loss": 1.7279,
      "loss/crossentropy": 3.2712498903274536,
      "loss/fcd": 1.05078125,
      "loss/logits": 0.6771342754364014,
      "step": 9
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.15625,
      "learning_rate": 1e-05,
      "loss": 1.6636,
      "loss/crossentropy": 3.31475031375885,
      "loss/fcd": 1.037109375,
      "loss/logits": 0.6264936327934265,
      "step": 10
    },
    {
      "epoch": 0.011,
      "grad_norm": 5.1875,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 1.6084,
      "loss/crossentropy": 3.252698540687561,
      "loss/fcd": 0.99609375,
      "loss/logits": 0.612336277961731,
      "step": 11
    },
    {
      "epoch": 0.012,
      "grad_norm": 4.53125,
      "learning_rate": 1.2e-05,
      "loss": 1.5937,
      "loss/crossentropy": 3.2348395586013794,
      "loss/fcd": 0.994140625,
      "loss/logits": 0.5995941162109375,
      "step": 12
    },
    {
      "epoch": 0.013,
      "grad_norm": 3.75,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 1.5844,
      "loss/crossentropy": 3.2545889616012573,
      "loss/fcd": 0.990234375,
      "loss/logits": 0.5941979587078094,
      "step": 13
    },
    {
      "epoch": 0.014,
      "grad_norm": 3.09375,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 1.5826,
      "loss/crossentropy": 3.3384639024734497,
      "loss/fcd": 0.970703125,
      "loss/logits": 0.6118558943271637,
      "step": 14
    },
    {
      "epoch": 0.015,
      "grad_norm": 2.65625,
      "learning_rate": 1.5e-05,
      "loss": 1.4314,
      "loss/crossentropy": 3.279728889465332,
      "loss/fcd": 0.90234375,
      "loss/logits": 0.529065728187561,
      "step": 15
    },
    {
      "epoch": 0.016,
      "grad_norm": 2.28125,
      "grad_norm_var": 3.1285441080729166,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.4926,
      "loss/crossentropy": 3.225893259048462,
      "loss/fcd": 0.919921875,
      "loss/logits": 0.572645902633667,
      "step": 16
    },
    {
      "epoch": 0.017,
      "grad_norm": 2.078125,
      "grad_norm_var": 3.593244425455729,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 1.4306,
      "loss/crossentropy": 3.2493473291397095,
      "loss/fcd": 0.896484375,
      "loss/logits": 0.534164160490036,
      "step": 17
    },
    {
      "epoch": 0.018,
      "grad_norm": 1.75,
      "grad_norm_var": 4.015469360351562,
      "learning_rate": 1.8e-05,
      "loss": 1.366,
      "loss/crossentropy": 3.346846103668213,
      "loss/fcd": 0.861328125,
      "loss/logits": 0.504672035574913,
      "step": 18
    },
    {
      "epoch": 0.019,
      "grad_norm": 1.671875,
      "grad_norm_var": 4.275484212239584,
      "learning_rate": 1.9e-05,
      "loss": 1.3969,
      "loss/crossentropy": 3.273995280265808,
      "loss/fcd": 0.865234375,
      "loss/logits": 0.5316321551799774,
      "step": 19
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.609375,
      "grad_norm_var": 3.8440338134765626,
      "learning_rate": 2e-05,
      "loss": 1.2777,
      "loss/crossentropy": 3.2759294509887695,
      "loss/fcd": 0.802734375,
      "loss/logits": 0.47498229146003723,
      "step": 20
    },
    {
      "epoch": 0.021,
      "grad_norm": 1.4765625,
      "grad_norm_var": 3.6140703837076824,
      "learning_rate": 2.1e-05,
      "loss": 1.3094,
      "loss/crossentropy": 3.299839496612549,
      "loss/fcd": 0.81640625,
      "loss/logits": 0.49299056828022003,
      "step": 21
    },
    {
      "epoch": 0.022,
      "grad_norm": 1.5546875,
      "grad_norm_var": 3.4289784749348957,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 1.3074,
      "loss/crossentropy": 3.3400200605392456,
      "loss/fcd": 0.826171875,
      "loss/logits": 0.4812265634536743,
      "step": 22
    },
    {
      "epoch": 0.023,
      "grad_norm": 1.3046875,
      "grad_norm_var": 2.816125233968099,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 1.2273,
      "loss/crossentropy": 3.1525696516036987,
      "loss/fcd": 0.78125,
      "loss/logits": 0.4460318386554718,
      "step": 23
    },
    {
      "epoch": 0.024,
      "grad_norm": 1.4140625,
      "grad_norm_var": 2.500584920247396,
      "learning_rate": 2.4e-05,
      "loss": 1.2879,
      "loss/crossentropy": 3.345754623413086,
      "loss/fcd": 0.810546875,
      "loss/logits": 0.4773627370595932,
      "step": 24
    },
    {
      "epoch": 0.025,
      "grad_norm": 1.203125,
      "grad_norm_var": 1.9206502278645834,
      "learning_rate": 2.5e-05,
      "loss": 1.1679,
      "loss/crossentropy": 3.041555643081665,
      "loss/fcd": 0.740234375,
      "loss/logits": 0.42767438292503357,
      "step": 25
    },
    {
      "epoch": 0.026,
      "grad_norm": 1.1328125,
      "grad_norm_var": 1.5315345764160155,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 1.203,
      "loss/crossentropy": 3.267328143119812,
      "loss/fcd": 0.763671875,
      "loss/logits": 0.43937453627586365,
      "step": 26
    },
    {
      "epoch": 0.027,
      "grad_norm": 1.03125,
      "grad_norm_var": 1.007403310139974,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 1.1558,
      "loss/crossentropy": 3.1087806224823,
      "loss/fcd": 0.7421875,
      "loss/logits": 0.41364721953868866,
      "step": 27
    },
    {
      "epoch": 0.028,
      "grad_norm": 1.0,
      "grad_norm_var": 0.610827382405599,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 1.1024,
      "loss/crossentropy": 3.087596893310547,
      "loss/fcd": 0.71484375,
      "loss/logits": 0.3875562846660614,
      "step": 28
    },
    {
      "epoch": 0.029,
      "grad_norm": 0.9765625,
      "grad_norm_var": 0.3752838134765625,
      "learning_rate": 2.9e-05,
      "loss": 1.0592,
      "loss/crossentropy": 2.9776238203048706,
      "loss/fcd": 0.689453125,
      "loss/logits": 0.36973273754119873,
      "step": 29
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.03125,
      "grad_norm_var": 0.2412750244140625,
      "learning_rate": 3e-05,
      "loss": 1.1152,
      "loss/crossentropy": 3.046830892562866,
      "loss/fcd": 0.728515625,
      "loss/logits": 0.38670212030410767,
      "step": 30
    },
    {
      "epoch": 0.031,
      "grad_norm": 1.046875,
      "grad_norm_var": 0.1573486328125,
      "learning_rate": 3.1e-05,
      "loss": 1.1022,
      "loss/crossentropy": 3.1873209476470947,
      "loss/fcd": 0.705078125,
      "loss/logits": 0.39707766473293304,
      "step": 31
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.8984375,
      "grad_norm_var": 0.11625137329101562,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 1.0413,
      "loss/crossentropy": 3.0968486070632935,
      "loss/fcd": 0.685546875,
      "loss/logits": 0.3557591587305069,
      "step": 32
    },
    {
      "epoch": 0.033,
      "grad_norm": 0.8828125,
      "grad_norm_var": 0.0853179931640625,
      "learning_rate": 3.3e-05,
      "loss": 1.0181,
      "loss/crossentropy": 3.0179919004440308,
      "loss/fcd": 0.671875,
      "loss/logits": 0.34625276923179626,
      "step": 33
    },
    {
      "epoch": 0.034,
      "grad_norm": 0.82421875,
      "grad_norm_var": 0.0770456314086914,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 1.023,
      "loss/crossentropy": 2.9905418157577515,
      "loss/fcd": 0.669921875,
      "loss/logits": 0.3531098961830139,
      "step": 34
    },
    {
      "epoch": 0.035,
      "grad_norm": 0.8359375,
      "grad_norm_var": 0.06714064280192057,
      "learning_rate": 3.5e-05,
      "loss": 1.0004,
      "loss/crossentropy": 2.974391222000122,
      "loss/fcd": 0.666015625,
      "loss/logits": 0.3343943804502487,
      "step": 35
    },
    {
      "epoch": 0.036,
      "grad_norm": 0.90234375,
      "grad_norm_var": 0.05403340657552083,
      "learning_rate": 3.6e-05,
      "loss": 1.0024,
      "loss/crossentropy": 3.1857047080993652,
      "loss/fcd": 0.6796875,
      "loss/logits": 0.3227563053369522,
      "step": 36
    },
    {
      "epoch": 0.037,
      "grad_norm": 0.890625,
      "grad_norm_var": 0.045660146077473956,
      "learning_rate": 3.7e-05,
      "loss": 1.0341,
      "loss/crossentropy": 3.1201231479644775,
      "loss/fcd": 0.68359375,
      "loss/logits": 0.35050980746746063,
      "step": 37
    },
    {
      "epoch": 0.038,
      "grad_norm": 0.78125,
      "grad_norm_var": 0.03183797200520833,
      "learning_rate": 3.8e-05,
      "loss": 0.9339,
      "loss/crossentropy": 3.1853790283203125,
      "loss/fcd": 0.623046875,
      "loss/logits": 0.3108416050672531,
      "step": 38
    },
    {
      "epoch": 0.039,
      "grad_norm": 0.87109375,
      "grad_norm_var": 0.02653802235921224,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 1.0078,
      "loss/crossentropy": 3.111618757247925,
      "loss/fcd": 0.654296875,
      "loss/logits": 0.35348545014858246,
      "step": 39
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.83984375,
      "grad_norm_var": 0.014117177327473958,
      "learning_rate": 4e-05,
      "loss": 0.9597,
      "loss/crossentropy": 3.0982449054718018,
      "loss/fcd": 0.64453125,
      "loss/logits": 0.3151911199092865,
      "step": 40
    },
    {
      "epoch": 0.041,
      "grad_norm": 0.72265625,
      "grad_norm_var": 0.012123044331868489,
      "learning_rate": 4.1e-05,
      "loss": 0.8549,
      "loss/crossentropy": 3.0434144735336304,
      "loss/fcd": 0.58984375,
      "loss/logits": 0.2650972008705139,
      "step": 41
    },
    {
      "epoch": 0.042,
      "grad_norm": 0.78125,
      "grad_norm_var": 0.009719785054524739,
      "learning_rate": 4.2e-05,
      "loss": 0.8934,
      "loss/crossentropy": 2.959277391433716,
      "loss/fcd": 0.609375,
      "loss/logits": 0.28400175273418427,
      "step": 42
    },
    {
      "epoch": 0.043,
      "grad_norm": 0.7421875,
      "grad_norm_var": 0.009682146708170573,
      "learning_rate": 4.3e-05,
      "loss": 0.8814,
      "loss/crossentropy": 2.9370049238204956,
      "loss/fcd": 0.59765625,
      "loss/logits": 0.28376346826553345,
      "step": 43
    },
    {
      "epoch": 0.044,
      "grad_norm": 0.75390625,
      "grad_norm_var": 0.009421793619791667,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.8301,
      "loss/crossentropy": 3.0939581394195557,
      "loss/fcd": 0.568359375,
      "loss/logits": 0.2617700919508934,
      "step": 44
    },
    {
      "epoch": 0.045,
      "grad_norm": 0.67578125,
      "grad_norm_var": 0.010454750061035157,
      "learning_rate": 4.5e-05,
      "loss": 0.8153,
      "loss/crossentropy": 3.077065944671631,
      "loss/fcd": 0.5625,
      "loss/logits": 0.2528259754180908,
      "step": 45
    },
    {
      "epoch": 0.046,
      "grad_norm": 0.71484375,
      "grad_norm_var": 0.008750152587890626,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.8464,
      "loss/crossentropy": 3.009420156478882,
      "loss/fcd": 0.580078125,
      "loss/logits": 0.26630888134241104,
      "step": 46
    },
    {
      "epoch": 0.047,
      "grad_norm": 0.703125,
      "grad_norm_var": 0.005863189697265625,
      "learning_rate": 4.7e-05,
      "loss": 0.79,
      "loss/crossentropy": 2.916282534599304,
      "loss/fcd": 0.541015625,
      "loss/logits": 0.24900944530963898,
      "step": 47
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.79296875,
      "grad_norm_var": 0.005191993713378906,
      "learning_rate": 4.8e-05,
      "loss": 0.866,
      "loss/crossentropy": 2.9061275720596313,
      "loss/fcd": 0.591796875,
      "loss/logits": 0.27421246469020844,
      "step": 48
    },
    {
      "epoch": 0.049,
      "grad_norm": 0.6953125,
      "grad_norm_var": 0.005185890197753906,
      "learning_rate": 4.9e-05,
      "loss": 0.8133,
      "loss/crossentropy": 2.9824827909469604,
      "loss/fcd": 0.560546875,
      "loss/logits": 0.25276362150907516,
      "step": 49
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.703125,
      "grad_norm_var": 0.0054361979166666664,
      "learning_rate": 5e-05,
      "loss": 0.7799,
      "loss/crossentropy": 2.917865753173828,
      "loss/fcd": 0.537109375,
      "loss/logits": 0.24279683083295822,
      "step": 50
    },
    {
      "epoch": 0.051,
      "grad_norm": 0.671875,
      "grad_norm_var": 0.005794016520182291,
      "learning_rate": 5.1000000000000006e-05,
      "loss": 0.7346,
      "loss/crossentropy": 2.9159669876098633,
      "loss/fcd": 0.5107421875,
      "loss/logits": 0.22386203706264496,
      "step": 51
    },
    {
      "epoch": 0.052,
      "grad_norm": 0.71484375,
      "grad_norm_var": 0.004561106363932292,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 0.7484,
      "loss/crossentropy": 2.892116069793701,
      "loss/fcd": 0.517578125,
      "loss/logits": 0.2308187410235405,
      "step": 52
    },
    {
      "epoch": 0.053,
      "grad_norm": 0.6640625,
      "grad_norm_var": 0.0036244710286458333,
      "learning_rate": 5.300000000000001e-05,
      "loss": 0.7393,
      "loss/crossentropy": 2.996389389038086,
      "loss/fcd": 0.521484375,
      "loss/logits": 0.21783078461885452,
      "step": 53
    },
    {
      "epoch": 0.054,
      "grad_norm": 0.65234375,
      "grad_norm_var": 0.00394128163655599,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 0.7563,
      "loss/crossentropy": 2.931565999984741,
      "loss/fcd": 0.53125,
      "loss/logits": 0.22508002817630768,
      "step": 54
    },
    {
      "epoch": 0.055,
      "grad_norm": 1.2265625,
      "grad_norm_var": 0.01846898396809896,
      "learning_rate": 5.500000000000001e-05,
      "loss": 0.8031,
      "loss/crossentropy": 3.2799376249313354,
      "loss/fcd": 0.5703125,
      "loss/logits": 0.23281589150428772,
      "step": 55
    },
    {
      "epoch": 0.056,
      "grad_norm": 0.95703125,
      "grad_norm_var": 0.020677693684895835,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 0.7495,
      "loss/crossentropy": 2.97863233089447,
      "loss/fcd": 0.529296875,
      "loss/logits": 0.22019048035144806,
      "step": 56
    },
    {
      "epoch": 0.057,
      "grad_norm": 0.71484375,
      "grad_norm_var": 0.02072118123372396,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 0.6926,
      "loss/crossentropy": 2.888186454772949,
      "loss/fcd": 0.48828125,
      "loss/logits": 0.20430795848369598,
      "step": 57
    },
    {
      "epoch": 0.058,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.021622149149576823,
      "learning_rate": 5.8e-05,
      "loss": 0.6907,
      "loss/crossentropy": 3.028933882713318,
      "loss/fcd": 0.4921875,
      "loss/logits": 0.1985393539071083,
      "step": 58
    },
    {
      "epoch": 0.059,
      "grad_norm": 0.640625,
      "grad_norm_var": 0.02238915761311849,
      "learning_rate": 5.9e-05,
      "loss": 0.6926,
      "loss/crossentropy": 2.909870743751526,
      "loss/fcd": 0.494140625,
      "loss/logits": 0.19847750663757324,
      "step": 59
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.76953125,
      "grad_norm_var": 0.022423235575358073,
      "learning_rate": 6e-05,
      "loss": 0.7102,
      "loss/crossentropy": 3.0572707653045654,
      "loss/fcd": 0.4931640625,
      "loss/logits": 0.21706663817167282,
      "step": 60
    },
    {
      "epoch": 0.061,
      "grad_norm": 0.7265625,
      "grad_norm_var": 0.0221099853515625,
      "learning_rate": 6.1e-05,
      "loss": 0.7628,
      "loss/crossentropy": 3.054797649383545,
      "loss/fcd": 0.53515625,
      "loss/logits": 0.22761277854442596,
      "step": 61
    },
    {
      "epoch": 0.062,
      "grad_norm": 0.83203125,
      "grad_norm_var": 0.022434234619140625,
      "learning_rate": 6.2e-05,
      "loss": 0.7159,
      "loss/crossentropy": 2.89387047290802,
      "loss/fcd": 0.501953125,
      "loss/logits": 0.21391130983829498,
      "step": 62
    },
    {
      "epoch": 0.063,
      "grad_norm": 0.69140625,
      "grad_norm_var": 0.022525978088378907,
      "learning_rate": 6.3e-05,
      "loss": 0.7102,
      "loss/crossentropy": 3.0660817623138428,
      "loss/fcd": 0.50390625,
      "loss/logits": 0.20626945048570633,
      "step": 63
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.71875,
      "grad_norm_var": 0.0225006103515625,
      "learning_rate": 6.400000000000001e-05,
      "loss": 0.6924,
      "loss/crossentropy": 3.02998948097229,
      "loss/fcd": 0.494140625,
      "loss/logits": 0.19828403741121292,
      "step": 64
    },
    {
      "epoch": 0.065,
      "grad_norm": 0.79296875,
      "grad_norm_var": 0.022371864318847655,
      "learning_rate": 6.500000000000001e-05,
      "loss": 0.6387,
      "loss/crossentropy": 2.9985952377319336,
      "loss/fcd": 0.45703125,
      "loss/logits": 0.18168944120407104,
      "step": 65
    },
    {
      "epoch": 0.066,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.02295220692952474,
      "learning_rate": 6.6e-05,
      "loss": 0.6224,
      "loss/crossentropy": 2.8418047428131104,
      "loss/fcd": 0.455078125,
      "loss/logits": 0.1672864779829979,
      "step": 66
    },
    {
      "epoch": 0.067,
      "grad_norm": 0.66015625,
      "grad_norm_var": 0.023088582356770835,
      "learning_rate": 6.7e-05,
      "loss": 0.6538,
      "loss/crossentropy": 3.1374388933181763,
      "loss/fcd": 0.4736328125,
      "loss/logits": 0.18021433055400848,
      "step": 67
    },
    {
      "epoch": 0.068,
      "grad_norm": 0.6171875,
      "grad_norm_var": 0.02418053944905599,
      "learning_rate": 6.800000000000001e-05,
      "loss": 0.6242,
      "loss/crossentropy": 2.95910382270813,
      "loss/fcd": 0.455078125,
      "loss/logits": 0.1690843105316162,
      "step": 68
    },
    {
      "epoch": 0.069,
      "grad_norm": 0.71875,
      "grad_norm_var": 0.02376397450764974,
      "learning_rate": 6.9e-05,
      "loss": 0.6089,
      "loss/crossentropy": 2.8932254314422607,
      "loss/fcd": 0.4453125,
      "loss/logits": 0.16358838975429535,
      "step": 69
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6875,
      "grad_norm_var": 0.02338231404622396,
      "learning_rate": 7e-05,
      "loss": 0.6128,
      "loss/crossentropy": 2.9293062686920166,
      "loss/fcd": 0.4443359375,
      "loss/logits": 0.16845793277025223,
      "step": 70
    },
    {
      "epoch": 0.071,
      "grad_norm": 0.7265625,
      "grad_norm_var": 0.007399241129557292,
      "learning_rate": 7.1e-05,
      "loss": 0.6994,
      "loss/crossentropy": 3.072464942932129,
