{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.566261678495529,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0012831308392477645,
      "grad_norm": 9.537115097045898,
      "learning_rate": 4.9975e-05,
      "loss": 15.9223,
      "step": 1
    },
    {
      "epoch": 0.002566261678495529,
      "grad_norm": 14.003229141235352,
      "learning_rate": 4.995e-05,
      "loss": 17.9005,
      "step": 2
    },
    {
      "epoch": 0.0038493925177432936,
      "grad_norm": 13.004029273986816,
      "learning_rate": 4.992500000000001e-05,
      "loss": 16.3292,
      "step": 3
    },
    {
      "epoch": 0.005132523356991058,
      "grad_norm": 26.871122360229492,
      "learning_rate": 4.99e-05,
      "loss": 15.5715,
      "step": 4
    },
    {
      "epoch": 0.006415654196238823,
      "grad_norm": 25.53430938720703,
      "learning_rate": 4.9875000000000006e-05,
      "loss": 14.103,
      "step": 5
    },
    {
      "epoch": 0.007698785035486587,
      "grad_norm": 26.39861297607422,
      "learning_rate": 4.9850000000000006e-05,
      "loss": 13.6807,
      "step": 6
    },
    {
      "epoch": 0.008981915874734351,
      "grad_norm": 20.607765197753906,
      "learning_rate": 4.9825000000000005e-05,
      "loss": 11.4559,
      "step": 7
    },
    {
      "epoch": 0.010265046713982116,
      "grad_norm": 23.218873977661133,
      "learning_rate": 4.9800000000000004e-05,
      "loss": 12.4352,
      "step": 8
    },
    {
      "epoch": 0.01154817755322988,
      "grad_norm": 28.108366012573242,
      "learning_rate": 4.9775000000000004e-05,
      "loss": 10.7017,
      "step": 9
    },
    {
      "epoch": 0.012831308392477645,
      "grad_norm": 28.258567810058594,
      "learning_rate": 4.975e-05,
      "loss": 10.6566,
      "step": 10
    },
    {
      "epoch": 0.01411443923172541,
      "grad_norm": 23.8648681640625,
      "learning_rate": 4.9725e-05,
      "loss": 8.1933,
      "step": 11
    },
    {
      "epoch": 0.015397570070973175,
      "grad_norm": 28.521028518676758,
      "learning_rate": 4.97e-05,
      "loss": 8.2035,
      "step": 12
    },
    {
      "epoch": 0.01668070091022094,
      "grad_norm": 30.17255210876465,
      "learning_rate": 4.967500000000001e-05,
      "loss": 7.7512,
      "step": 13
    },
    {
      "epoch": 0.017963831749468702,
      "grad_norm": 28.181726455688477,
      "learning_rate": 4.965e-05,
      "loss": 6.0859,
      "step": 14
    },
    {
      "epoch": 0.019246962588716467,
      "grad_norm": 30.046680450439453,
      "learning_rate": 4.962500000000001e-05,
      "loss": 5.7812,
      "step": 15
    },
    {
      "epoch": 0.020530093427964233,
      "grad_norm": 21.437732696533203,
      "learning_rate": 4.96e-05,
      "loss": 4.9606,
      "step": 16
    },
    {
      "epoch": 0.021813224267211998,
      "grad_norm": 14.27261734008789,
      "learning_rate": 4.9575000000000006e-05,
      "loss": 3.8279,
      "step": 17
    },
    {
      "epoch": 0.02309635510645976,
      "grad_norm": 11.45578670501709,
      "learning_rate": 4.9550000000000005e-05,
      "loss": 3.522,
      "step": 18
    },
    {
      "epoch": 0.024379485945707526,
      "grad_norm": 10.85922622680664,
      "learning_rate": 4.9525000000000004e-05,
      "loss": 3.3099,
      "step": 19
    },
    {
      "epoch": 0.02566261678495529,
      "grad_norm": 10.32077407836914,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 2.9669,
      "step": 20
    },
    {
      "epoch": 0.026945747624203056,
      "grad_norm": 10.480860710144043,
      "learning_rate": 4.9475e-05,
      "loss": 2.6655,
      "step": 21
    },
    {
      "epoch": 0.02822887846345082,
      "grad_norm": 11.026902198791504,
      "learning_rate": 4.945e-05,
      "loss": 2.675,
      "step": 22
    },
    {
      "epoch": 0.029512009302698584,
      "grad_norm": 11.019874572753906,
      "learning_rate": 4.9425e-05,
      "loss": 2.2989,
      "step": 23
    },
    {
      "epoch": 0.03079514014194635,
      "grad_norm": 6.715106010437012,
      "learning_rate": 4.94e-05,
      "loss": 1.9904,
      "step": 24
    },
    {
      "epoch": 0.03207827098119411,
      "grad_norm": 3.500138521194458,
      "learning_rate": 4.937500000000001e-05,
      "loss": 1.8062,
      "step": 25
    },
    {
      "epoch": 0.03336140182044188,
      "grad_norm": 3.117760419845581,
      "learning_rate": 4.935e-05,
      "loss": 1.9541,
      "step": 26
    },
    {
      "epoch": 0.03464453265968964,