      "loss/fcd": 0.501953125,
      "loss/logits": 0.19741064310073853,
      "step": 71
    },
    {
      "epoch": 0.072,
      "grad_norm": 0.61328125,
      "grad_norm_var": 0.0039751688639322914,
      "learning_rate": 7.2e-05,
      "loss": 0.6146,
      "loss/crossentropy": 2.8726388216018677,
      "loss/fcd": 0.44921875,
      "loss/logits": 0.16534685343503952,
      "step": 72
    },
    {
      "epoch": 0.073,
      "grad_norm": 0.703125,
      "grad_norm_var": 0.0039601008097330725,
      "learning_rate": 7.3e-05,
      "loss": 0.585,
      "loss/crossentropy": 2.7776870727539062,
      "loss/fcd": 0.427734375,
      "loss/logits": 0.1572359874844551,
      "step": 73
    },
    {
      "epoch": 0.074,
      "grad_norm": 0.66015625,
      "grad_norm_var": 0.003799883524576823,
      "learning_rate": 7.4e-05,
      "loss": 0.668,
      "loss/crossentropy": 3.068588614463806,
      "loss/fcd": 0.4814453125,
      "loss/logits": 0.18657152354717255,
      "step": 74
    },
    {
      "epoch": 0.075,
      "grad_norm": 0.625,
      "grad_norm_var": 0.0039397557576497395,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.5844,
      "loss/crossentropy": 2.985379934310913,
      "loss/fcd": 0.4296875,
      "loss/logits": 0.1547524631023407,
      "step": 75
    },
    {
      "epoch": 0.076,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.0037249247233072918,
      "learning_rate": 7.6e-05,
      "loss": 0.5738,
      "loss/crossentropy": 2.9654282331466675,
      "loss/fcd": 0.419921875,
      "loss/logits": 0.1539217308163643,
      "step": 76
    },
    {
      "epoch": 0.077,
      "grad_norm": 0.59375,
      "grad_norm_var": 0.004213460286458333,
      "learning_rate": 7.7e-05,
      "loss": 0.5822,
      "loss/crossentropy": 3.0354888439178467,
      "loss/fcd": 0.4189453125,
      "loss/logits": 0.16322524845600128,
      "step": 77
    },
    {
      "epoch": 0.078,
      "grad_norm": 0.65625,
      "grad_norm_var": 0.0026656468709309895,
      "learning_rate": 7.800000000000001e-05,
      "loss": 0.6238,
      "loss/crossentropy": 2.976304769515991,
      "loss/fcd": 0.447265625,
      "loss/logits": 0.17654582858085632,
      "step": 78
    },
    {
      "epoch": 0.079,
      "grad_norm": 0.5859375,
      "grad_norm_var": 0.0030965169270833334,
      "learning_rate": 7.900000000000001e-05,
      "loss": 0.5371,
      "loss/crossentropy": 2.884138584136963,
      "loss/fcd": 0.3974609375,
      "loss/logits": 0.13963794708251953,
      "step": 79
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6015625,
      "grad_norm_var": 0.0031308492024739584,
      "learning_rate": 8e-05,
      "loss": 0.5553,
      "loss/crossentropy": 2.6886497735977173,
      "loss/fcd": 0.4140625,
      "loss/logits": 0.14122631400823593,
      "step": 80
    },
    {
      "epoch": 0.081,
      "grad_norm": 0.6015625,
      "grad_norm_var": 0.001993751525878906,
      "learning_rate": 8.1e-05,
      "loss": 0.5781,
      "loss/crossentropy": 2.8824827671051025,
      "loss/fcd": 0.4267578125,
      "loss/logits": 0.1513620838522911,
      "step": 81
    },
    {
      "epoch": 0.082,
      "grad_norm": 0.609375,
      "grad_norm_var": 0.002080217997233073,
      "learning_rate": 8.2e-05,
      "loss": 0.5702,
      "loss/crossentropy": 2.934627056121826,
      "loss/fcd": 0.419921875,
      "loss/logits": 0.15031378716230392,
      "step": 82
    },
    {
      "epoch": 0.083,
      "grad_norm": 0.68359375,
      "grad_norm_var": 0.002164141337076823,
      "learning_rate": 8.3e-05,
      "loss": 0.592,
      "loss/crossentropy": 3.0252050161361694,
      "loss/fcd": 0.4345703125,
      "loss/logits": 0.157474547624588,
      "step": 83
    },
    {
      "epoch": 0.084,
      "grad_norm": 0.7890625,
      "grad_norm_var": 0.003355852762858073,
      "learning_rate": 8.4e-05,
      "loss": 0.5616,
      "loss/crossentropy": 3.0032228231430054,
      "loss/fcd": 0.41796875,
      "loss/logits": 0.1436692625284195,
      "step": 84
    },
    {
      "epoch": 0.085,
      "grad_norm": 0.70703125,
      "grad_norm_var": 0.0032671610514322918,
      "learning_rate": 8.5e-05,
      "loss": 0.5471,
      "loss/crossentropy": 2.9243968725204468,
      "loss/fcd": 0.4033203125,
      "loss/logits": 0.14377443492412567,
      "step": 85
    },
    {
      "epoch": 0.086,
      "grad_norm": 0.6640625,
      "grad_norm_var": 0.0032023111979166668,
      "learning_rate": 8.6e-05,
      "loss": 0.5416,
      "loss/crossentropy": 2.8646360635757446,
      "loss/fcd": 0.4033203125,
      "loss/logits": 0.13825497776269913,
      "step": 86
    },
    {
      "epoch": 0.087,
      "grad_norm": 0.609375,
      "grad_norm_var": 0.0029314676920572918,
      "learning_rate": 8.7e-05,
      "loss": 0.5441,
      "loss/crossentropy": 2.7806389331817627,
      "loss/fcd": 0.4052734375,
      "loss/logits": 0.13886360824108124,
      "step": 87
    },
    {
      "epoch": 0.088,
      "grad_norm": 0.69140625,
      "grad_norm_var": 0.0029619852701822918,
      "learning_rate": 8.800000000000001e-05,
      "loss": 0.5696,
      "loss/crossentropy": 2.923311948776245,
      "loss/fcd": 0.4189453125,
      "loss/logits": 0.15060903877019882,
      "step": 88
    },
    {
      "epoch": 0.089,
      "grad_norm": 0.65234375,
      "grad_norm_var": 0.002776018778483073,
      "learning_rate": 8.900000000000001e-05,
      "loss": 0.5715,
      "loss/crossentropy": 2.9447062015533447,
      "loss/fcd": 0.4169921875,
      "loss/logits": 0.15446894615888596,
      "step": 89
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6171875,
      "grad_norm_var": 0.0028256734212239584,
      "learning_rate": 9e-05,
      "loss": 0.5464,
      "loss/crossentropy": 2.8143328428268433,
      "loss/fcd": 0.4033203125,
      "loss/logits": 0.14312554895877838,
      "step": 90
    },
    {
      "epoch": 0.091,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.0027943929036458332,
      "learning_rate": 9.1e-05,
      "loss": 0.4956,
      "loss/crossentropy": 2.8094122409820557,
      "loss/fcd": 0.375,
      "loss/logits": 0.12059126049280167,
      "step": 91
    },
    {
      "epoch": 0.092,
      "grad_norm": 0.671875,
      "grad_norm_var": 0.0028317769368489585,
      "learning_rate": 9.200000000000001e-05,
      "loss": 0.56,
      "loss/crossentropy": 2.8966939449310303,
      "loss/fcd": 0.4169921875,
      "loss/logits": 0.14303536713123322,
      "step": 92
    },
    {
      "epoch": 0.093,
      "grad_norm": 0.6796875,
      "grad_norm_var": 0.0026611328125,
      "learning_rate": 9.300000000000001e-05,
      "loss": 0.4895,
      "loss/crossentropy": 2.8516520261764526,
      "loss/fcd": 0.3662109375,
      "loss/logits": 0.12323926761746407,
      "step": 93
    },
    {
      "epoch": 0.094,
      "grad_norm": 0.8046875,
      "grad_norm_var": 0.004076894124348958,
      "learning_rate": 9.4e-05,
      "loss": 0.6001,
      "loss/crossentropy": 2.906438112258911,
      "loss/fcd": 0.4208984375,
      "loss/logits": 0.1791856661438942,
      "step": 94
    },
    {
      "epoch": 0.095,
      "grad_norm": 0.64453125,
      "grad_norm_var": 0.0036849339803059896,
      "learning_rate": 9.5e-05,
      "loss": 0.4952,
      "loss/crossentropy": 2.7804044485092163,
      "loss/fcd": 0.37109375,
      "loss/logits": 0.1240886002779007,
      "step": 95
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.81640625,
      "grad_norm_var": 0.0046885172526041664,
      "learning_rate": 9.6e-05,
      "loss": 0.5221,
      "loss/crossentropy": 2.7704015970230103,
      "loss/fcd": 0.3837890625,
      "loss/logits": 0.13829708099365234,
      "step": 96
    },
    {
      "epoch": 0.097,
      "grad_norm": 0.66796875,
      "grad_norm_var": 0.004263750712076823,
      "learning_rate": 9.7e-05,
      "loss": 0.4998,
      "loss/crossentropy": 3.006615400314331,
      "loss/fcd": 0.375,
      "loss/logits": 0.12483775615692139,
      "step": 97
    },
    {
      "epoch": 0.098,
      "grad_norm": 0.99609375,
      "grad_norm_var": 0.009720865885416667,
      "learning_rate": 9.8e-05,
      "loss": 0.5155,
      "loss/crossentropy": 2.900958776473999,
      "loss/fcd": 0.3837890625,
      "loss/logits": 0.13175079226493835,
      "step": 98
    },
    {
      "epoch": 0.099,
      "grad_norm": 1.03125,
      "grad_norm_var": 0.01609795888264974,
      "learning_rate": 9.900000000000001e-05,
      "loss": 0.5303,
      "loss/crossentropy": 2.6889891624450684,
      "loss/fcd": 0.3896484375,
      "loss/logits": 0.14069970324635506,
      "step": 99
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.703125,
      "grad_norm_var": 0.01589094797770182,
      "learning_rate": 0.0001,
      "loss": 0.5076,
      "loss/crossentropy": 2.856358528137207,
      "loss/fcd": 0.3857421875,
      "loss/logits": 0.1218658983707428,
      "step": 100
    },
    {
      "epoch": 0.101,
      "grad_norm": 1.2109375,
      "grad_norm_var": 0.030530802408854165,
      "learning_rate": 0.0001,
      "loss": 0.6061,
      "loss/crossentropy": 2.836049795150757,
      "loss/fcd": 0.3857421875,
      "loss/logits": 0.22031500563025475,
      "step": 101
    },
    {
      "epoch": 0.102,
      "grad_norm": 0.7890625,
      "grad_norm_var": 0.029961140950520833,
      "learning_rate": 0.0001,
      "loss": 0.5018,
      "loss/crossentropy": 2.9406790733337402,
      "loss/fcd": 0.375,
      "loss/logits": 0.12680745497345924,
      "step": 102
    },
    {
      "epoch": 0.103,
      "grad_norm": 0.6328125,
      "grad_norm_var": 0.02951024373372396,
      "learning_rate": 0.0001,
      "loss": 0.4978,
      "loss/crossentropy": 2.9559073448181152,
      "loss/fcd": 0.369140625,
      "loss/logits": 0.12867781147360802,
      "step": 103
    },
    {
      "epoch": 0.104,
      "grad_norm": 0.51953125,
      "grad_norm_var": 0.03306859334309896,
      "learning_rate": 0.0001,
      "loss": 0.4568,
      "loss/crossentropy": 2.858797073364258,
      "loss/fcd": 0.3525390625,
      "loss/logits": 0.10423312336206436,
      "step": 104
    },
    {
      "epoch": 0.105,
      "grad_norm": 0.59375,
      "grad_norm_var": 0.03408807118733724,
      "learning_rate": 0.0001,
      "loss": 0.5192,
      "loss/crossentropy": 2.960332751274109,
      "loss/fcd": 0.3876953125,
      "loss/logits": 0.13145895302295685,
      "step": 105
    },
    {
      "epoch": 0.106,
      "grad_norm": 0.6796875,
      "grad_norm_var": 0.03321119944254557,
      "learning_rate": 0.0001,
      "loss": 0.5741,
      "loss/crossentropy": 2.82626473903656,
      "loss/fcd": 0.427734375,
      "loss/logits": 0.14631643146276474,
      "step": 106
    },
    {
      "epoch": 0.107,
      "grad_norm": 0.60546875,
      "grad_norm_var": 0.033940633138020836,
      "learning_rate": 0.0001,
      "loss": 0.4548,
      "loss/crossentropy": 2.8321757316589355,
      "loss/fcd": 0.3486328125,
      "loss/logits": 0.10612062737345695,
      "step": 107
    },
    {
      "epoch": 0.108,
      "grad_norm": 0.6796875,
      "grad_norm_var": 0.03386001586914063,
      "learning_rate": 0.0001,
      "loss": 0.4943,
      "loss/crossentropy": 2.782582402229309,
      "loss/fcd": 0.3681640625,
      "loss/logits": 0.12610362470149994,
      "step": 108
    },
    {
      "epoch": 0.109,
      "grad_norm": 0.609375,
      "grad_norm_var": 0.0348602294921875,
      "learning_rate": 0.0001,
      "loss": 0.4827,
      "loss/crossentropy": 2.8280365467071533,
      "loss/fcd": 0.359375,
      "loss/logits": 0.1233636736869812,
      "step": 109
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6015625,
      "grad_norm_var": 0.035931396484375,
      "learning_rate": 0.0001,
      "loss": 0.5,
      "loss/crossentropy": 2.964377284049988,
      "loss/fcd": 0.3681640625,
      "loss/logits": 0.13179466873407364,
      "step": 110
    },
    {
      "epoch": 0.111,
      "grad_norm": 0.69140625,
      "grad_norm_var": 0.0354949951171875,
      "learning_rate": 0.0001,
      "loss": 0.4909,
      "loss/crossentropy": 2.8390966653823853,
      "loss/fcd": 0.3662109375,
      "loss/logits": 0.12465999647974968,
      "step": 111
    },
    {
      "epoch": 0.112,
      "grad_norm": 0.7578125,
      "grad_norm_var": 0.035106849670410153,
      "learning_rate": 0.0001,
      "loss": 0.527,
      "loss/crossentropy": 2.828911542892456,
      "loss/fcd": 0.392578125,
      "loss/logits": 0.13445420563220978,
      "step": 112
    },
    {
      "epoch": 0.113,
      "grad_norm": 0.58984375,
      "grad_norm_var": 0.036192766825358075,
      "learning_rate": 0.0001,
      "loss": 0.4785,
      "loss/crossentropy": 2.9939907789230347,
      "loss/fcd": 0.3671875,
      "loss/logits": 0.11130134016275406,
      "step": 113
    },
    {
      "epoch": 0.114,
      "grad_norm": 0.578125,
      "grad_norm_var": 0.032321929931640625,
      "learning_rate": 0.0001,
      "loss": 0.4889,
      "loss/crossentropy": 2.901363730430603,
      "loss/fcd": 0.3671875,
      "loss/logits": 0.12166618555784225,
      "step": 114
    },
    {
      "epoch": 0.115,
      "grad_norm": 0.578125,
      "grad_norm_var": 0.025418853759765624,
      "learning_rate": 0.0001,
      "loss": 0.5046,
      "loss/crossentropy": 2.9244019985198975,
      "loss/fcd": 0.3759765625,
      "loss/logits": 0.1286001205444336,
      "step": 115
    },
    {
      "epoch": 0.116,
      "grad_norm": 0.55859375,
      "grad_norm_var": 0.02620690663655599,
      "learning_rate": 0.0001,
      "loss": 0.4725,
      "loss/crossentropy": 2.8373483419418335,
      "loss/fcd": 0.3564453125,
      "loss/logits": 0.11607612296938896,
      "step": 116
    },
    {
      "epoch": 0.117,
      "grad_norm": 0.5859375,
      "grad_norm_var": 0.005312538146972657,
      "learning_rate": 0.0001,
      "loss": 0.4802,
      "loss/crossentropy": 2.862138271331787,
      "loss/fcd": 0.3583984375,
      "loss/logits": 0.1218356192111969,
      "step": 117
    },
    {
      "epoch": 0.118,
      "grad_norm": 0.5625,
      "grad_norm_var": 0.0036605199178059897,
      "learning_rate": 0.0001,
      "loss": 0.4794,
      "loss/crossentropy": 2.8210216760635376,
      "loss/fcd": 0.36328125,
      "loss/logits": 0.1161160059273243,
      "step": 118
    },
    {
      "epoch": 0.119,
      "grad_norm": 0.54296875,
      "grad_norm_var": 0.0039398193359375,
      "learning_rate": 0.0001,
      "loss": 0.4864,
      "loss/crossentropy": 2.880854606628418,
      "loss/fcd": 0.365234375,
      "loss/logits": 0.12117999419569969,
      "step": 119
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.55078125,
      "grad_norm_var": 0.0036305745442708334,
      "learning_rate": 0.0001,
      "loss": 0.4729,
      "loss/crossentropy": 2.908434748649597,
      "loss/fcd": 0.3564453125,
      "loss/logits": 0.11646857485175133,
      "step": 120
    },
    {
      "epoch": 0.121,
      "grad_norm": 0.5390625,
      "grad_norm_var": 0.003938547770182292,
      "learning_rate": 0.0001,
      "loss": 0.4538,
      "loss/crossentropy": 2.8156157732009888,
      "loss/fcd": 0.345703125,
      "loss/logits": 0.108062494546175,
      "step": 121
    },
    {
      "epoch": 0.122,
      "grad_norm": 0.51953125,
      "grad_norm_var": 0.003988075256347656,
      "learning_rate": 0.0001,
      "loss": 0.4501,
      "loss/crossentropy": 2.8814542293548584,
      "loss/fcd": 0.33984375,
      "loss/logits": 0.11024175211787224,
      "step": 122
    },
    {
      "epoch": 0.123,
      "grad_norm": 0.5390625,
      "grad_norm_var": 0.004188028971354166,
      "learning_rate": 0.0001,
      "loss": 0.4605,
      "loss/crossentropy": 2.794196367263794,
      "loss/fcd": 0.3544921875,
      "loss/logits": 0.10602544993162155,
      "step": 123
    },
    {
      "epoch": 0.124,
      "grad_norm": 0.56640625,
      "grad_norm_var": 0.0036773045857747396,
      "learning_rate": 0.0001,
      "loss": 0.4203,
      "loss/crossentropy": 2.7264811992645264,
      "loss/fcd": 0.3232421875,
      "loss/logits": 0.09708360582590103,
      "step": 124
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.51171875,
      "grad_norm_var": 0.003964996337890625,
      "learning_rate": 0.0001,
      "loss": 0.4584,
      "loss/crossentropy": 2.739405870437622,
      "loss/fcd": 0.3515625,
      "loss/logits": 0.10686014965176582,
      "step": 125
    },
    {
      "epoch": 0.126,
      "grad_norm": 0.546875,
      "grad_norm_var": 0.00399169921875,
      "learning_rate": 0.0001,
      "loss": 0.4395,
      "loss/crossentropy": 2.9202369451522827,
      "loss/fcd": 0.3359375,
      "loss/logits": 0.10352493822574615,
      "step": 126
    },
    {
      "epoch": 0.127,
      "grad_norm": 0.6796875,
      "grad_norm_var": 0.0038202285766601564,
      "learning_rate": 0.0001,
      "loss": 0.4886,
      "loss/crossentropy": 2.872094511985779,
      "loss/fcd": 0.3642578125,
      "loss/logits": 0.12433834001421928,
      "step": 127
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.5859375,
      "grad_norm_var": 0.0014871597290039063,
      "learning_rate": 0.0001,
      "loss": 0.4743,
      "loss/crossentropy": 3.0496490001678467,
      "loss/fcd": 0.3583984375,
      "loss/logits": 0.11591921001672745,
      "step": 128
    },
    {
      "epoch": 0.129,
      "grad_norm": 0.53515625,
      "grad_norm_var": 0.0014907201131184897,
      "learning_rate": 0.0001,
      "loss": 0.4772,
      "loss/crossentropy": 3.0315643548965454,
      "loss/fcd": 0.3603515625,
      "loss/logits": 0.11689409613609314,
      "step": 129
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.625,
      "grad_norm_var": 0.0017333348592122396,
      "learning_rate": 0.0001,
      "loss": 0.4397,
      "loss/crossentropy": 2.784233808517456,
      "loss/fcd": 0.33984375,
      "loss/logits": 0.0998244546353817,
      "step": 130
    },
    {
      "epoch": 0.131,
      "grad_norm": 0.59765625,
      "grad_norm_var": 0.0017934163411458333,
      "learning_rate": 0.0001,
      "loss": 0.4349,
      "loss/crossentropy": 2.7653247117996216,
      "loss/fcd": 0.3310546875,
      "loss/logits": 0.10387945547699928,
      "step": 131
    },
    {
      "epoch": 0.132,
      "grad_norm": 0.61328125,
      "grad_norm_var": 0.0019304911295572917,
      "learning_rate": 0.0001,
      "loss": 0.4506,
      "loss/crossentropy": 2.861555814743042,
      "loss/fcd": 0.3408203125,
      "loss/logits": 0.10979178920388222,
      "step": 132
    },
    {
      "epoch": 0.133,
      "grad_norm": 0.609375,
      "grad_norm_var": 0.002018229166666667,
      "learning_rate": 0.0001,
      "loss": 0.451,
      "loss/crossentropy": 2.8154799938201904,
      "loss/fcd": 0.34375,
      "loss/logits": 0.10724714770913124,
      "step": 133
    },
    {
      "epoch": 0.134,
      "grad_norm": 0.67578125,
      "grad_norm_var": 0.0027022679646809896,
      "learning_rate": 0.0001,
      "loss": 0.4704,
      "loss/crossentropy": 2.9504034519195557,
      "loss/fcd": 0.3525390625,
      "loss/logits": 0.11785488575696945,
      "step": 134
    },
    {
      "epoch": 0.135,
      "grad_norm": 0.6328125,
      "grad_norm_var": 0.0027943929036458332,
      "learning_rate": 0.0001,
      "loss": 0.4539,
      "loss/crossentropy": 2.909873127937317,
      "loss/fcd": 0.345703125,
      "loss/logits": 0.1082424707710743,
      "step": 135
    },
    {
      "epoch": 0.136,
      "grad_norm": 0.921875,
      "grad_norm_var": 0.00980676015218099,
      "learning_rate": 0.0001,
      "loss": 0.4273,
      "loss/crossentropy": 2.8628796339035034,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.10214787721633911,
      "step": 136
    },
    {
      "epoch": 0.137,
      "grad_norm": 0.50390625,
      "grad_norm_var": 0.010198720296223958,
      "learning_rate": 0.0001,
      "loss": 0.4094,
      "loss/crossentropy": 2.8360742330551147,
      "loss/fcd": 0.314453125,
      "loss/logits": 0.0949702151119709,
      "step": 137
    },
    {
      "epoch": 0.138,
      "grad_norm": 0.58203125,
      "grad_norm_var": 0.009738922119140625,
      "learning_rate": 0.0001,
      "loss": 0.4563,
      "loss/crossentropy": 2.9145069122314453,
      "loss/fcd": 0.3427734375,
      "loss/logits": 0.1135670654475689,
      "step": 138
    },
    {
      "epoch": 0.139,
      "grad_norm": 0.6015625,
      "grad_norm_var": 0.009409332275390625,
      "learning_rate": 0.0001,
      "loss": 0.4629,
      "loss/crossentropy": 2.915167808532715,
      "loss/fcd": 0.357421875,
      "loss/logits": 0.10544831678271294,
      "step": 139
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5625,
      "grad_norm_var": 0.009433937072753907,
      "learning_rate": 0.0001,
      "loss": 0.433,
      "loss/crossentropy": 2.8979417085647583,
      "loss/fcd": 0.330078125,
      "loss/logits": 0.10290464386343956,
      "step": 140
    },
    {
      "epoch": 0.141,
      "grad_norm": 0.7890625,
      "grad_norm_var": 0.010548909505208334,
      "learning_rate": 0.0001,
      "loss": 0.5202,
      "loss/crossentropy": 2.953768253326416,
      "loss/fcd": 0.3828125,
      "loss/logits": 0.13743700832128525,
      "step": 141
    },
    {
      "epoch": 0.142,
      "grad_norm": 0.890625,
      "grad_norm_var": 0.014174397786458333,
      "learning_rate": 0.0001,
      "loss": 0.5144,
      "loss/crossentropy": 2.8743395805358887,
      "loss/fcd": 0.376953125,
      "loss/logits": 0.13742298260331154,
      "step": 142
    },
    {
      "epoch": 0.143,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.014113362630208333,
      "learning_rate": 0.0001,
      "loss": 0.4276,
      "loss/crossentropy": 2.8286346197128296,
      "loss/fcd": 0.328125,
      "loss/logits": 0.09949736669659615,
      "step": 143
    },
    {
      "epoch": 0.144,
      "grad_norm": 0.79296875,
      "grad_norm_var": 0.015066973368326823,
      "learning_rate": 0.0001,
      "loss": 0.4918,
      "loss/crossentropy": 2.908802032470703,
      "loss/fcd": 0.3671875,
      "loss/logits": 0.12464811280369759,
      "step": 144
    },
    {
      "epoch": 0.145,
      "grad_norm": 1.0625,
      "grad_norm_var": 0.023572794596354165,
      "learning_rate": 0.0001,
      "loss": 0.4983,
      "loss/crossentropy": 3.0399140119552612,
      "loss/fcd": 0.3662109375,
      "loss/logits": 0.13205987215042114,
      "step": 145
    },
    {
      "epoch": 0.146,
      "grad_norm": 0.85546875,
      "grad_norm_var": 0.02476189931233724,
      "learning_rate": 0.0001,
      "loss": 0.4498,
      "loss/crossentropy": 2.7846896648406982,
      "loss/fcd": 0.3447265625,
      "loss/logits": 0.10509144142270088,
      "step": 146
    },
    {
      "epoch": 0.147,
      "grad_norm": 0.62109375,
      "grad_norm_var": 0.02444909413655599,
      "learning_rate": 0.0001,
      "loss": 0.4078,
      "loss/crossentropy": 2.7551146745681763,
      "loss/fcd": 0.3203125,
      "loss/logits": 0.0875249132514,
      "step": 147
    },
    {
      "epoch": 0.148,
      "grad_norm": 0.62890625,
      "grad_norm_var": 0.024262428283691406,
      "learning_rate": 0.0001,
      "loss": 0.4494,
      "loss/crossentropy": 3.0209089517593384,
      "loss/fcd": 0.3427734375,
      "loss/logits": 0.10662630200386047,
      "step": 148
    },
    {
      "epoch": 0.149,
      "grad_norm": 0.703125,
      "grad_norm_var": 0.023539161682128905,
      "learning_rate": 0.0001,
      "loss": 0.4343,
      "loss/crossentropy": 2.762128233909607,
      "loss/fcd": 0.330078125,
      "loss/logits": 0.10424429923295975,
      "step": 149
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.671875,
      "grad_norm_var": 0.023561604817708335,
      "learning_rate": 0.0001,
      "loss": 0.4631,
      "loss/crossentropy": 2.9137667417526245,
      "loss/fcd": 0.3466796875,
      "loss/logits": 0.11645066738128662,
      "step": 150
    },
    {
      "epoch": 0.151,
      "grad_norm": 0.5390625,
      "grad_norm_var": 0.025160725911458334,
      "learning_rate": 0.0001,
      "loss": 0.4427,
      "loss/crossentropy": 2.881830334663391,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.10573653504252434,
      "step": 151
    },
    {
      "epoch": 0.152,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.02222283681233724,
      "learning_rate": 0.0001,
      "loss": 0.5156,
      "loss/crossentropy": 2.850717067718506,
      "loss/fcd": 0.3798828125,
      "loss/logits": 0.1356773041188717,
      "step": 152
    },
    {
      "epoch": 0.153,
      "grad_norm": 0.55859375,
      "grad_norm_var": 0.021030108133951824,
      "learning_rate": 0.0001,
      "loss": 0.4536,
      "loss/crossentropy": 2.8608055114746094,
      "loss/fcd": 0.3447265625,
      "loss/logits": 0.10892104357481003,
      "step": 153
    },
    {
      "epoch": 0.154,
      "grad_norm": 0.55078125,
      "grad_norm_var": 0.02156823476155599,
      "learning_rate": 0.0001,
      "loss": 0.4315,
      "loss/crossentropy": 2.800734758377075,
      "loss/fcd": 0.3310546875,
      "loss/logits": 0.1004045195877552,
      "step": 154
    },
    {
      "epoch": 0.155,
      "grad_norm": 0.58984375,
      "grad_norm_var": 0.021722157796223957,
      "learning_rate": 0.0001,
      "loss": 0.4312,
      "loss/crossentropy": 2.7723549604415894,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.09424581006169319,
      "step": 155
    },
    {
      "epoch": 0.156,
      "grad_norm": 0.6015625,
      "grad_norm_var": 0.0211334228515625,
      "learning_rate": 0.0001,
      "loss": 0.4285,
      "loss/crossentropy": 2.7112516164779663,
      "loss/fcd": 0.333984375,
      "loss/logits": 0.09450496360659599,
      "step": 156
    },
    {
      "epoch": 0.157,
      "grad_norm": 0.55078125,
      "grad_norm_var": 0.02173455556233724,
      "learning_rate": 0.0001,
      "loss": 0.405,
      "loss/crossentropy": 2.8753061294555664,
      "loss/fcd": 0.30859375,
      "loss/logits": 0.0963931567966938,
      "step": 157
    },
    {
      "epoch": 0.158,
      "grad_norm": 0.59765625,
      "grad_norm_var": 0.018925984700520832,
      "learning_rate": 0.0001,
      "loss": 0.4793,
      "loss/crossentropy": 2.859338641166687,
      "loss/fcd": 0.3623046875,
      "loss/logits": 0.1170136034488678,
      "step": 158
    },
    {
      "epoch": 0.159,
      "grad_norm": 0.59765625,
      "grad_norm_var": 0.01918633778889974,
      "learning_rate": 0.0001,
      "loss": 0.4427,
      "loss/crossentropy": 2.9007763862609863,
      "loss/fcd": 0.33984375,
      "loss/logits": 0.10285023972392082,
      "step": 159
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.54296875,
      "grad_norm_var": 0.018657366434733074,
      "learning_rate": 0.0001,
      "loss": 0.4145,
      "loss/crossentropy": 2.764789581298828,
      "loss/fcd": 0.3193359375,
      "loss/logits": 0.09517998620867729,
      "step": 160
    },
    {
      "epoch": 0.161,
      "grad_norm": 0.55859375,
      "grad_norm_var": 0.006428782145182292,
      "learning_rate": 0.0001,
      "loss": 0.4301,
      "loss/crossentropy": 2.9352102279663086,
      "loss/fcd": 0.3291015625,
      "loss/logits": 0.10102442279458046,
      "step": 161
    },
    {
      "epoch": 0.162,
      "grad_norm": 0.7109375,
      "grad_norm_var": 0.0030577977498372397,
      "learning_rate": 0.0001,
      "loss": 0.4884,
      "loss/crossentropy": 2.8804049491882324,
      "loss/fcd": 0.3720703125,
      "loss/logits": 0.11631228029727936,
      "step": 162
    },
    {
      "epoch": 0.163,
      "grad_norm": 0.59765625,
      "grad_norm_var": 0.0030379613240559896,
      "learning_rate": 0.0001,
      "loss": 0.4418,
      "loss/crossentropy": 2.9045521020889282,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.10484013333916664,
      "step": 163
    },
    {
      "epoch": 0.164,
      "grad_norm": 0.51953125,
      "grad_norm_var": 0.003397560119628906,
      "learning_rate": 0.0001,
      "loss": 0.4228,
      "loss/crossentropy": 2.84330677986145,
      "loss/fcd": 0.3232421875,
      "loss/logits": 0.09951141849160194,
      "step": 164
    },
    {
      "epoch": 0.165,
      "grad_norm": 0.6171875,
      "grad_norm_var": 0.0026254653930664062,
      "learning_rate": 0.0001,
      "loss": 0.4573,
      "loss/crossentropy": 2.9289658069610596,
      "loss/fcd": 0.34765625,
      "loss/logits": 0.1096344031393528,
      "step": 165
    },
    {
      "epoch": 0.166,
      "grad_norm": 0.62109375,
      "grad_norm_var": 0.0022328694661458335,
      "learning_rate": 0.0001,
      "loss": 0.4325,
      "loss/crossentropy": 2.9142534732818604,
      "loss/fcd": 0.3359375,
      "loss/logits": 0.09654596820473671,
      "step": 166
    },
    {
      "epoch": 0.167,
      "grad_norm": 0.59765625,
      "grad_norm_var": 0.0020736058553059897,
      "learning_rate": 0.0001,
      "loss": 0.4141,
      "loss/crossentropy": 2.722415328025818,
      "loss/fcd": 0.318359375,
      "loss/logits": 0.09574693441390991,
      "step": 167
    },
    {
      "epoch": 0.168,
      "grad_norm": 0.48046875,
      "grad_norm_var": 0.0026381810506184897,
      "learning_rate": 0.0001,
      "loss": 0.4442,
      "loss/crossentropy": 2.9160542488098145,
      "loss/fcd": 0.337890625,
      "loss/logits": 0.10633675754070282,
      "step": 168
    },
    {
      "epoch": 0.169,
      "grad_norm": 0.53125,
      "grad_norm_var": 0.002765909830729167,
      "learning_rate": 0.0001,
      "loss": 0.437,
      "loss/crossentropy": 2.8252073526382446,
      "loss/fcd": 0.333984375,
      "loss/logits": 0.10301430523395538,
      "step": 169
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.474609375,
      "grad_norm_var": 0.0034161726633707683,
      "learning_rate": 0.0001,
      "loss": 0.4145,
      "loss/crossentropy": 2.8485448360443115,
      "loss/fcd": 0.3203125,
      "loss/logits": 0.09413985908031464,
      "step": 170
    },
    {
      "epoch": 0.171,
      "grad_norm": 0.6953125,
      "grad_norm_var": 0.004329411188761393,
      "learning_rate": 0.0001,
      "loss": 0.4442,
      "loss/crossentropy": 2.7965201139450073,
      "loss/fcd": 0.3427734375,
      "loss/logits": 0.1014043353497982,
      "step": 171
    },
    {
      "epoch": 0.172,
      "grad_norm": 0.6328125,
      "grad_norm_var": 0.004476404190063477,
      "learning_rate": 0.0001,
      "loss": 0.4499,
      "loss/crossentropy": 2.9560967683792114,
      "loss/fcd": 0.3447265625,
      "loss/logits": 0.1051417626440525,
      "step": 172
    },
    {
      "epoch": 0.173,
      "grad_norm": 0.64453125,
      "grad_norm_var": 0.004624414443969727,
      "learning_rate": 0.0001,
      "loss": 0.4659,
      "loss/crossentropy": 2.8695074319839478,
      "loss/fcd": 0.35546875,
      "loss/logits": 0.11045186966657639,
      "step": 173
    },
    {
      "epoch": 0.174,
      "grad_norm": 0.515625,
      "grad_norm_var": 0.004947519302368164,
      "learning_rate": 0.0001,
      "loss": 0.3876,
      "loss/crossentropy": 2.6555715799331665,
      "loss/fcd": 0.3056640625,
      "loss/logits": 0.08189895376563072,
      "step": 174
    },
    {
      "epoch": 0.175,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.005116001764933268,
      "learning_rate": 0.0001,
      "loss": 0.4016,
      "loss/crossentropy": 2.8286584615707397,
      "loss/fcd": 0.310546875,
      "loss/logits": 0.09104615077376366,
      "step": 175
    },
    {
      "epoch": 0.176,
      "grad_norm": 0.5390625,
      "grad_norm_var": 0.005139398574829102,
      "learning_rate": 0.0001,
      "loss": 0.4242,
      "loss/crossentropy": 2.8632349967956543,
      "loss/fcd": 0.3232421875,
      "loss/logits": 0.10094088688492775,
      "step": 176
    },
    {
      "epoch": 0.177,
      "grad_norm": 0.55859375,
      "grad_norm_var": 0.005139398574829102,
      "learning_rate": 0.0001,
      "loss": 0.4185,
      "loss/crossentropy": 2.834444522857666,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09620294347405434,
      "step": 177
    },
    {
      "epoch": 0.178,
      "grad_norm": 0.5625,
      "grad_norm_var": 0.004040129979451497,
      "learning_rate": 0.0001,
      "loss": 0.3848,
      "loss/crossentropy": 2.7522603273391724,
      "loss/fcd": 0.3017578125,
      "loss/logits": 0.08301593363285065,
      "step": 178
    },
    {
      "epoch": 0.179,
      "grad_norm": 0.5625,
      "grad_norm_var": 0.004018386205037435,
      "learning_rate": 0.0001,
      "loss": 0.4228,
      "loss/crossentropy": 2.8634541034698486,
      "loss/fcd": 0.3232421875,
      "loss/logits": 0.09959971159696579,
      "step": 179
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.52734375,
      "grad_norm_var": 0.003965107599894205,
      "learning_rate": 0.0001,
      "loss": 0.4212,
      "loss/crossentropy": 2.8671987056732178,
      "loss/fcd": 0.32421875,
      "loss/logits": 0.09694879502058029,
      "step": 180
    },
    {
      "epoch": 0.181,
      "grad_norm": 0.578125,
      "grad_norm_var": 0.0038398583730061848,
      "learning_rate": 0.0001,
      "loss": 0.4259,
      "loss/crossentropy": 2.9206082820892334,
      "loss/fcd": 0.32421875,
      "loss/logits": 0.10163932293653488,
      "step": 181
    },
    {
      "epoch": 0.182,
      "grad_norm": 0.609375,
      "grad_norm_var": 0.0037723382314046225,
      "learning_rate": 0.0001,
      "loss": 0.4385,
      "loss/crossentropy": 2.8616329431533813,
      "loss/fcd": 0.33203125,
      "loss/logits": 0.10650401189923286,
      "step": 182
    },
    {
      "epoch": 0.183,
      "grad_norm": 0.56640625,
      "grad_norm_var": 0.003725035985310872,
      "learning_rate": 0.0001,
      "loss": 0.4238,
      "loss/crossentropy": 2.743476986885071,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.09862633794546127,
      "step": 183
    },
    {
      "epoch": 0.184,
      "grad_norm": 0.5625,
      "grad_norm_var": 0.0031696160634358725,
      "learning_rate": 0.0001,
      "loss": 0.4195,
      "loss/crossentropy": 2.822865128517151,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09720025584101677,
      "step": 184
    },
    {
      "epoch": 0.185,
      "grad_norm": 0.51953125,
      "grad_norm_var": 0.0032462914784749347,
      "learning_rate": 0.0001,
      "loss": 0.395,
      "loss/crossentropy": 2.875390648841858,
      "loss/fcd": 0.3076171875,
      "loss/logits": 0.08735854551196098,
      "step": 185
    },
    {
      "epoch": 0.186,
      "grad_norm": 0.56640625,
      "grad_norm_var": 0.0025552749633789063,
      "learning_rate": 0.0001,
      "loss": 0.45,
      "loss/crossentropy": 2.821836471557617,
      "loss/fcd": 0.341796875,
      "loss/logits": 0.10815925523638725,
      "step": 186
    },
    {
      "epoch": 0.187,
      "grad_norm": 0.59375,
      "grad_norm_var": 0.001636187235514323,
      "learning_rate": 0.0001,
      "loss": 0.4167,
      "loss/crossentropy": 2.9735742807388306,
      "loss/fcd": 0.318359375,
      "loss/logits": 0.09838059172034264,
      "step": 187
    },
    {
      "epoch": 0.188,
      "grad_norm": 0.57421875,
      "grad_norm_var": 0.0013872782389322917,
      "learning_rate": 0.0001,
      "loss": 0.4461,
      "loss/crossentropy": 2.8691996335983276,
      "loss/fcd": 0.341796875,
      "loss/logits": 0.10430673509836197,
      "step": 188
    },
    {
      "epoch": 0.189,
      "grad_norm": 0.515625,
      "grad_norm_var": 0.001141802469889323,
      "learning_rate": 0.0001,
      "loss": 0.4334,
      "loss/crossentropy": 2.7962933778762817,
      "loss/fcd": 0.3330078125,
      "loss/logits": 0.1004236750304699,
      "step": 189
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5625,
      "grad_norm_var": 0.0009907404581705728,
      "learning_rate": 0.0001,
      "loss": 0.4124,
      "loss/crossentropy": 2.9746346473693848,
      "loss/fcd": 0.3173828125,
      "loss/logits": 0.09501497820019722,
      "step": 190
    },
    {
      "epoch": 0.191,
      "grad_norm": 0.6328125,
      "grad_norm_var": 0.0009541829427083333,
      "learning_rate": 0.0001,
      "loss": 0.4909,
      "loss/crossentropy": 3.0425490140914917,
      "loss/fcd": 0.369140625,
      "loss/logits": 0.12177145853638649,
      "step": 191
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.6328125,
      "grad_norm_var": 0.0011861165364583333,
      "learning_rate": 0.0001,
      "loss": 0.4268,
      "loss/crossentropy": 2.8151696920394897,
      "loss/fcd": 0.33203125,
      "loss/logits": 0.09481121972203255,
      "step": 192
    },
    {
      "epoch": 0.193,
      "grad_norm": 0.70703125,
      "grad_norm_var": 0.0023312886555989583,
      "learning_rate": 0.0001,
      "loss": 0.4791,
      "loss/crossentropy": 2.8669657707214355,
      "loss/fcd": 0.3662109375,
      "loss/logits": 0.11289490759372711,
      "step": 193
    },
    {
      "epoch": 0.194,
      "grad_norm": 0.66796875,
      "grad_norm_var": 0.0027861913045247396,
      "learning_rate": 0.0001,
      "loss": 0.3946,
      "loss/crossentropy": 2.7116475105285645,
      "loss/fcd": 0.30859375,
      "loss/logits": 0.0860372893512249,
      "step": 194
    },
    {
      "epoch": 0.195,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.0029764175415039062,
      "learning_rate": 0.0001,
      "loss": 0.4165,
      "loss/crossentropy": 2.870499610900879,
      "loss/fcd": 0.3212890625,
      "loss/logits": 0.09520205855369568,
      "step": 195
    },
    {
      "epoch": 0.196,
      "grad_norm": 0.6875,
      "grad_norm_var": 0.003208414713541667,
      "learning_rate": 0.0001,
      "loss": 0.419,
      "loss/crossentropy": 2.9276784658432007,
      "loss/fcd": 0.318359375,
      "loss/logits": 0.10059179738163948,
      "step": 196
    },
    {
      "epoch": 0.197,
      "grad_norm": 0.7109375,
      "grad_norm_var": 0.0038958231608072916,
      "learning_rate": 0.0001,
      "loss": 0.4725,
      "loss/crossentropy": 2.845921277999878,
      "loss/fcd": 0.3505859375,
      "loss/logits": 0.12191061675548553,
      "step": 197
    },
    {
      "epoch": 0.198,
      "grad_norm": 0.71484375,
      "grad_norm_var": 0.004584185282389323,
      "learning_rate": 0.0001,
      "loss": 0.453,
      "loss/crossentropy": 2.8804662227630615,
      "loss/fcd": 0.3359375,
      "loss/logits": 0.11705750972032547,
      "step": 198
    },
    {
      "epoch": 0.199,
      "grad_norm": 0.53515625,
      "grad_norm_var": 0.00485375722249349,
      "learning_rate": 0.0001,
      "loss": 0.3999,
      "loss/crossentropy": 2.681739091873169,
      "loss/fcd": 0.3115234375,
      "loss/logits": 0.08841734752058983,
      "step": 199
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.57421875,
      "grad_norm_var": 0.004781087239583333,
      "learning_rate": 0.0001,
      "loss": 0.4227,
      "loss/crossentropy": 2.968848705291748,
      "loss/fcd": 0.3212890625,
      "loss/logits": 0.10139760375022888,
      "step": 200
    },
    {
      "epoch": 0.201,
      "grad_norm": 0.609375,
      "grad_norm_var": 0.004139137268066406,
      "learning_rate": 0.0001,
      "loss": 0.4464,
      "loss/crossentropy": 2.8944156169891357,
      "loss/fcd": 0.3388671875,
      "loss/logits": 0.1075349859893322,
      "step": 201
    },
    {
      "epoch": 0.202,
      "grad_norm": 0.62890625,
      "grad_norm_var": 0.0039295832316080725,
      "learning_rate": 0.0001,
      "loss": 0.4596,
      "loss/crossentropy": 2.9402053356170654,
      "loss/fcd": 0.3447265625,
      "loss/logits": 0.11490758880972862,
      "step": 202
    },
    {
      "epoch": 0.203,
      "grad_norm": 0.5859375,
      "grad_norm_var": 0.003965695699055989,
      "learning_rate": 0.0001,
      "loss": 0.4385,
      "loss/crossentropy": 2.930693745613098,
      "loss/fcd": 0.3330078125,
      "loss/logits": 0.10551033169031143,
      "step": 203
    },
    {
      "epoch": 0.204,
      "grad_norm": 0.671875,
      "grad_norm_var": 0.003910064697265625,
      "learning_rate": 0.0001,
      "loss": 0.4432,
      "loss/crossentropy": 2.9876062870025635,
      "loss/fcd": 0.3349609375,
      "loss/logits": 0.10823972150683403,
      "step": 204
    },
    {
      "epoch": 0.205,
      "grad_norm": 0.828125,
      "grad_norm_var": 0.005232493082682292,
      "learning_rate": 0.0001,
      "loss": 0.4812,
      "loss/crossentropy": 2.9890397787094116,
      "loss/fcd": 0.3583984375,
      "loss/logits": 0.12285104021430016,
      "step": 205
    },
    {
      "epoch": 0.206,
      "grad_norm": 0.765625,
      "grad_norm_var": 0.005444081624348959,
      "learning_rate": 0.0001,
      "loss": 0.3894,
      "loss/crossentropy": 2.867851734161377,
      "loss/fcd": 0.3037109375,
      "loss/logits": 0.08565087616443634,
      "step": 206
    },
    {
      "epoch": 0.207,
      "grad_norm": 0.89453125,
      "grad_norm_var": 0.008685747782389322,
      "learning_rate": 0.0001,
      "loss": 0.441,
      "loss/crossentropy": 3.0004996061325073,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.10407992079854012,
      "step": 207
    },
    {
      "epoch": 0.208,
      "grad_norm": 0.82421875,
      "grad_norm_var": 0.009797922770182292,
      "learning_rate": 0.0001,
      "loss": 0.4171,
      "loss/crossentropy": 2.8082648515701294,
      "loss/fcd": 0.3212890625,
      "loss/logits": 0.09578917548060417,
      "step": 208
    },
    {
      "epoch": 0.209,
      "grad_norm": 0.69921875,
      "grad_norm_var": 0.009784952799479166,
      "learning_rate": 0.0001,
      "loss": 0.4368,
      "loss/crossentropy": 2.928029775619507,
      "loss/fcd": 0.33203125,
      "loss/logits": 0.10473304614424706,
      "step": 209
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.494140625,
      "grad_norm_var": 0.012194045384724935,
      "learning_rate": 0.0001,
      "loss": 0.3936,
      "loss/crossentropy": 2.699643611907959,
      "loss/fcd": 0.30859375,
      "loss/logits": 0.0850089080631733,
      "step": 210
    },
    {
      "epoch": 0.211,
      "grad_norm": 0.57421875,
      "grad_norm_var": 0.01284635861714681,
      "learning_rate": 0.0001,
      "loss": 0.4002,
      "loss/crossentropy": 2.853596329689026,
      "loss/fcd": 0.3076171875,
      "loss/logits": 0.09262469410896301,
      "step": 211
    },
    {
      "epoch": 0.212,
      "grad_norm": 0.62890625,
      "grad_norm_var": 0.012962706883748372,
      "learning_rate": 0.0001,
      "loss": 0.4106,
      "loss/crossentropy": 2.7878748178482056,
      "loss/fcd": 0.3212890625,
      "loss/logits": 0.0892898328602314,
      "step": 212
    },
    {
      "epoch": 0.213,
      "grad_norm": 1.0859375,
      "grad_norm_var": 0.023735411961873374,
      "learning_rate": 0.0001,
      "loss": 0.516,
      "loss/crossentropy": 2.8341736793518066,
      "loss/fcd": 0.373046875,
      "loss/logits": 0.1429123878479004,
      "step": 213
    },
    {
      "epoch": 0.214,
      "grad_norm": 0.71484375,
      "grad_norm_var": 0.023735411961873374,
      "learning_rate": 0.0001,
      "loss": 0.417,
      "loss/crossentropy": 2.848291754722595,
      "loss/fcd": 0.3193359375,
      "loss/logits": 0.09763862192630768,
      "step": 214
    },
    {
      "epoch": 0.215,
      "grad_norm": 0.74609375,
      "grad_norm_var": 0.02202909787495931,
      "learning_rate": 0.0001,
      "loss": 0.4259,
      "loss/crossentropy": 3.0413947105407715,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.10072237998247147,
      "step": 215
    },
    {
      "epoch": 0.216,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.0210506280263265,
      "learning_rate": 0.0001,
      "loss": 0.4208,
      "loss/crossentropy": 2.9306408166885376,
      "loss/fcd": 0.318359375,
      "loss/logits": 0.10242888703942299,
      "step": 216
    },
    {
      "epoch": 0.217,
      "grad_norm": 0.80078125,
      "grad_norm_var": 0.02070794105529785,
      "learning_rate": 0.0001,
      "loss": 0.4511,
      "loss/crossentropy": 2.8223299980163574,
      "loss/fcd": 0.3408203125,
      "loss/logits": 0.11031396687030792,
      "step": 217
    },
    {
      "epoch": 0.218,
      "grad_norm": 0.6171875,
      "grad_norm_var": 0.020865869522094727,
      "learning_rate": 0.0001,
      "loss": 0.3949,
      "loss/crossentropy": 2.7668732404708862,
      "loss/fcd": 0.3095703125,
      "loss/logits": 0.08537604659795761,
      "step": 218
    },
    {
      "epoch": 0.219,
      "grad_norm": 0.6640625,
      "grad_norm_var": 0.019811741511027017,
      "learning_rate": 0.0001,
      "loss": 0.4237,
      "loss/crossentropy": 2.9836037158966064,
      "loss/fcd": 0.318359375,
      "loss/logits": 0.10530032590031624,
      "step": 219
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.66796875,
      "grad_norm_var": 0.019842259089152017,
      "learning_rate": 0.0001,
      "loss": 0.3983,
      "loss/crossentropy": 2.747257947921753,
      "loss/fcd": 0.3076171875,
      "loss/logits": 0.09065764769911766,
      "step": 220
    },
    {
      "epoch": 0.221,
      "grad_norm": 0.6640625,
      "grad_norm_var": 0.019342915217081705,
      "learning_rate": 0.0001,
      "loss": 0.43,
      "loss/crossentropy": 2.845228672027588,
      "loss/fcd": 0.330078125,
      "loss/logits": 0.09995642304420471,
      "step": 221
    },
    {
      "epoch": 0.222,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.019459263483683268,
      "learning_rate": 0.0001,
      "loss": 0.4125,
      "loss/crossentropy": 2.8502864837646484,
      "loss/fcd": 0.3203125,
      "loss/logits": 0.09222512319684029,
      "step": 222
    },
    {
      "epoch": 0.223,
      "grad_norm": 0.53515625,
      "grad_norm_var": 0.018728113174438475,
      "learning_rate": 0.0001,
      "loss": 0.4058,
      "loss/crossentropy": 2.924101948738098,
      "loss/fcd": 0.3115234375,
      "loss/logits": 0.0942913256585598,
      "step": 223
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.58984375,
      "grad_norm_var": 0.01791558265686035,
      "learning_rate": 0.0001,
      "loss": 0.4231,
      "loss/crossentropy": 2.9752752780914307,
      "loss/fcd": 0.326171875,
      "loss/logits": 0.09692845121026039,
      "step": 224
    },
    {
      "epoch": 0.225,
      "grad_norm": 0.60546875,
      "grad_norm_var": 0.018145990371704102,
      "learning_rate": 0.0001,
      "loss": 0.4163,
      "loss/crossentropy": 2.9902013540267944,
      "loss/fcd": 0.3203125,
      "loss/logits": 0.09601914137601852,
      "step": 225
    },
    {
      "epoch": 0.226,
      "grad_norm": 0.60546875,
      "grad_norm_var": 0.0163421630859375,
      "learning_rate": 0.0001,
      "loss": 0.4282,
      "loss/crossentropy": 2.9544960260391235,
      "loss/fcd": 0.32421875,
      "loss/logits": 0.10400371253490448,
      "step": 226
    },
    {
      "epoch": 0.227,
      "grad_norm": 0.52734375,
      "grad_norm_var": 0.017108154296875,
      "learning_rate": 0.0001,
      "loss": 0.4189,
      "loss/crossentropy": 2.9443225860595703,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09667930379509926,
      "step": 227
    },
    {
      "epoch": 0.228,
      "grad_norm": 0.5625,
      "grad_norm_var": 0.01776421864827474,
      "learning_rate": 0.0001,
      "loss": 0.4086,
      "loss/crossentropy": 2.8471500873565674,
      "loss/fcd": 0.3154296875,
      "loss/logits": 0.09314657375216484,
      "step": 228
    },
    {
      "epoch": 0.229,
      "grad_norm": 0.78515625,
      "grad_norm_var": 0.006646474202473958,
      "learning_rate": 0.0001,
      "loss": 0.4747,
      "loss/crossentropy": 3.0068620443344116,
      "loss/fcd": 0.353515625,
      "loss/logits": 0.1211743988096714,
      "step": 229
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0390625,
      "grad_norm_var": 0.01606591542561849,
      "learning_rate": 0.0001,
      "loss": 0.5054,
      "loss/crossentropy": 2.904601573944092,
      "loss/fcd": 0.357421875,
      "loss/logits": 0.1479528360068798,
      "step": 230
    },
    {
      "epoch": 0.231,
      "grad_norm": 0.71484375,
      "grad_norm_var": 0.01580651601155599,
      "learning_rate": 0.0001,
      "loss": 0.413,
      "loss/crossentropy": 2.8541531562805176,
      "loss/fcd": 0.314453125,
      "loss/logits": 0.09850187227129936,
      "step": 231
    },
    {
      "epoch": 0.232,
      "grad_norm": 0.8359375,
      "grad_norm_var": 0.01753381093343099,
      "learning_rate": 0.0001,
      "loss": 0.4665,
      "loss/crossentropy": 3.0907140970230103,
      "loss/fcd": 0.3447265625,
      "loss/logits": 0.12178562209010124,
      "step": 232
    },
    {
      "epoch": 0.233,
      "grad_norm": 0.78125,
      "grad_norm_var": 0.017240397135416665,
      "learning_rate": 0.0001,
      "loss": 0.4282,
      "loss/crossentropy": 2.7091563940048218,
      "loss/fcd": 0.3310546875,
      "loss/logits": 0.09712602570652962,
      "step": 233
    },
    {
      "epoch": 0.234,
      "grad_norm": 1.3046875,
      "grad_norm_var": 0.04123128255208333,
      "learning_rate": 0.0001,
      "loss": 0.504,
      "loss/crossentropy": 2.912622332572937,
      "loss/fcd": 0.3564453125,
      "loss/logits": 0.14751752838492393,
      "step": 234
    },
    {
      "epoch": 0.235,
      "grad_norm": 0.7890625,
      "grad_norm_var": 0.04126383463541667,
      "learning_rate": 0.0001,
      "loss": 0.4265,
      "loss/crossentropy": 2.7788244485855103,
      "loss/fcd": 0.3232421875,
      "loss/logits": 0.10323498025536537,
      "step": 235
    },
    {
      "epoch": 0.236,
      "grad_norm": 0.984375,
      "grad_norm_var": 0.044966570536295575,
      "learning_rate": 0.0001,
      "loss": 0.4324,
      "loss/crossentropy": 2.8856834173202515,
      "loss/fcd": 0.3291015625,
      "loss/logits": 0.10333634912967682,
      "step": 236
    },
    {
      "epoch": 0.237,
      "grad_norm": 1.0234375,
      "grad_norm_var": 0.049002520243326825,
      "learning_rate": 0.0001,
      "loss": 0.4056,
      "loss/crossentropy": 2.8181627988815308,
      "loss/fcd": 0.3125,
      "loss/logits": 0.09313894063234329,
      "step": 237
    },
    {
      "epoch": 0.238,
      "grad_norm": 0.95703125,
      "grad_norm_var": 0.049921671549479164,
      "learning_rate": 0.0001,
      "loss": 0.4479,
      "loss/crossentropy": 2.959288477897644,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.11102355644106865,
      "step": 238
    },
    {
      "epoch": 0.239,
      "grad_norm": 0.8515625,
      "grad_norm_var": 0.04542586008707682,
      "learning_rate": 0.0001,
      "loss": 0.4428,
      "loss/crossentropy": 2.8065025806427,
      "loss/fcd": 0.333984375,
      "loss/logits": 0.10878817737102509,
      "step": 239
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.76171875,
      "grad_norm_var": 0.042231178283691405,
      "learning_rate": 0.0001,
      "loss": 0.396,
      "loss/crossentropy": 2.7715532779693604,
      "loss/fcd": 0.3095703125,
      "loss/logits": 0.08645889163017273,
      "step": 240
    },
    {
      "epoch": 0.241,
      "grad_norm": 0.75,
      "grad_norm_var": 0.03939183553059896,
      "learning_rate": 0.0001,
      "loss": 0.4564,
      "loss/crossentropy": 2.829145073890686,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.11950911581516266,
      "step": 241
    },
    {
      "epoch": 0.242,
      "grad_norm": 0.71484375,
      "grad_norm_var": 0.036871083577473956,
      "learning_rate": 0.0001,
      "loss": 0.4192,
      "loss/crossentropy": 2.8319283723831177,
      "loss/fcd": 0.326171875,
      "loss/logits": 0.09307562187314034,
      "step": 242
    },
    {
      "epoch": 0.243,
      "grad_norm": 0.65625,
      "grad_norm_var": 0.032597287495930986,
      "learning_rate": 0.0001,
      "loss": 0.4242,
      "loss/crossentropy": 2.817041516304016,
      "loss/fcd": 0.32421875,
      "loss/logits": 0.09995916113257408,
      "step": 243
    },
    {
      "epoch": 0.244,
      "grad_norm": 0.64453125,
      "grad_norm_var": 0.029933675130208334,
      "learning_rate": 0.0001,
      "loss": 0.4462,
      "loss/crossentropy": 2.806472897529602,
      "loss/fcd": 0.3408203125,
      "loss/logits": 0.10536561161279678,
      "step": 244
    },
    {
      "epoch": 0.245,
      "grad_norm": 0.55078125,
      "grad_norm_var": 0.03538106282552083,
      "learning_rate": 0.0001,
      "loss": 0.3982,
      "loss/crossentropy": 2.800318717956543,
      "loss/fcd": 0.3076171875,
      "loss/logits": 0.09055132418870926,
      "step": 245
    },
    {
      "epoch": 0.246,
      "grad_norm": 0.58984375,
      "grad_norm_var": 0.03576857248942057,
      "learning_rate": 0.0001,
      "loss": 0.402,
      "loss/crossentropy": 2.7864032983779907,
      "loss/fcd": 0.3095703125,
      "loss/logits": 0.09244050830602646,
      "step": 246
    },
    {
      "epoch": 0.247,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.03710880279541016,
      "learning_rate": 0.0001,
      "loss": 0.4454,
      "loss/crossentropy": 2.8640553951263428,
      "loss/fcd": 0.3388671875,
      "loss/logits": 0.10653172433376312,
      "step": 247
    },
    {
      "epoch": 0.248,
      "grad_norm": 0.68359375,
      "grad_norm_var": 0.037870025634765624,
      "learning_rate": 0.0001,
      "loss": 0.4301,
      "loss/crossentropy": 2.8637115955352783,
      "loss/fcd": 0.3271484375,
      "loss/logits": 0.10292865708470345,
      "step": 248
    },
    {
      "epoch": 0.249,
      "grad_norm": 0.80078125,
      "grad_norm_var": 0.03786462148030599,
      "learning_rate": 0.0001,
      "loss": 0.3903,
      "loss/crossentropy": 2.8047688007354736,
      "loss/fcd": 0.3046875,
      "loss/logits": 0.08557453379034996,
      "step": 249
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.57421875,
      "grad_norm_var": 0.021445719401041667,
      "learning_rate": 0.0001,
      "loss": 0.4041,
      "loss/crossentropy": 2.837961196899414,
      "loss/fcd": 0.3125,
      "loss/logits": 0.09161163493990898,
      "step": 250
    },
    {
      "epoch": 0.251,
      "grad_norm": 0.84765625,
      "grad_norm_var": 0.021980730692545573,
      "learning_rate": 0.0001,
      "loss": 0.5253,
      "loss/crossentropy": 2.749048352241516,
      "loss/fcd": 0.3818359375,
      "loss/logits": 0.1434529758989811,
      "step": 251
    },
    {
      "epoch": 0.252,
      "grad_norm": 0.60546875,
      "grad_norm_var": 0.019199371337890625,
      "learning_rate": 0.0001,
      "loss": 0.4083,
      "loss/crossentropy": 3.039583444595337,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.09189995378255844,
      "step": 252
    },
    {
      "epoch": 0.253,
      "grad_norm": 0.65234375,
      "grad_norm_var": 0.01318963368733724,
      "learning_rate": 0.0001,
      "loss": 0.4154,
      "loss/crossentropy": 2.83537220954895,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.09897411242127419,
      "step": 253
    },
    {
      "epoch": 0.254,
      "grad_norm": 0.67578125,
      "grad_norm_var": 0.008676083882649739,
      "learning_rate": 0.0001,
      "loss": 0.4261,
      "loss/crossentropy": 3.0131027698516846,
      "loss/fcd": 0.326171875,
      "loss/logits": 0.09990369901061058,
      "step": 254
    },
    {
      "epoch": 0.255,
      "grad_norm": 0.70703125,
      "grad_norm_var": 0.006815338134765625,
      "learning_rate": 0.0001,
      "loss": 0.4328,
      "loss/crossentropy": 2.93803608417511,
      "loss/fcd": 0.3291015625,
      "loss/logits": 0.10372024774551392,
      "step": 255
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.578125,
      "grad_norm_var": 0.006878089904785156,
      "learning_rate": 0.0001,
      "loss": 0.3991,
      "loss/crossentropy": 2.9447672367095947,
      "loss/fcd": 0.3056640625,
      "loss/logits": 0.09344978258013725,
      "step": 256
    },
    {
      "epoch": 0.257,
      "grad_norm": 0.78125,
      "grad_norm_var": 0.00728600819905599,
      "learning_rate": 0.0001,
      "loss": 0.4649,
      "loss/crossentropy": 2.7877269983291626,
      "loss/fcd": 0.3447265625,
      "loss/logits": 0.12016481161117554,
      "step": 257
    },
    {
      "epoch": 0.258,
      "grad_norm": 1.0625,
      "grad_norm_var": 0.016978963216145834,
      "learning_rate": 0.0001,
      "loss": 0.4578,
      "loss/crossentropy": 2.8194340467453003,
      "loss/fcd": 0.3515625,
      "loss/logits": 0.10628331080079079,
      "step": 258
    },
    {
      "epoch": 0.259,
      "grad_norm": 0.59765625,
      "grad_norm_var": 0.01746056874593099,
      "learning_rate": 0.0001,
      "loss": 0.4067,
      "loss/crossentropy": 2.864375114440918,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.09034289047122002,
      "step": 259
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6171875,
      "grad_norm_var": 0.017661285400390626,
      "learning_rate": 0.0001,
      "loss": 0.3902,
      "loss/crossentropy": 2.8615986108779907,
      "loss/fcd": 0.302734375,
      "loss/logits": 0.08747856691479683,
      "step": 260
    },
    {
      "epoch": 0.261,
      "grad_norm": 0.64453125,
      "grad_norm_var": 0.016532135009765626,
      "learning_rate": 0.0001,
      "loss": 0.4094,
      "loss/crossentropy": 2.824509024620056,
      "loss/fcd": 0.3115234375,
      "loss/logits": 0.09783903509378433,
      "step": 261
    },
    {
      "epoch": 0.262,
      "grad_norm": 0.6328125,
      "grad_norm_var": 0.016068458557128906,
      "learning_rate": 0.0001,
      "loss": 0.4327,
      "loss/crossentropy": 2.802370309829712,
      "loss/fcd": 0.3310546875,
      "loss/logits": 0.10163304209709167,
      "step": 262
    },
    {
      "epoch": 0.263,
      "grad_norm": 0.7109375,
      "grad_norm_var": 0.015849812825520834,
      "learning_rate": 0.0001,
      "loss": 0.4364,
      "loss/crossentropy": 2.9660686254501343,
      "loss/fcd": 0.328125,
      "loss/logits": 0.10826770216226578,
      "step": 263
    },
    {
      "epoch": 0.264,
      "grad_norm": 0.640625,
      "grad_norm_var": 0.01604913075764974,
      "learning_rate": 0.0001,
      "loss": 0.4306,
      "loss/crossentropy": 2.970041871070862,
      "loss/fcd": 0.328125,
      "loss/logits": 0.10252038761973381,
      "step": 264
    },
    {
      "epoch": 0.265,
      "grad_norm": 0.57421875,
      "grad_norm_var": 0.016078631083170574,
      "learning_rate": 0.0001,
      "loss": 0.3905,
      "loss/crossentropy": 2.7347593307495117,
      "loss/fcd": 0.30078125,
      "loss/logits": 0.08967656269669533,
      "step": 265
    },
    {
      "epoch": 0.266,
      "grad_norm": 0.62890625,
      "grad_norm_var": 0.015484046936035157,
      "learning_rate": 0.0001,
      "loss": 0.3998,
      "loss/crossentropy": 2.6828304529190063,
      "loss/fcd": 0.3115234375,
      "loss/logits": 0.08822760730981827,
      "step": 266
    },
    {
      "epoch": 0.267,
      "grad_norm": 0.73046875,
      "grad_norm_var": 0.013797950744628907,
      "learning_rate": 0.0001,
      "loss": 0.5116,
      "loss/crossentropy": 3.124196410179138,
      "loss/fcd": 0.376953125,
      "loss/logits": 0.13468829542398453,
      "step": 267
    },
    {
      "epoch": 0.268,
      "grad_norm": 0.73828125,
      "grad_norm_var": 0.013625017801920573,
      "learning_rate": 0.0001,
      "loss": 0.4071,
      "loss/crossentropy": 2.963875889778137,
      "loss/fcd": 0.3134765625,
      "loss/logits": 0.09359101951122284,
      "step": 268
    },
    {
      "epoch": 0.269,
      "grad_norm": 0.6640625,
      "grad_norm_var": 0.013581339518229167,
      "learning_rate": 0.0001,
      "loss": 0.4062,
      "loss/crossentropy": 2.852897524833679,
      "loss/fcd": 0.3134765625,
      "loss/logits": 0.09275222197175026,
      "step": 269
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.76171875,
      "grad_norm_var": 0.013919830322265625,
      "learning_rate": 0.0001,
      "loss": 0.4204,
      "loss/crossentropy": 2.8601616621017456,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09812301397323608,
      "step": 270
    },
    {
      "epoch": 0.271,
      "grad_norm": 0.7578125,
      "grad_norm_var": 0.014183489481608073,
      "learning_rate": 0.0001,
      "loss": 0.4126,
      "loss/crossentropy": 2.713176965713501,
      "loss/fcd": 0.3154296875,
      "loss/logits": 0.09714118391275406,
      "step": 271
    },
    {
      "epoch": 0.272,
      "grad_norm": 0.671875,
      "grad_norm_var": 0.013271013895670572,
      "learning_rate": 0.0001,
      "loss": 0.4449,
      "loss/crossentropy": 3.02414071559906,
      "loss/fcd": 0.337890625,
      "loss/logits": 0.10696037858724594,
      "step": 272
    },
    {
      "epoch": 0.273,
      "grad_norm": 0.6640625,
      "grad_norm_var": 0.012874285380045572,
      "learning_rate": 0.0001,
      "loss": 0.422,
      "loss/crossentropy": 2.842344880104065,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.09679682552814484,
      "step": 273
    },
    {
      "epoch": 0.274,
      "grad_norm": 0.83203125,
      "grad_norm_var": 0.004858144124348958,
      "learning_rate": 0.0001,
      "loss": 0.4451,
      "loss/crossentropy": 2.876818299293518,
      "loss/fcd": 0.33984375,
      "loss/logits": 0.10525127500295639,
      "step": 274
    },
    {
      "epoch": 0.275,
      "grad_norm": 0.91796875,
      "grad_norm_var": 0.0077880859375,
      "learning_rate": 0.0001,
      "loss": 0.4179,
      "loss/crossentropy": 2.785101294517517,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09567511081695557,
      "step": 275
    },
    {
      "epoch": 0.276,
      "grad_norm": 0.734375,
      "grad_norm_var": 0.007364654541015625,
      "learning_rate": 0.0001,
      "loss": 0.4274,
      "loss/crossentropy": 2.839262843132019,
      "loss/fcd": 0.33203125,
      "loss/logits": 0.09535057097673416,
      "step": 276
    },
    {
      "epoch": 0.277,
      "grad_norm": 0.68359375,
      "grad_norm_var": 0.0071370442708333336,
      "learning_rate": 0.0001,
      "loss": 0.4264,
      "loss/crossentropy": 2.8711353540420532,
      "loss/fcd": 0.326171875,
      "loss/logits": 0.10018761828541756,
      "step": 277
    },
    {
      "epoch": 0.278,
      "grad_norm": 0.6328125,
      "grad_norm_var": 0.0071370442708333336,
      "learning_rate": 0.0001,
      "loss": 0.4129,
      "loss/crossentropy": 2.871408224105835,
      "loss/fcd": 0.3173828125,
      "loss/logits": 0.09553812071681023,
      "step": 278
    },
    {
      "epoch": 0.279,
      "grad_norm": 0.62890625,
      "grad_norm_var": 0.007536252339680989,
      "learning_rate": 0.0001,
      "loss": 0.4575,
      "loss/crossentropy": 2.876176953315735,
      "loss/fcd": 0.341796875,
      "loss/logits": 0.11570525541901588,
      "step": 279
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.61328125,
      "grad_norm_var": 0.007813517252604167,
      "learning_rate": 0.0001,
      "loss": 0.4354,
      "loss/crossentropy": 2.8264540433883667,
      "loss/fcd": 0.3330078125,
      "loss/logits": 0.1023903377354145,
      "step": 280
    },
    {
      "epoch": 0.281,
      "grad_norm": 0.69140625,
      "grad_norm_var": 0.006672922770182292,
      "learning_rate": 0.0001,
      "loss": 0.3931,
      "loss/crossentropy": 2.744624972343445,
      "loss/fcd": 0.3046875,
      "loss/logits": 0.08837807923555374,
      "step": 281
    },
    {
      "epoch": 0.282,
      "grad_norm": 0.7109375,
      "grad_norm_var": 0.006212298075358073,
      "learning_rate": 0.0001,
      "loss": 0.4133,
      "loss/crossentropy": 2.892166256904602,
      "loss/fcd": 0.314453125,
      "loss/logits": 0.09885082766413689,
      "step": 282
    },
    {
      "epoch": 0.283,
      "grad_norm": 0.7578125,
      "grad_norm_var": 0.0063168843587239586,
      "learning_rate": 0.0001,
      "loss": 0.4083,
      "loss/crossentropy": 2.8837363719940186,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.0918973907828331,
      "step": 283
    },
    {
      "epoch": 0.284,
      "grad_norm": 0.67578125,
      "grad_norm_var": 0.0063779195149739586,
      "learning_rate": 0.0001,
      "loss": 0.4308,
      "loss/crossentropy": 2.7492740154266357,
      "loss/fcd": 0.3359375,
      "loss/logits": 0.09483079984784126,
      "step": 284
    },
    {
      "epoch": 0.285,
      "grad_norm": 0.55078125,
      "grad_norm_var": 0.00791009267171224,
      "learning_rate": 0.0001,
      "loss": 0.4066,
      "loss/crossentropy": 2.8545846939086914,
      "loss/fcd": 0.314453125,
      "loss/logits": 0.09213941171765327,
      "step": 285
    },
    {
      "epoch": 0.286,
      "grad_norm": 0.61328125,
      "grad_norm_var": 0.008171017964680989,
      "learning_rate": 0.0001,
      "loss": 0.4126,
      "loss/crossentropy": 2.67995822429657,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.08735696226358414,
      "step": 286
    },
    {
      "epoch": 0.287,
      "grad_norm": 0.80078125,
      "grad_norm_var": 0.008640289306640625,
      "learning_rate": 0.0001,
      "loss": 0.3921,
      "loss/crossentropy": 2.8901337385177612,
      "loss/fcd": 0.3056640625,
      "loss/logits": 0.08640312403440475,
      "step": 287
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.609375,
      "grad_norm_var": 0.009108225504557291,
      "learning_rate": 0.0001,
      "loss": 0.4164,
      "loss/crossentropy": 2.8247963190078735,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.10003437474370003,
      "step": 288
    },
    {
      "epoch": 0.289,
      "grad_norm": 0.5546875,
      "grad_norm_var": 0.010304514567057292,
      "learning_rate": 0.0001,
      "loss": 0.4182,
      "loss/crossentropy": 2.943819999694824,
      "loss/fcd": 0.3232421875,
      "loss/logits": 0.09490966796875,
      "step": 289
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.515625,
      "grad_norm_var": 0.010484759012858074,
      "learning_rate": 0.0001,
      "loss": 0.3826,
      "loss/crossentropy": 2.9070965051651,
      "loss/fcd": 0.2939453125,
      "loss/logits": 0.0886867605149746,
      "step": 290
    },
    {
      "epoch": 0.291,
      "grad_norm": 0.62109375,
      "grad_norm_var": 0.006107012430826823,
      "learning_rate": 0.0001,
      "loss": 0.4199,
      "loss/crossentropy": 2.8656623363494873,
      "loss/fcd": 0.318359375,
      "loss/logits": 0.10157756507396698,
      "step": 291
    },
    {
      "epoch": 0.292,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.005597877502441406,
      "learning_rate": 0.0001,
      "loss": 0.3961,
      "loss/crossentropy": 2.8809529542922974,
      "loss/fcd": 0.3095703125,
      "loss/logits": 0.08648988232016563,
      "step": 292
    },
    {
      "epoch": 0.293,
      "grad_norm": 0.765625,
      "grad_norm_var": 0.0064483642578125,
      "learning_rate": 0.0001,
      "loss": 0.383,
      "loss/crossentropy": 2.6984453201293945,
      "loss/fcd": 0.3017578125,
      "loss/logits": 0.08123517036437988,
      "step": 293
    },
    {
      "epoch": 0.294,
      "grad_norm": 0.734375,
      "grad_norm_var": 0.006868235270182292,
      "learning_rate": 0.0001,
      "loss": 0.4093,
      "loss/crossentropy": 2.7411580085754395,
      "loss/fcd": 0.3154296875,
      "loss/logits": 0.09388710558414459,
      "step": 294
    },
    {
      "epoch": 0.295,
      "grad_norm": 0.64453125,
      "grad_norm_var": 0.006827545166015625,
      "learning_rate": 0.0001,
      "loss": 0.4139,
      "loss/crossentropy": 2.761058211326599,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.09746293723583221,
      "step": 295
    },
    {
      "epoch": 0.296,
      "grad_norm": 0.8671875,
      "grad_norm_var": 0.009385617574055989,
      "learning_rate": 0.0001,
      "loss": 0.4587,
      "loss/crossentropy": 2.882654905319214,
      "loss/fcd": 0.345703125,
      "loss/logits": 0.11303438991308212,
      "step": 296
    },
    {
      "epoch": 0.297,
      "grad_norm": 0.75390625,
      "grad_norm_var": 0.009786415100097656,
      "learning_rate": 0.0001,
      "loss": 0.4656,
      "loss/crossentropy": 2.759294867515564,
      "loss/fcd": 0.353515625,
      "loss/logits": 0.11211657896637917,
      "step": 297
    },
    {
      "epoch": 0.298,
      "grad_norm": 0.68359375,
      "grad_norm_var": 0.0097076416015625,
      "learning_rate": 0.0001,
      "loss": 0.4354,
      "loss/crossentropy": 2.732961416244507,
      "loss/fcd": 0.3349609375,
      "loss/logits": 0.10048427432775497,
      "step": 298
    },
    {
      "epoch": 0.299,
      "grad_norm": 0.6171875,
      "grad_norm_var": 0.00938720703125,
      "learning_rate": 0.0001,
      "loss": 0.4397,
      "loss/crossentropy": 2.8611546754837036,
      "loss/fcd": 0.337890625,
      "loss/logits": 0.10177960991859436,
      "step": 299
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8203125,
      "grad_norm_var": 0.010880978902180989,
      "learning_rate": 0.0001,
      "loss": 0.4757,
      "loss/crossentropy": 2.784830689430237,
      "loss/fcd": 0.3515625,
      "loss/logits": 0.1241227462887764,
      "step": 300
    },
    {
      "epoch": 0.301,
      "grad_norm": 0.65234375,
      "grad_norm_var": 0.009842872619628906,
      "learning_rate": 0.0001,
      "loss": 0.429,
      "loss/crossentropy": 2.9347182512283325,
      "loss/fcd": 0.328125,
      "loss/logits": 0.10089103132486343,
      "step": 301
    },
    {
      "epoch": 0.302,
      "grad_norm": 0.6875,
      "grad_norm_var": 0.009513092041015626,
      "learning_rate": 0.0001,
      "loss": 0.4071,
      "loss/crossentropy": 2.816414952278137,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.09071046859025955,
      "step": 302
    },
    {
      "epoch": 0.303,
      "grad_norm": 0.640625,
      "grad_norm_var": 0.008665911356608073,
      "learning_rate": 0.0001,
      "loss": 0.4173,
      "loss/crossentropy": 2.8991583585739136,
      "loss/fcd": 0.318359375,
      "loss/logits": 0.09893063083291054,
      "step": 303
    },
    {
      "epoch": 0.304,
      "grad_norm": 0.5546875,
      "grad_norm_var": 0.009338823954264323,
      "learning_rate": 0.0001,
      "loss": 0.4005,
      "loss/crossentropy": 2.894463539123535,
      "loss/fcd": 0.3125,
      "loss/logits": 0.08798391744494438,
      "step": 304
    },
    {
      "epoch": 0.305,
      "grad_norm": 0.62109375,
      "grad_norm_var": 0.008570353190104166,
      "learning_rate": 0.0001,
      "loss": 0.4035,
      "loss/crossentropy": 2.914915442466736,
      "loss/fcd": 0.3125,
      "loss/logits": 0.09100022539496422,
      "step": 305
    },
    {
      "epoch": 0.306,
      "grad_norm": 0.7109375,
      "grad_norm_var": 0.006758371988932292,
      "learning_rate": 0.0001,
      "loss": 0.4032,
      "loss/crossentropy": 2.9412468671798706,
      "loss/fcd": 0.314453125,
      "loss/logits": 0.08878887072205544,
      "step": 306
    },
    {
      "epoch": 0.307,
      "grad_norm": 0.64453125,
      "grad_norm_var": 0.0065806070963541664,
      "learning_rate": 0.0001,
      "loss": 0.3907,
      "loss/crossentropy": 2.829257607460022,
      "loss/fcd": 0.3056640625,
      "loss/logits": 0.08507547527551651,
      "step": 307
    },
    {
      "epoch": 0.308,
      "grad_norm": 0.6015625,
      "grad_norm_var": 0.006980387369791666,
      "learning_rate": 0.0001,
      "loss": 0.3899,
      "loss/crossentropy": 2.7131818532943726,
      "loss/fcd": 0.30078125,
      "loss/logits": 0.08915158361196518,
      "step": 308
    },
    {
      "epoch": 0.309,
      "grad_norm": 0.57421875,
      "grad_norm_var": 0.007276344299316406,
      "learning_rate": 0.0001,
      "loss": 0.3931,
      "loss/crossentropy": 2.820746898651123,
      "loss/fcd": 0.30859375,
      "loss/logits": 0.0845017246901989,
      "step": 309
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.65625,
      "grad_norm_var": 0.0070449193318684895,
      "learning_rate": 0.0001,
      "loss": 0.4006,
      "loss/crossentropy": 2.7722835540771484,
      "loss/fcd": 0.3115234375,
      "loss/logits": 0.08902696147561073,
      "step": 310
    },
    {
      "epoch": 0.311,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.007075945536295573,
      "learning_rate": 0.0001,
      "loss": 0.4239,
      "loss/crossentropy": 2.867286205291748,
      "loss/fcd": 0.3271484375,
      "loss/logits": 0.09677955880761147,
      "step": 311
    },
    {
      "epoch": 0.312,
      "grad_norm": 0.734375,
      "grad_norm_var": 0.004689470926920573,
      "learning_rate": 0.0001,
      "loss": 0.4179,
      "loss/crossentropy": 2.7744059562683105,
      "loss/fcd": 0.318359375,
      "loss/logits": 0.09955108538269997,
      "step": 312
    },
    {
      "epoch": 0.313,
      "grad_norm": 0.66796875,
      "grad_norm_var": 0.004096412658691406,
      "learning_rate": 0.0001,
      "loss": 0.4488,
      "loss/crossentropy": 2.7806670665740967,
      "loss/fcd": 0.3427734375,
      "loss/logits": 0.10600494965910912,
      "step": 313
    },
    {
      "epoch": 0.314,
      "grad_norm": 0.56640625,
      "grad_norm_var": 0.004531288146972656,
      "learning_rate": 0.0001,
      "loss": 0.3871,
      "loss/crossentropy": 2.8486790657043457,
      "loss/fcd": 0.3037109375,
      "loss/logits": 0.08334493264555931,
      "step": 314
    },
    {
      "epoch": 0.315,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.004471842447916667,
      "learning_rate": 0.0001,
      "loss": 0.4111,
      "loss/crossentropy": 2.9870445728302,
      "loss/fcd": 0.318359375,
      "loss/logits": 0.0927395410835743,
      "step": 315
    },
    {
      "epoch": 0.316,
      "grad_norm": 0.6015625,
      "grad_norm_var": 0.002506510416666667,
      "learning_rate": 0.0001,
      "loss": 0.3772,
      "loss/crossentropy": 2.8333510160446167,
      "loss/fcd": 0.2958984375,
      "loss/logits": 0.0812879391014576,
      "step": 316
    },
    {
      "epoch": 0.317,
      "grad_norm": 0.6875,
      "grad_norm_var": 0.002657000223795573,
      "learning_rate": 0.0001,
      "loss": 0.4304,
      "loss/crossentropy": 2.783381938934326,
      "loss/fcd": 0.330078125,
      "loss/logits": 0.10034845769405365,
      "step": 317
    },
    {
      "epoch": 0.318,
      "grad_norm": 0.60546875,
      "grad_norm_var": 0.0025461832682291668,
      "learning_rate": 0.0001,
      "loss": 0.4136,
      "loss/crossentropy": 2.886239528656006,
      "loss/fcd": 0.3173828125,
      "loss/logits": 0.09618546813726425,
      "step": 318
    },
    {
      "epoch": 0.319,
      "grad_norm": 0.625,
      "grad_norm_var": 0.0025472005208333334,
      "learning_rate": 0.0001,
      "loss": 0.4161,
      "loss/crossentropy": 2.9301581382751465,
      "loss/fcd": 0.32421875,
      "loss/logits": 0.09190791472792625,
      "step": 319
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.546875,
      "grad_norm_var": 0.002632395426432292,
      "learning_rate": 0.0001,
      "loss": 0.3861,
      "loss/crossentropy": 2.813088059425354,
      "loss/fcd": 0.30078125,
      "loss/logits": 0.08532163128256798,
      "step": 320
    },
    {
      "epoch": 0.321,
      "grad_norm": 0.52734375,
      "grad_norm_var": 0.0033220926920572917,
      "learning_rate": 0.0001,
      "loss": 0.3855,
      "loss/crossentropy": 2.8616257905960083,
      "loss/fcd": 0.298828125,
      "loss/logits": 0.08671489730477333,
      "step": 321
    },
    {
      "epoch": 0.322,
      "grad_norm": 0.58984375,
      "grad_norm_var": 0.0028746922810872397,
      "learning_rate": 0.0001,
      "loss": 0.4115,
      "loss/crossentropy": 2.859944701194763,
      "loss/fcd": 0.3193359375,
      "loss/logits": 0.09216945245862007,
      "step": 322
    },
    {
      "epoch": 0.323,
      "grad_norm": 0.65625,
      "grad_norm_var": 0.0029233296712239585,
      "learning_rate": 0.0001,
      "loss": 0.4164,
      "loss/crossentropy": 2.8817009925842285,
      "loss/fcd": 0.3193359375,
      "loss/logits": 0.09702347591519356,
      "step": 323
    },
    {
      "epoch": 0.324,
      "grad_norm": 0.75,
      "grad_norm_var": 0.00394287109375,
      "learning_rate": 0.0001,
      "loss": 0.4215,
      "loss/crossentropy": 2.9102847576141357,
      "loss/fcd": 0.330078125,
      "loss/logits": 0.09144152700901031,
      "step": 324
    },
    {
      "epoch": 0.325,
      "grad_norm": 0.89453125,
      "grad_norm_var": 0.008019765218098959,
      "learning_rate": 0.0001,
      "loss": 0.3933,
      "loss/crossentropy": 2.7305492162704468,
      "loss/fcd": 0.3056640625,
      "loss/logits": 0.08765168115496635,
      "step": 325
    },
    {
      "epoch": 0.326,
      "grad_norm": 0.796875,
      "grad_norm_var": 0.009393056233723959,
      "learning_rate": 0.0001,
      "loss": 0.3895,
      "loss/crossentropy": 2.8607096672058105,
      "loss/fcd": 0.3017578125,
      "loss/logits": 0.08771565556526184,
      "step": 326
    },
    {
      "epoch": 0.327,
      "grad_norm": 0.6328125,
      "grad_norm_var": 0.009404945373535156,
      "learning_rate": 0.0001,
      "loss": 0.4095,
      "loss/crossentropy": 2.9553240537643433,
      "loss/fcd": 0.3212890625,
      "loss/logits": 0.08823421597480774,
      "step": 327
    },
    {
      "epoch": 0.328,
      "grad_norm": 0.6953125,
      "grad_norm_var": 0.009099769592285156,
      "learning_rate": 0.0001,
      "loss": 0.4631,
      "loss/crossentropy": 2.838402509689331,
      "loss/fcd": 0.34765625,
      "loss/logits": 0.1154676228761673,
      "step": 328
    },
    {
      "epoch": 0.329,
      "grad_norm": 1.75,
      "grad_norm_var": 0.08414103190104166,
      "learning_rate": 0.0001,
      "loss": 0.4394,
      "loss/crossentropy": 2.756170868873596,
      "loss/fcd": 0.314453125,
      "loss/logits": 0.1249246709048748,
      "step": 329
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.765625,
      "grad_norm_var": 0.08247114817301432,
      "learning_rate": 0.0001,
      "loss": 0.422,
      "loss/crossentropy": 2.8995450735092163,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.10559750720858574,
      "step": 330
    },
    {
      "epoch": 0.331,
      "grad_norm": 0.796875,
      "grad_norm_var": 0.0819732666015625,
      "learning_rate": 0.0001,
      "loss": 0.4665,
      "loss/crossentropy": 2.8580288887023926,
      "loss/fcd": 0.341796875,
      "loss/logits": 0.12471328675746918,
      "step": 331
    },
    {
      "epoch": 0.332,
      "grad_norm": 0.796875,
      "grad_norm_var": 0.08061904907226562,
      "learning_rate": 0.0001,
      "loss": 0.4448,
      "loss/crossentropy": 2.831050157546997,
      "loss/fcd": 0.330078125,
      "loss/logits": 0.11475277319550514,
      "step": 332
    },
    {
      "epoch": 0.333,
      "grad_norm": 0.7421875,
      "grad_norm_var": 0.08029683430989583,
      "learning_rate": 0.0001,
      "loss": 0.4326,
      "loss/crossentropy": 2.8131872415542603,
      "loss/fcd": 0.328125,
      "loss/logits": 0.10444386303424835,
      "step": 333
    },
    {
      "epoch": 0.334,
      "grad_norm": 0.6875,
      "grad_norm_var": 0.07901910146077475,
      "learning_rate": 0.0001,
      "loss": 0.4318,
      "loss/crossentropy": 2.822066903114319,
      "loss/fcd": 0.32421875,
      "loss/logits": 0.10761934518814087,
      "step": 334
    },
    {
      "epoch": 0.335,
      "grad_norm": 0.79296875,
      "grad_norm_var": 0.0776275634765625,
      "learning_rate": 0.0001,
      "loss": 0.4357,
      "loss/crossentropy": 2.8042834997177124,
      "loss/fcd": 0.3359375,
      "loss/logits": 0.09971516206860542,
      "step": 335
    },
    {
      "epoch": 0.336,
      "grad_norm": 0.6328125,
      "grad_norm_var": 0.07545954386393229,
      "learning_rate": 0.0001,
      "loss": 0.4202,
      "loss/crossentropy": 2.8359432220458984,
      "loss/fcd": 0.32421875,
      "loss/logits": 0.0959857665002346,
      "step": 336
    },
    {
      "epoch": 0.337,
      "grad_norm": 0.57421875,
      "grad_norm_var": 0.0740069071451823,
      "learning_rate": 0.0001,
      "loss": 0.4091,
      "loss/crossentropy": 2.869974374771118,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.0926552303135395,
      "step": 337
    },
    {
      "epoch": 0.338,
      "grad_norm": 0.58984375,
      "grad_norm_var": 0.0740069071451823,
      "learning_rate": 0.0001,
      "loss": 0.3979,
      "loss/crossentropy": 2.807898163795471,
      "loss/fcd": 0.306640625,
      "loss/logits": 0.09129861742258072,
      "step": 338
    },
    {
      "epoch": 0.339,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.07436517079671225,
      "learning_rate": 0.0001,
      "loss": 0.4291,
      "loss/crossentropy": 2.8850966691970825,
      "loss/fcd": 0.328125,
      "loss/logits": 0.10096083208918571,
      "step": 339
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7265625,
      "grad_norm_var": 0.07450402577718099,
      "learning_rate": 0.0001,
      "loss": 0.4518,
      "loss/crossentropy": 2.9171935319900513,
      "loss/fcd": 0.34375,
      "loss/logits": 0.10800491645932198,
      "step": 340
    },
    {
      "epoch": 0.341,
      "grad_norm": 0.703125,
      "grad_norm_var": 0.07392145792643229,
      "learning_rate": 0.0001,
      "loss": 0.4566,
      "loss/crossentropy": 2.8163551092147827,
      "loss/fcd": 0.34375,
      "loss/logits": 0.11288850009441376,
      "step": 341
    },
    {
      "epoch": 0.342,
      "grad_norm": 0.55859375,
      "grad_norm_var": 0.07661685943603516,
      "learning_rate": 0.0001,
      "loss": 0.3953,
      "loss/crossentropy": 2.788830041885376,
      "loss/fcd": 0.3056640625,
      "loss/logits": 0.08966437354683876,
      "step": 342
    },
    {
      "epoch": 0.343,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.07655410766601563,
      "learning_rate": 0.0001,
      "loss": 0.423,
      "loss/crossentropy": 2.919311761856079,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.0978211909532547,
      "step": 343
    },
    {
      "epoch": 0.344,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.07706680297851562,
      "learning_rate": 0.0001,
      "loss": 0.404,
      "loss/crossentropy": 2.8302316665649414,
      "loss/fcd": 0.3154296875,
      "loss/logits": 0.08858633041381836,
      "step": 344
    },
    {
      "epoch": 0.345,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.006453895568847656,
      "learning_rate": 0.0001,
      "loss": 0.4177,
      "loss/crossentropy": 2.918962240219116,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09547547250986099,
      "step": 345
    },
    {
      "epoch": 0.346,
      "grad_norm": 0.55859375,
      "grad_norm_var": 0.00684814453125,
      "learning_rate": 0.0001,
      "loss": 0.4338,
      "loss/crossentropy": 2.921096920967102,
      "loss/fcd": 0.33203125,
      "loss/logits": 0.10181862115859985,
      "step": 346
    },
    {
      "epoch": 0.347,
      "grad_norm": 0.58203125,
      "grad_norm_var": 0.0060963312784830725,
      "learning_rate": 0.0001,
      "loss": 0.4104,
      "loss/crossentropy": 2.638503074645996,
      "loss/fcd": 0.3173828125,
      "loss/logits": 0.09305207803845406,
      "step": 347
    },
    {
      "epoch": 0.348,
      "grad_norm": 0.73828125,
      "grad_norm_var": 0.005214182535807291,
      "learning_rate": 0.0001,
      "loss": 0.4034,
      "loss/crossentropy": 2.8353854417800903,
      "loss/fcd": 0.3125,
      "loss/logits": 0.09088730812072754,
      "step": 348
    },
    {
      "epoch": 0.349,
      "grad_norm": 0.7265625,
      "grad_norm_var": 0.0050432840983072914,
      "learning_rate": 0.0001,
      "loss": 0.4215,
      "loss/crossentropy": 2.875319242477417,
      "loss/fcd": 0.3271484375,
      "loss/logits": 0.09431769698858261,
      "step": 349
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.58203125,
      "grad_norm_var": 0.005237261454264323,
      "learning_rate": 0.0001,
      "loss": 0.4012,
      "loss/crossentropy": 2.716060757637024,
      "loss/fcd": 0.314453125,
      "loss/logits": 0.08677634596824646,
      "step": 350
    },
    {
      "epoch": 0.351,
      "grad_norm": 0.61328125,
      "grad_norm_var": 0.0037164688110351562,
      "learning_rate": 0.0001,
      "loss": 0.4161,
      "loss/crossentropy": 2.9706015586853027,
      "loss/fcd": 0.3271484375,
      "loss/logits": 0.08898995071649551,
      "step": 351
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.72265625,
      "grad_norm_var": 0.004206339518229167,
      "learning_rate": 0.0001,
      "loss": 0.4454,
      "loss/crossentropy": 2.9135972261428833,
      "loss/fcd": 0.3408203125,
      "loss/logits": 0.10461591929197311,
      "step": 352
    },
    {
      "epoch": 0.353,
      "grad_norm": 0.71875,
      "grad_norm_var": 0.00425103505452474,
      "learning_rate": 0.0001,
      "loss": 0.4235,
      "loss/crossentropy": 2.9371761083602905,
      "loss/fcd": 0.326171875,
      "loss/logits": 0.09735563024878502,
      "step": 353
    },
    {
      "epoch": 0.354,
      "grad_norm": 0.69140625,
      "grad_norm_var": 0.004098955790201823,
      "learning_rate": 0.0001,
      "loss": 0.4207,
      "loss/crossentropy": 2.9942837953567505,
      "loss/fcd": 0.326171875,
      "loss/logits": 0.09450989216566086,
      "step": 354
    },
    {
      "epoch": 0.355,
      "grad_norm": 0.6171875,
      "grad_norm_var": 0.004170481363932292,
      "learning_rate": 0.0001,
      "loss": 0.4433,
      "loss/crossentropy": 2.820738196372986,
      "loss/fcd": 0.3349609375,
      "loss/logits": 0.10836119577288628,
      "step": 355
    },
    {
      "epoch": 0.356,
      "grad_norm": 0.65625,
      "grad_norm_var": 0.0037974039713541668,
      "learning_rate": 0.0001,
      "loss": 0.4336,
      "loss/crossentropy": 2.8830113410949707,
      "loss/fcd": 0.3291015625,
      "loss/logits": 0.10447230935096741,
      "step": 356
    },
    {
      "epoch": 0.357,
      "grad_norm": 0.87890625,
      "grad_norm_var": 0.006987444559733073,
      "learning_rate": 0.0001,
      "loss": 0.4988,
      "loss/crossentropy": 2.950507402420044,
      "loss/fcd": 0.365234375,
      "loss/logits": 0.13359853625297546,
      "step": 357
    },
    {
      "epoch": 0.358,
      "grad_norm": 0.6640625,
      "grad_norm_var": 0.006251017252604167,
      "learning_rate": 0.0001,
      "loss": 0.4254,
      "loss/crossentropy": 3.050377130508423,
      "loss/fcd": 0.328125,
      "loss/logits": 0.09723382443189621,
      "step": 358
    },
    {
      "epoch": 0.359,
      "grad_norm": 0.61328125,
      "grad_norm_var": 0.006379954020182292,
      "learning_rate": 0.0001,
      "loss": 0.4174,
      "loss/crossentropy": 2.872576355934143,
      "loss/fcd": 0.3173828125,
      "loss/logits": 0.09998740255832672,
      "step": 359
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.64453125,
      "grad_norm_var": 0.006389808654785156,
      "learning_rate": 0.0001,
      "loss": 0.4129,
      "loss/crossentropy": 2.636136770248413,
      "loss/fcd": 0.3193359375,
      "loss/logits": 0.09356288239359856,
      "step": 360
    },
    {
      "epoch": 0.361,
      "grad_norm": 0.61328125,
      "grad_norm_var": 0.006513404846191406,
      "learning_rate": 0.0001,
      "loss": 0.4126,
      "loss/crossentropy": 2.9116278886795044,
      "loss/fcd": 0.3173828125,
      "loss/logits": 0.09523866325616837,
      "step": 361
    },
    {
      "epoch": 0.362,
      "grad_norm": 0.62109375,
      "grad_norm_var": 0.005880673726399739,
      "learning_rate": 0.0001,
      "loss": 0.4159,
      "loss/crossentropy": 2.89151668548584,
      "loss/fcd": 0.3173828125,
      "loss/logits": 0.09850744158029556,
      "step": 362
    },
    {
      "epoch": 0.363,
      "grad_norm": 0.875,
      "grad_norm_var": 0.007897694905598959,
      "learning_rate": 0.0001,
      "loss": 0.4513,
      "loss/crossentropy": 2.9357181787490845,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.11434905976057053,
      "step": 363
    },
    {
      "epoch": 0.364,
      "grad_norm": 0.74609375,
      "grad_norm_var": 0.0079559326171875,
      "learning_rate": 0.0001,
      "loss": 0.4275,
      "loss/crossentropy": 3.013663649559021,
      "loss/fcd": 0.330078125,
      "loss/logits": 0.0974624715745449,
      "step": 364
    },
    {
      "epoch": 0.365,
      "grad_norm": 0.7265625,
      "grad_norm_var": 0.0079559326171875,
      "learning_rate": 0.0001,
      "loss": 0.4141,
      "loss/crossentropy": 2.7465943098068237,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09180082008242607,
      "step": 365
    },
    {
      "epoch": 0.366,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.007380930582682291,
      "learning_rate": 0.0001,
      "loss": 0.4263,
      "loss/crossentropy": 2.9539248943328857,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.10109733417630196,
      "step": 366
    },
    {
      "epoch": 0.367,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.0071756998697916664,
      "learning_rate": 0.0001,
      "loss": 0.4324,
      "loss/crossentropy": 2.89021635055542,
      "loss/fcd": 0.33203125,
      "loss/logits": 0.10038217157125473,
      "step": 367
    },
    {
      "epoch": 0.368,
      "grad_norm": 0.5703125,
      "grad_norm_var": 0.007991472880045572,
      "learning_rate": 0.0001,
      "loss": 0.4051,
      "loss/crossentropy": 2.7650424242019653,
      "loss/fcd": 0.3125,
      "loss/logits": 0.09255194291472435,
      "step": 368
    },
    {
      "epoch": 0.369,
      "grad_norm": 0.53125,
      "grad_norm_var": 0.009267107645670573,
      "learning_rate": 0.0001,
      "loss": 0.3949,
      "loss/crossentropy": 2.841827392578125,
      "loss/fcd": 0.3076171875,
      "loss/logits": 0.0872911661863327,
      "step": 369
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.50390625,
      "grad_norm_var": 0.010933367411295573,
      "learning_rate": 0.0001,
      "loss": 0.3829,
      "loss/crossentropy": 2.776681661605835,
      "loss/fcd": 0.298828125,
      "loss/logits": 0.08403456583619118,
      "step": 370
    },
    {
      "epoch": 0.371,
      "grad_norm": 0.61328125,
      "grad_norm_var": 0.010955810546875,
      "learning_rate": 0.0001,
      "loss": 0.4569,
      "loss/crossentropy": 2.904550552368164,
      "loss/fcd": 0.349609375,
      "loss/logits": 0.10726456344127655,
      "step": 371
    },
    {
      "epoch": 0.372,
      "grad_norm": 0.625,
      "grad_norm_var": 0.011024983723958333,
      "learning_rate": 0.0001,
      "loss": 0.4291,
      "loss/crossentropy": 2.7700815200805664,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.09217642992734909,
      "step": 372
    },
    {
      "epoch": 0.373,
      "grad_norm": 0.68359375,
      "grad_norm_var": 0.007610829671223959,
      "learning_rate": 0.0001,
      "loss": 0.4092,
      "loss/crossentropy": 2.7967922687530518,
      "loss/fcd": 0.3193359375,
      "loss/logits": 0.0898539088666439,
      "step": 373
    },
    {
      "epoch": 0.374,
      "grad_norm": 0.87890625,
      "grad_norm_var": 0.011069170633951823,
      "learning_rate": 0.0001,
      "loss": 0.4631,
      "loss/crossentropy": 2.9314931631088257,
      "loss/fcd": 0.34765625,
      "loss/logits": 0.11540572345256805,
      "step": 374
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.8125,
      "grad_norm_var": 0.012375895182291667,
      "learning_rate": 0.0001,
      "loss": 0.4492,
      "loss/crossentropy": 2.9116827249526978,
      "loss/fcd": 0.34375,
      "loss/logits": 0.10542852059006691,
      "step": 375
    },
    {
      "epoch": 0.376,
      "grad_norm": 0.81640625,
      "grad_norm_var": 0.013640340169270833,
      "learning_rate": 0.0001,
      "loss": 0.4443,
      "loss/crossentropy": 2.7188583612442017,
      "loss/fcd": 0.349609375,
      "loss/logits": 0.09467883035540581,
      "step": 376
    },
    {
      "epoch": 0.377,
      "grad_norm": 0.66015625,
      "grad_norm_var": 0.013356526692708334,
      "learning_rate": 0.0001,
      "loss": 0.4221,
      "loss/crossentropy": 2.844428539276123,
      "loss/fcd": 0.32421875,
      "loss/logits": 0.09790046513080597,
      "step": 377
    },
    {
      "epoch": 0.378,
      "grad_norm": 0.734375,
      "grad_norm_var": 0.013214556376139323,
      "learning_rate": 0.0001,
      "loss": 0.4092,
      "loss/crossentropy": 2.787495493888855,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.09276452288031578,
      "step": 378
    },
    {
      "epoch": 0.379,
      "grad_norm": 0.7421875,
      "grad_norm_var": 0.011052894592285156,
      "learning_rate": 0.0001,
      "loss": 0.4362,
      "loss/crossentropy": 2.9763654470443726,
      "loss/fcd": 0.3349609375,
      "loss/logits": 0.10128062590956688,
      "step": 379
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.85546875,
      "grad_norm_var": 0.012729835510253907,
      "learning_rate": 0.0001,
      "loss": 0.417,
      "loss/crossentropy": 2.846543788909912,
      "loss/fcd": 0.3232421875,
      "loss/logits": 0.09378386661410332,
      "step": 380
    },
    {
      "epoch": 0.381,
      "grad_norm": 0.83203125,
      "grad_norm_var": 0.013950347900390625,
      "learning_rate": 0.0001,
      "loss": 0.4281,
      "loss/crossentropy": 2.728231906890869,
      "loss/fcd": 0.3310546875,
      "loss/logits": 0.09700711071491241,
      "step": 381
    },
    {
      "epoch": 0.382,
      "grad_norm": 1.21875,
      "grad_norm_var": 0.030537859598795573,
      "learning_rate": 0.0001,
      "loss": 0.5004,
      "loss/crossentropy": 2.8327022790908813,
      "loss/fcd": 0.3544921875,
      "loss/logits": 0.1459062695503235,
      "step": 382
    },
    {
      "epoch": 0.383,
      "grad_norm": 0.91015625,
      "grad_norm_var": 0.03173058827718099,
      "learning_rate": 0.0001,
      "loss": 0.4301,
      "loss/crossentropy": 2.865984559059143,
      "loss/fcd": 0.3232421875,
      "loss/logits": 0.10680827498435974,
      "step": 383
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.828125,
      "grad_norm_var": 0.02973321278889974,
      "learning_rate": 0.0001,
      "loss": 0.4345,
      "loss/crossentropy": 2.8023486137390137,
      "loss/fcd": 0.3271484375,
      "loss/logits": 0.10733743384480476,
      "step": 384
    },
    {
      "epoch": 0.385,
      "grad_norm": 0.89453125,
      "grad_norm_var": 0.026640828450520834,
      "learning_rate": 0.0001,
      "loss": 0.4466,
      "loss/crossentropy": 2.9903111457824707,
      "loss/fcd": 0.328125,
      "loss/logits": 0.11846619471907616,
      "step": 385
    },
    {
      "epoch": 0.386,
      "grad_norm": 0.7421875,
      "grad_norm_var": 0.02116082509358724,
      "learning_rate": 0.0001,
      "loss": 0.4408,
      "loss/crossentropy": 2.8822553157806396,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.10392668098211288,
      "step": 386
    },
    {
      "epoch": 0.387,
      "grad_norm": 0.87890625,
      "grad_norm_var": 0.01885217030843099,
      "learning_rate": 0.0001,
      "loss": 0.4332,
      "loss/crossentropy": 3.0039666891098022,
      "loss/fcd": 0.330078125,
      "loss/logits": 0.10309557244181633,
      "step": 387
    },
    {
      "epoch": 0.388,
      "grad_norm": 0.91796875,
      "grad_norm_var": 0.016615804036458334,
      "learning_rate": 0.0001,
      "loss": 0.4587,
      "loss/crossentropy": 2.811259388923645,
      "loss/fcd": 0.3505859375,
      "loss/logits": 0.10812849923968315,
      "step": 388
    },
    {
      "epoch": 0.389,
      "grad_norm": 0.69140625,
      "grad_norm_var": 0.016458892822265626,
      "learning_rate": 0.0001,
      "loss": 0.4188,
      "loss/crossentropy": 2.794381260871887,
      "loss/fcd": 0.3193359375,
      "loss/logits": 0.0994318462908268,
      "step": 389
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.69140625,
      "grad_norm_var": 0.017642974853515625,
      "learning_rate": 0.0001,
      "loss": 0.4247,
      "loss/crossentropy": 2.789844274520874,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.09947730973362923,
      "step": 390
    },
    {
      "epoch": 0.391,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.019906044006347656,
      "learning_rate": 0.0001,
      "loss": 0.4485,
      "loss/crossentropy": 2.736423969268799,
      "loss/fcd": 0.337890625,
      "loss/logits": 0.11057012155652046,
      "step": 391
    },
    {
      "epoch": 0.392,
      "grad_norm": 0.66015625,
      "grad_norm_var": 0.021416664123535156,
      "learning_rate": 0.0001,
      "loss": 0.4431,
      "loss/crossentropy": 2.9826775789260864,
      "loss/fcd": 0.3330078125,
      "loss/logits": 0.11012021079659462,
      "step": 392
    },
    {
      "epoch": 0.393,
      "grad_norm": 0.66015625,
      "grad_norm_var": 0.021416664123535156,
      "learning_rate": 0.0001,
      "loss": 0.4056,
      "loss/crossentropy": 2.818847894668579,
      "loss/fcd": 0.3115234375,
      "loss/logits": 0.09412503615021706,
      "step": 393
    },
    {
      "epoch": 0.394,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.022944132486979168,
      "learning_rate": 0.0001,
      "loss": 0.4532,
      "loss/crossentropy": 2.912561774253845,
      "loss/fcd": 0.3408203125,
      "loss/logits": 0.11239226907491684,
      "step": 394
    },
    {
      "epoch": 0.395,
      "grad_norm": 0.71875,
      "grad_norm_var": 0.023158518473307292,
      "learning_rate": 0.0001,
      "loss": 0.4436,
      "loss/crossentropy": 2.7931829690933228,
      "loss/fcd": 0.34375,
      "loss/logits": 0.09981722384691238,
      "step": 395
    },
    {
      "epoch": 0.396,
      "grad_norm": 0.94140625,
      "grad_norm_var": 0.024274698893229165,
      "learning_rate": 0.0001,
      "loss": 0.4378,
      "loss/crossentropy": 2.9140548706054688,
      "loss/fcd": 0.333984375,
      "loss/logits": 0.10377934947609901,
      "step": 396
    },
    {
      "epoch": 0.397,
      "grad_norm": 0.7109375,
      "grad_norm_var": 0.024733924865722658,
      "learning_rate": 0.0001,
      "loss": 0.4385,
      "loss/crossentropy": 2.8483548164367676,
      "loss/fcd": 0.33203125,
      "loss/logits": 0.10648469254374504,
      "step": 397
    },
    {
      "epoch": 0.398,
      "grad_norm": 0.6953125,
      "grad_norm_var": 0.012363624572753907,
      "learning_rate": 0.0001,
      "loss": 0.4473,
      "loss/crossentropy": 2.7623294591903687,
      "loss/fcd": 0.33984375,
      "loss/logits": 0.10747401416301727,
      "step": 398
    },
    {
      "epoch": 0.399,
      "grad_norm": 0.5859375,
      "grad_norm_var": 0.012590535481770833,
      "learning_rate": 0.0001,
      "loss": 0.4196,
      "loss/crossentropy": 2.924271821975708,
      "loss/fcd": 0.326171875,
      "loss/logits": 0.09341751039028168,
      "step": 399
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5703125,
      "grad_norm_var": 0.013824208577473959,
      "learning_rate": 0.0001,
      "loss": 0.4132,
      "loss/crossentropy": 2.8939170837402344,
      "loss/fcd": 0.3232421875,
      "loss/logits": 0.08991139009594917,
      "step": 400
    },
    {
      "epoch": 0.401,
      "grad_norm": 0.65625,
      "grad_norm_var": 0.012051836649576823,
      "learning_rate": 0.0001,
      "loss": 0.4504,
      "loss/crossentropy": 2.948604702949524,
      "loss/fcd": 0.337890625,
      "loss/logits": 0.1125587709248066,
      "step": 401
    },
    {
      "epoch": 0.402,
      "grad_norm": 0.671875,
      "grad_norm_var": 0.012079302469889324,
      "learning_rate": 0.0001,
      "loss": 0.4236,
      "loss/crossentropy": 2.953108549118042,
      "loss/fcd": 0.326171875,
      "loss/logits": 0.09738374873995781,
      "step": 402
    },
    {
      "epoch": 0.403,
      "grad_norm": 0.640625,
      "grad_norm_var": 0.010190582275390625,
      "learning_rate": 0.0001,
      "loss": 0.4453,
      "loss/crossentropy": 2.8220441341400146,
      "loss/fcd": 0.341796875,
      "loss/logits": 0.10346821323037148,
      "step": 403
    },
    {
      "epoch": 0.404,
      "grad_norm": 0.72265625,
      "grad_norm_var": 0.006712849934895833,
      "learning_rate": 0.0001,
      "loss": 0.4319,
      "loss/crossentropy": 2.9475324153900146,
      "loss/fcd": 0.3310546875,
      "loss/logits": 0.10082582384347916,
      "step": 404
    },
    {
      "epoch": 0.405,
      "grad_norm": 0.828125,
      "grad_norm_var": 0.008076922098795573,
      "learning_rate": 0.0001,
      "loss": 0.4165,
      "loss/crossentropy": 2.79871141910553,
      "loss/fcd": 0.3193359375,
      "loss/logits": 0.09713438898324966,
      "step": 405
    },
    {
      "epoch": 0.406,
      "grad_norm": 0.89453125,
      "grad_norm_var": 0.010715166727701822,
      "learning_rate": 0.0001,
      "loss": 0.425,
      "loss/crossentropy": 2.78818678855896,
      "loss/fcd": 0.3271484375,
      "loss/logits": 0.09782487899065018,
      "step": 406
    },
    {
      "epoch": 0.407,
      "grad_norm": 0.73046875,
      "grad_norm_var": 0.010449663798014323,
      "learning_rate": 0.0001,
      "loss": 0.4082,
      "loss/crossentropy": 2.843753933906555,
      "loss/fcd": 0.3154296875,
      "loss/logits": 0.09279356896877289,
      "step": 407
    },
    {
      "epoch": 0.408,
      "grad_norm": 0.5703125,
      "grad_norm_var": 0.011524454752604166,
      "learning_rate": 0.0001,
      "loss": 0.4299,
      "loss/crossentropy": 2.911107063293457,
      "loss/fcd": 0.330078125,
      "loss/logits": 0.09986836463212967,
      "step": 408
    },
    {
      "epoch": 0.409,
      "grad_norm": 0.61328125,
      "grad_norm_var": 0.011924235026041667,
      "learning_rate": 0.0001,
      "loss": 0.4285,
      "loss/crossentropy": 2.840214490890503,
      "loss/fcd": 0.3310546875,
      "loss/logits": 0.09741492196917534,
      "step": 409
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6015625,
      "grad_norm_var": 0.012294451395670572,
      "learning_rate": 0.0001,
      "loss": 0.4203,
      "loss/crossentropy": 2.8811668157577515,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09806715697050095,
      "step": 410
    },
    {
      "epoch": 0.411,
      "grad_norm": 0.59765625,
      "grad_norm_var": 0.012860107421875,
      "learning_rate": 0.0001,
      "loss": 0.4131,
      "loss/crossentropy": 2.6903880834579468,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09085888788104057,
      "step": 411
    },
    {
      "epoch": 0.412,
      "grad_norm": 0.75,
      "grad_norm_var": 0.008719825744628906,
      "learning_rate": 0.0001,
      "loss": 0.4222,
      "loss/crossentropy": 2.8993114233016968,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.09704534709453583,
      "step": 412
    },
    {
      "epoch": 0.413,
      "grad_norm": 0.73828125,
      "grad_norm_var": 0.008888498942057291,
      "learning_rate": 0.0001,
      "loss": 0.4299,
      "loss/crossentropy": 2.9046199321746826,
      "loss/fcd": 0.328125,
      "loss/logits": 0.10177389904856682,
      "step": 413
    },
    {
      "epoch": 0.414,
      "grad_norm": 0.734375,
      "grad_norm_var": 0.009067789713541666,
      "learning_rate": 0.0001,
      "loss": 0.4429,
      "loss/crossentropy": 2.827233672142029,
      "loss/fcd": 0.33984375,
      "loss/logits": 0.10300642624497414,
      "step": 414
    },
    {
      "epoch": 0.415,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.008514404296875,
      "learning_rate": 0.0001,
      "loss": 0.4356,
      "loss/crossentropy": 2.8794647455215454,
      "loss/fcd": 0.3359375,
      "loss/logits": 0.09965472295880318,
      "step": 415
    },
    {
      "epoch": 0.416,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.0077697118123372395,
      "learning_rate": 0.0001,
      "loss": 0.4302,
      "loss/crossentropy": 2.7547887563705444,
      "loss/fcd": 0.3330078125,
      "loss/logits": 0.09721554070711136,
      "step": 416
    },
    {
      "epoch": 0.417,
      "grad_norm": 0.59765625,
      "grad_norm_var": 0.008245595296223958,
      "learning_rate": 0.0001,
      "loss": 0.4237,
      "loss/crossentropy": 2.985141396522522,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.09846670553088188,
      "step": 417
    },
    {
      "epoch": 0.418,
      "grad_norm": 0.5859375,
      "grad_norm_var": 0.008869425455729166,
      "learning_rate": 0.0001,
      "loss": 0.4615,
      "loss/crossentropy": 2.7390782833099365,
      "loss/fcd": 0.3544921875,
      "loss/logits": 0.10705351829528809,
      "step": 418
    },
    {
      "epoch": 0.419,
      "grad_norm": 0.87890625,
      "grad_norm_var": 0.011145973205566406,
      "learning_rate": 0.0001,
      "loss": 0.4159,
      "loss/crossentropy": 2.7943382263183594,
      "loss/fcd": 0.32421875,
      "loss/logits": 0.09164649993181229,
      "step": 419
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.59375,
      "grad_norm_var": 0.01171875,
      "learning_rate": 0.0001,
      "loss": 0.4377,
      "loss/crossentropy": 2.912733554840088,
      "loss/fcd": 0.3359375,
      "loss/logits": 0.10175374150276184,
      "step": 420
    },
    {
      "epoch": 0.421,
      "grad_norm": 0.625,
      "grad_norm_var": 0.0104888916015625,
      "learning_rate": 0.0001,
      "loss": 0.4021,
      "loss/crossentropy": 2.6603671312332153,
      "loss/fcd": 0.3125,
      "loss/logits": 0.08958860114216805,
      "step": 421
    },
    {
      "epoch": 0.422,
      "grad_norm": 0.84375,
      "grad_norm_var": 0.009162330627441406,
      "learning_rate": 0.0001,
      "loss": 0.4461,
      "loss/crossentropy": 2.68258535861969,
      "loss/fcd": 0.34375,
      "loss/logits": 0.10239701718091965,
      "step": 422
    },
    {
      "epoch": 0.423,
      "grad_norm": 0.7734375,
      "grad_norm_var": 0.009614817301432292,
      "learning_rate": 0.0001,
      "loss": 0.4318,
      "loss/crossentropy": 2.7980507612228394,
      "loss/fcd": 0.330078125,
      "loss/logits": 0.10173183679580688,
      "step": 423
    },
    {
      "epoch": 0.424,
      "grad_norm": 0.75390625,
      "grad_norm_var": 0.009175554911295573,
      "learning_rate": 0.0001,
      "loss": 0.512,
      "loss/crossentropy": 2.661671280860901,
      "loss/fcd": 0.3876953125,
      "loss/logits": 0.12431911006569862,
      "step": 424
    },
    {
      "epoch": 0.425,
      "grad_norm": 0.84765625,
      "grad_norm_var": 0.010342852274576823,
      "learning_rate": 0.0001,
      "loss": 0.4684,
      "loss/crossentropy": 2.962968587875366,
      "loss/fcd": 0.3515625,
      "loss/logits": 0.11681900918483734,
      "step": 425
    },
    {
      "epoch": 0.426,
      "grad_norm": 0.74609375,
      "grad_norm_var": 0.009742991129557291,
      "learning_rate": 0.0001,
      "loss": 0.4187,
      "loss/crossentropy": 2.812212586402893,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09639938920736313,
      "step": 426
    },
    {
      "epoch": 0.427,
      "grad_norm": 0.5625,
      "grad_norm_var": 0.010344378153483073,
      "learning_rate": 0.0001,
      "loss": 0.422,
      "loss/crossentropy": 2.7403759956359863,
      "loss/fcd": 0.328125,
      "loss/logits": 0.09389032050967216,
      "step": 427
    },
    {
      "epoch": 0.428,
      "grad_norm": 0.62109375,
      "grad_norm_var": 0.010648600260416667,
      "learning_rate": 0.0001,
      "loss": 0.4535,
      "loss/crossentropy": 2.9029101133346558,
      "loss/fcd": 0.3466796875,
      "loss/logits": 0.10677453130483627,
      "step": 428
    },
    {
      "epoch": 0.429,
      "grad_norm": 0.625,
      "grad_norm_var": 0.010860633850097657,
      "learning_rate": 0.0001,
      "loss": 0.4213,
      "loss/crossentropy": 2.9899282455444336,
      "loss/fcd": 0.3251953125,
      "loss/logits": 0.09613670036196709,
      "step": 429
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.66015625,
      "grad_norm_var": 0.010786946614583333,
      "learning_rate": 0.0001,
      "loss": 0.4325,
      "loss/crossentropy": 2.871562957763672,
      "loss/fcd": 0.3349609375,
      "loss/logits": 0.09752309322357178,
      "step": 430
    },
    {
      "epoch": 0.431,
      "grad_norm": 0.68359375,
      "grad_norm_var": 0.010681088765462239,
      "learning_rate": 0.0001,
      "loss": 0.4306,
      "loss/crossentropy": 2.6593871116638184,
      "loss/fcd": 0.333984375,
      "loss/logits": 0.09662756696343422,
      "step": 431
    },
    {
      "epoch": 0.432,
      "grad_norm": 0.7421875,
      "grad_norm_var": 0.010631306966145834,
      "learning_rate": 0.0001,
      "loss": 0.4289,
      "loss/crossentropy": 2.8526699542999268,
      "loss/fcd": 0.3310546875,
      "loss/logits": 0.09788180887699127,
      "step": 432
    },
    {
      "epoch": 0.433,
      "grad_norm": 0.8203125,
      "grad_norm_var": 0.01080163319905599,
      "learning_rate": 0.0001,
      "loss": 0.447,
      "loss/crossentropy": 2.845671057701111,
      "loss/fcd": 0.341796875,
      "loss/logits": 0.10521851852536201,
      "step": 433
    },
    {
      "epoch": 0.434,
      "grad_norm": 0.734375,
      "grad_norm_var": 0.009719276428222656,
      "learning_rate": 0.0001,
      "loss": 0.4569,
      "loss/crossentropy": 2.7953325510025024,
      "loss/fcd": 0.349609375,
      "loss/logits": 0.10731937736272812,
      "step": 434
    },
    {
      "epoch": 0.435,
      "grad_norm": 0.71484375,
      "grad_norm_var": 0.007914161682128907,
      "learning_rate": 0.0001,
      "loss": 0.4052,
      "loss/crossentropy": 2.741719603538513,
      "loss/fcd": 0.31640625,
      "loss/logits": 0.08881807327270508,
      "step": 435
    },
    {
      "epoch": 0.436,
      "grad_norm": 0.6328125,
      "grad_norm_var": 0.00740807851155599,
      "learning_rate": 0.0001,
      "loss": 0.4394,
      "loss/crossentropy": 2.8444197177886963,
      "loss/fcd": 0.337890625,
      "loss/logits": 0.10151328146457672,
      "step": 436
    },
    {
      "epoch": 0.437,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.007281239827473958,
      "learning_rate": 0.0001,
      "loss": 0.4549,
      "loss/crossentropy": 2.6685785055160522,
      "loss/fcd": 0.3525390625,
      "loss/logits": 0.1023550033569336,
      "step": 437
    },
    {
      "epoch": 0.438,
      "grad_norm": 0.74609375,
      "grad_norm_var": 0.006167030334472657,
      "learning_rate": 0.0001,
      "loss": 0.4555,
      "loss/crossentropy": 2.8580113649368286,
      "loss/fcd": 0.34375,
      "loss/logits": 0.11170928180217743,
      "step": 438
    },
    {
      "epoch": 0.439,
      "grad_norm": 0.765625,
      "grad_norm_var": 0.006100908915201823,
      "learning_rate": 0.0001,
      "loss": 0.4679,
      "loss/crossentropy": 2.7053956985473633,
      "loss/fcd": 0.357421875,
      "loss/logits": 0.11045508086681366,
      "step": 439
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.83203125,
      "grad_norm_var": 0.006983375549316407,
      "learning_rate": 0.0001,
      "loss": 0.4574,
      "loss/crossentropy": 3.011350154876709,
      "loss/fcd": 0.3515625,
      "loss/logits": 0.10579814016819,
      "step": 440
    },
    {
      "epoch": 0.441,
      "grad_norm": 0.9921875,
      "grad_norm_var": 0.0109283447265625,
      "learning_rate": 0.0001,
      "loss": 0.4559,
      "loss/crossentropy": 2.850973963737488,
      "loss/fcd": 0.3486328125,
      "loss/logits": 0.10725782811641693,
      "step": 441
    },
    {
      "epoch": 0.442,
      "grad_norm": 1.265625,
      "grad_norm_var": 0.029624366760253908,
      "learning_rate": 0.0001,
      "loss": 0.4703,
      "loss/crossentropy": 2.9250357151031494,
      "loss/fcd": 0.35546875,
      "loss/logits": 0.11481418460607529,
      "step": 442
    },
    {
      "epoch": 0.443,
      "grad_norm": 0.86328125,
      "grad_norm_var": 0.027671051025390626,
      "learning_rate": 0.0001,
      "loss": 0.4503,
      "loss/crossentropy": 2.8306068181991577,
      "loss/fcd": 0.34375,
      "loss/logits": 0.10658934339880943,
      "step": 443
    },
    {
      "epoch": 0.444,
      "grad_norm": 0.9296875,
      "grad_norm_var": 0.02745507558186849,
      "learning_rate": 0.0001,
      "loss": 0.4421,
      "loss/crossentropy": 2.835876703262329,
      "loss/fcd": 0.3359375,
      "loss/logits": 0.10615797340869904,
      "step": 444
    },
    {
      "epoch": 0.445,
      "grad_norm": 0.9375,
      "grad_norm_var": 0.026671791076660158,
      "learning_rate": 0.0001,
      "loss": 0.4746,
      "loss/crossentropy": 2.862402319908142,
      "loss/fcd": 0.36328125,
      "loss/logits": 0.11136271059513092,
      "step": 445
    },
    {
      "epoch": 0.446,
      "grad_norm": 0.796875,
      "grad_norm_var": 0.025111897786458334,
      "learning_rate": 0.0001,
      "loss": 0.4353,
      "loss/crossentropy": 2.6523581743240356,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.09838449582457542,
      "step": 446
    },
    {
      "epoch": 0.447,
      "grad_norm": 0.6875,
      "grad_norm_var": 0.02504266103108724,
      "learning_rate": 0.0001,
      "loss": 0.4331,
      "loss/crossentropy": 2.8293275833129883,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.09622528776526451,
      "step": 447
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.62109375,
      "grad_norm_var": 0.02719294230143229,
      "learning_rate": 0.0001,
      "loss": 0.4363,
      "loss/crossentropy": 2.827236771583557,
      "loss/fcd": 0.341796875,
      "loss/logits": 0.09454872459173203,
      "step": 448
    },
    {
      "epoch": 0.449,
      "grad_norm": 0.65234375,
      "grad_norm_var": 0.028748512268066406,
      "learning_rate": 0.0001,
      "loss": 0.4322,
      "loss/crossentropy": 2.9126251935958862,
      "loss/fcd": 0.3310546875,
      "loss/logits": 0.10111148655414581,
      "step": 449
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8203125,
      "grad_norm_var": 0.028451983133951822,
      "learning_rate": 0.0001,
      "loss": 0.4393,
      "loss/crossentropy": 2.932951331138611,
      "loss/fcd": 0.3359375,
      "loss/logits": 0.10334927961230278,
      "step": 450
    },
    {
      "epoch": 0.451,
      "grad_norm": 0.83203125,
      "grad_norm_var": 0.027887407938639322,
      "learning_rate": 0.0001,
      "loss": 0.4458,
      "loss/crossentropy": 2.7913864850997925,
      "loss/fcd": 0.341796875,
      "loss/logits": 0.1039762981235981,
      "step": 451
    },
    {
      "epoch": 0.452,
      "grad_norm": 0.70703125,
      "grad_norm_var": 0.02644627888997396,
      "learning_rate": 0.0001,
      "loss": 0.4415,
      "loss/crossentropy": 2.8713871240615845,
      "loss/fcd": 0.3408203125,
      "loss/logits": 0.10065218433737755,
      "step": 452
    },
    {
      "epoch": 0.453,
      "grad_norm": 1.0234375,
      "grad_norm_var": 0.02645257314046224,
      "learning_rate": 0.0001,
      "loss": 0.5023,
      "loss/crossentropy": 2.884295105934143,
      "loss/fcd": 0.36328125,
      "loss/logits": 0.1390550322830677,
      "step": 453
    },
    {
      "epoch": 0.454,
      "grad_norm": 0.70703125,
      "grad_norm_var": 0.02704766591389974,
      "learning_rate": 0.0001,
      "loss": 0.4461,
      "loss/crossentropy": 2.7804571390151978,
      "loss/fcd": 0.3447265625,
      "loss/logits": 0.10132800415158272,
      "step": 454
    },
    {
      "epoch": 0.455,
      "grad_norm": 0.7578125,
      "grad_norm_var": 0.02712853749593099,
      "learning_rate": 0.0001,
      "loss": 0.441,
      "loss/crossentropy": 2.756545662879944,
      "loss/fcd": 0.337890625,
      "loss/logits": 0.10309552773833275,
      "step": 455
    },
    {
      "epoch": 0.456,
      "grad_norm": 0.7890625,
      "grad_norm_var": 0.02728449503580729,
      "learning_rate": 0.0001,
      "loss": 0.4314,
      "loss/crossentropy": 2.8818628787994385,
      "loss/fcd": 0.3310546875,
      "loss/logits": 0.10030314698815346,
      "step": 456
    },
    {
      "epoch": 0.457,
      "grad_norm": 0.73046875,
      "grad_norm_var": 0.026130104064941408,
      "learning_rate": 0.0001,
      "loss": 0.4104,
      "loss/crossentropy": 2.8387563228607178,
      "loss/fcd": 0.3203125,
      "loss/logits": 0.09010699018836021,
      "step": 457
    },
    {
      "epoch": 0.458,
      "grad_norm": 0.78515625,
      "grad_norm_var": 0.0120147705078125,
      "learning_rate": 0.0001,
      "loss": 0.4569,
      "loss/crossentropy": 2.864213466644287,
      "loss/fcd": 0.3525390625,
      "loss/logits": 0.10435482114553452,
      "step": 458
    },
    {
      "epoch": 0.459,
      "grad_norm": 0.79296875,
      "grad_norm_var": 0.011637115478515625,
      "learning_rate": 0.0001,
      "loss": 0.423,
      "loss/crossentropy": 2.761037230491638,
      "loss/fcd": 0.3291015625,
      "loss/logits": 0.09392649680376053,
      "step": 459
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.73828125,
      "grad_norm_var": 0.01025079091389974,
      "learning_rate": 0.0001,
      "loss": 0.4241,
      "loss/crossentropy": 2.9691224098205566,
      "loss/fcd": 0.3291015625,
      "loss/logits": 0.09495318681001663,
      "step": 460
    },
    {
      "epoch": 0.461,
      "grad_norm": 0.6484375,
      "grad_norm_var": 0.009159278869628907,
      "learning_rate": 0.0001,
      "loss": 0.4355,
      "loss/crossentropy": 2.914411187171936,
      "loss/fcd": 0.3359375,
      "loss/logits": 0.09953488409519196,
      "step": 461
    },
    {
      "epoch": 0.462,
      "grad_norm": 0.8203125,
      "grad_norm_var": 0.009322547912597656,
      "learning_rate": 0.0001,
      "loss": 0.4917,
      "loss/crossentropy": 3.1634334325790405,
      "loss/fcd": 0.373046875,
      "loss/logits": 0.11867117509245872,
      "step": 462
    },
    {
      "epoch": 0.463,
      "grad_norm": 0.6953125,
      "grad_norm_var": 0.009253883361816406,
      "learning_rate": 0.0001,
      "loss": 0.4236,
      "loss/crossentropy": 2.865732789039612,
      "loss/fcd": 0.3271484375,
      "loss/logits": 0.09641865640878677,
      "step": 463
    },
    {
      "epoch": 0.464,
      "grad_norm": 0.63671875,
      "grad_norm_var": 0.008984820048014323,
      "learning_rate": 0.0001,
      "loss": 0.4393,
      "loss/crossentropy": 2.7939807176589966,
      "loss/fcd": 0.3369140625,
      "loss/logits": 0.10240520909428596,
      "step": 464
    },
    {
      "epoch": 0.465,
      "grad_norm": 0.65625,
      "grad_norm_var": 0.008930460611979166,
      "learning_rate": 0.0001,
      "loss": 0.4451,
      "loss/crossentropy": 2.7904140949249268,
      "loss/fcd": 0.341796875,
      "loss/logits": 0.10333988070487976,
      "step": 465
    },
    {
      "epoch": 0.466,
      "grad_norm": 0.71484375,
      "grad_norm_var": 0.008760515848795574,
      "learning_rate": 0.0001,
      "loss": 0.4413,
      "loss/crossentropy": 2.782050848007202,
      "loss/fcd": 0.33984375,
      "loss/logits": 0.10147681087255478,
      "step": 466
    },
    {
      "epoch": 0.467,
      "grad_norm": 0.75390625,
      "grad_norm_var": 0.008310381571451824,
      "learning_rate": 0.0001,
      "loss": 0.4635,
      "loss/crossentropy": 2.73897647857666,
      "loss/fcd": 0.3486328125,
      "loss/logits": 0.11481858789920807,
      "step": 467
    },
    {
      "epoch": 0.468,
      "grad_norm": 0.8203125,
      "grad_norm_var": 0.008503977457682292,
      "learning_rate": 0.0001,
      "loss": 0.4645,
      "loss/crossentropy": 2.7576816082000732,
      "loss/fcd": 0.353515625,
      "loss/logits": 0.11100797355175018,
      "step": 468
    },
    {
      "epoch": 0.469,
      "grad_norm": 0.953125,
      "grad_norm_var": 0.0062906901041666664,
      "learning_rate": 0.0001,
      "loss": 0.5082,
      "loss/crossentropy": 2.9028536081314087,
      "loss/fcd": 0.388671875,
      "loss/logits": 0.11947885528206825,
      "step": 469
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8046875,
      "grad_norm_var": 0.006327247619628907,
      "learning_rate": 0.0001,
      "loss": 0.4491,
      "loss/crossentropy": 2.803042411804199,
      "loss/fcd": 0.3466796875,
      "loss/logits": 0.10240250080823898,
      "step": 470
    },
    {
      "epoch": 0.471,
      "grad_norm": 0.6875,
      "grad_norm_var": 0.006620216369628906,
      "learning_rate": 0.0001,
      "loss": 0.4412,
      "loss/crossentropy": 2.86372172832489,
      "loss/fcd": 0.3408203125,
      "loss/logits": 0.10040333867073059,
      "step": 471
    },
    {
      "epoch": 0.472,
      "grad_norm": 0.67578125,
      "grad_norm_var": 0.006858062744140625,
      "learning_rate": 0.0001,
      "loss": 0.4593,
      "loss/crossentropy": 2.8334121704101562,
      "loss/fcd": 0.3505859375,
      "loss/logits": 0.10868023708462715,
      "step": 472
    },
    {
      "epoch": 0.473,
      "grad_norm": 1.9453125,
      "grad_norm_var": 0.09680474599202474,
      "learning_rate": 0.0001,
      "loss": 0.5875,
      "loss/crossentropy": 2.7985202074050903,
      "loss/fcd": 0.4072265625,
      "loss/logits": 0.18022966012358665,
      "step": 473
    },
    {
      "epoch": 0.474,
      "grad_norm": 1.3515625,
      "grad_norm_var": 0.11418228149414063,
      "learning_rate": 0.0001,
      "loss": 0.5436,
      "loss/crossentropy": 3.055509328842163,
      "loss/fcd": 0.392578125,
      "loss/logits": 0.15097329765558243,
      "step": 474
    },
    {
      "epoch": 0.475,
      "grad_norm": 1.0546875,
      "grad_norm_var": 0.11626529693603516,
      "learning_rate": 0.0001,
      "loss": 0.4853,
      "loss/crossentropy": 2.7848037481307983,
      "loss/fcd": 0.365234375,
      "loss/logits": 0.12005244940519333,
      "step": 475
    },
    {
      "epoch": 0.476,
      "grad_norm": 0.98828125,
      "grad_norm_var": 0.11570377349853515,
      "learning_rate": 0.0001,
      "loss": 0.506,
      "loss/crossentropy": 3.017175316810608,
      "loss/fcd": 0.380859375,
      "loss/logits": 0.1251136139035225,
      "step": 476
    },
    {
      "epoch": 0.477,
      "grad_norm": 1.2734375,
      "grad_norm_var": 0.12015933990478515,
      "learning_rate": 0.0001,
      "loss": 0.5155,
      "loss/crossentropy": 2.6267424821853638,
      "loss/fcd": 0.388671875,
      "loss/logits": 0.1268685646355152,
      "step": 477
    },
    {
      "epoch": 0.478,
      "grad_norm": 0.9453125,
      "grad_norm_var": 0.11935774485270183,
      "learning_rate": 0.0001,
      "loss": 0.4556,
      "loss/crossentropy": 2.7268201112747192,
      "loss/fcd": 0.353515625,
      "loss/logits": 0.10206079855561256,
      "step": 478
    },
    {
      "epoch": 0.479,
      "grad_norm": 0.84375,
      "grad_norm_var": 0.11599470774332682,
      "learning_rate": 0.0001,
      "loss": 0.4605,
      "loss/crossentropy": 2.879013180732727,
      "loss/fcd": 0.353515625,
      "loss/logits": 0.10701316967606544,
      "step": 479
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.78515625,
      "grad_norm_var": 0.11128838857014973,
      "learning_rate": 0.0001,
      "loss": 0.4878,
      "loss/crossentropy": 2.570293426513672,
      "loss/fcd": 0.3701171875,
      "loss/logits": 0.11772741004824638,
      "step": 480
    },
    {
      "epoch": 0.481,
      "grad_norm": 0.765625,
      "grad_norm_var": 0.10770308176676432,
      "learning_rate": 0.0001,
      "loss": 0.4587,
      "loss/crossentropy": 2.846108078956604,
      "loss/fcd": 0.349609375,
      "loss/logits": 0.10910140722990036,
      "step": 481
    },
    {
      "epoch": 0.482,
      "grad_norm": 0.9453125,
      "grad_norm_var": 0.10348307291666667,
      "learning_rate": 0.0001,
      "loss": 0.495,
      "loss/crossentropy": 3.0078119039535522,
      "loss/fcd": 0.3701171875,
      "loss/logits": 0.12487388029694557,
      "step": 482
    },
    {
      "epoch": 0.483,
      "grad_norm": 0.9296875,
      "grad_norm_var": 0.10024153391520182,
      "learning_rate": 0.0001,
      "loss": 0.4598,
      "loss/crossentropy": 2.8971179723739624,
      "loss/fcd": 0.3486328125,
      "loss/logits": 0.11118747293949127,
      "step": 483
    },
    {
      "epoch": 0.484,
      "grad_norm": 0.81640625,
      "grad_norm_var": 0.10032857259114583,
      "learning_rate": 0.0001,
      "loss": 0.4545,
      "loss/crossentropy": 2.7730342149734497,
      "loss/fcd": 0.353515625,
      "loss/logits": 0.10101194679737091,
      "step": 484
    },
    {
      "epoch": 0.485,
      "grad_norm": 0.73046875,
      "grad_norm_var": 0.10438378651936848,
      "learning_rate": 0.0001,
      "loss": 0.4642,
      "loss/crossentropy": 2.933692455291748,
      "loss/fcd": 0.3583984375,
      "loss/logits": 0.10578196868300438,
      "step": 485
    },
    {
      "epoch": 0.486,
      "grad_norm": 0.75,
      "grad_norm_var": 0.10578657786051432,
      "learning_rate": 0.0001,
      "loss": 0.4646,
      "loss/crossentropy": 2.8260008096694946,
      "loss/fcd": 0.35546875,
      "loss/logits": 0.1091219000518322,
      "step": 486
    },
    {
      "epoch": 0.487,
      "grad_norm": 0.765625,
      "grad_norm_var": 0.10324598948160807,
      "learning_rate": 0.0001,
      "loss": 0.4442,
      "loss/crossentropy": 2.8580541610717773,
      "loss/fcd": 0.33984375,
      "loss/logits": 0.10437064245343208,
      "step": 487
    },
    {
      "epoch": 0.488,
      "grad_norm": 0.70703125,
      "grad_norm_var": 0.10206902821858724,
      "learning_rate": 0.0001,
      "loss": 0.4382,
      "loss/crossentropy": 2.885003685951233,
      "loss/fcd": 0.337890625,
      "loss/logits": 0.10032079368829727,
      "step": 488
    },
    {
      "epoch": 0.489,
      "grad_norm": 0.79296875,
      "grad_norm_var": 0.035955556233723956,
      "learning_rate": 0.0001,
      "loss": 0.4776,
      "loss/crossentropy": 2.99890398979187,
      "loss/fcd": 0.3603515625,
      "loss/logits": 0.11722472310066223,
      "step": 489
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.73828125,
      "grad_norm_var": 0.022769610087076824,
      "learning_rate": 0.0001,
      "loss": 0.4513,
      "loss/crossentropy": 2.87681782245636,
      "loss/fcd": 0.345703125,
      "loss/logits": 0.10560647398233414,
      "step": 490
    },
    {
      "epoch": 0.491,
      "grad_norm": 0.7265625,
      "grad_norm_var": 0.02117811838785807,
      "learning_rate": 0.0001,
      "loss": 0.4427,
      "loss/crossentropy": 2.8826026916503906,
      "loss/fcd": 0.33984375,
      "loss/logits": 0.1029045432806015,
      "step": 491
    },
    {
      "epoch": 0.492,
      "grad_norm": 0.78515625,
      "grad_norm_var": 0.019849077860514323,
      "learning_rate": 0.0001,
      "loss": 0.4723,
      "loss/crossentropy": 2.90323805809021,
      "loss/fcd": 0.365234375,
      "loss/logits": 0.10705263912677765,
      "step": 492
    },
    {
      "epoch": 0.493,
      "grad_norm": 0.93359375,
      "grad_norm_var": 0.007033030192057292,
      "learning_rate": 0.0001,
      "loss": 0.5298,
      "loss/crossentropy": 2.849150061607361,
      "loss/fcd": 0.373046875,
      "loss/logits": 0.15670844912528992,
      "step": 493
    },
    {
      "epoch": 0.494,
      "grad_norm": 0.84765625,
      "grad_norm_var": 0.005867958068847656,
      "learning_rate": 0.0001,
      "loss": 0.4689,
      "loss/crossentropy": 2.828599691390991,
      "loss/fcd": 0.3564453125,
      "loss/logits": 0.11246544495224953,
      "step": 494
    },
    {
      "epoch": 0.495,
      "grad_norm": 0.87109375,
      "grad_norm_var": 0.006059773763020833,
      "learning_rate": 0.0001,
      "loss": 0.4789,
      "loss/crossentropy": 2.940881133079529,
      "loss/fcd": 0.359375,
      "loss/logits": 0.11956067010760307,
      "step": 495
    },
    {
      "epoch": 0.496,
      "grad_norm": 0.95703125,
      "grad_norm_var": 0.007436116536458333,
      "learning_rate": 0.0001,
      "loss": 0.4811,
      "loss/crossentropy": 2.7961113452911377,
      "loss/fcd": 0.3642578125,
      "loss/logits": 0.11681519448757172,
      "step": 496
    },
    {
      "epoch": 0.497,
      "grad_norm": 0.88671875,
      "grad_norm_var": 0.007532691955566407,
      "learning_rate": 0.0001,
      "loss": 0.4886,
      "loss/crossentropy": 2.9191471338272095,
      "loss/fcd": 0.3623046875,
      "loss/logits": 0.12631087005138397,
      "step": 497
    },
    {
      "epoch": 0.498,
      "grad_norm": 0.74609375,
      "grad_norm_var": 0.0067901611328125,
      "learning_rate": 0.0001,
      "loss": 0.4617,
      "loss/crossentropy": 2.8428783416748047,
      "loss/fcd": 0.3466796875,
      "loss/logits": 0.11499987542629242,
      "step": 498
    },
    {
      "epoch": 0.499,
      "grad_norm": 0.640625,
      "grad_norm_var": 0.007458241780598959,
      "learning_rate": 0.0001,
      "loss": 0.4178,
      "loss/crossentropy": 2.779066324234009,
      "loss/fcd": 0.322265625,
      "loss/logits": 0.09556294605135918,
      "step": 499
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.87890625,
      "grad_norm_var": 0.007893625895182292,
      "learning_rate": 0.0001,
      "loss": 0.5152,
      "loss/crossentropy": 2.9425406455993652,
      "loss/fcd": 0.380859375,
      "loss/logits": 0.13436781987547874,
      "step": 500
    }
  ],
  "logging_steps": 1,
  "max_steps": 1000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": true,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.2917680717824e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}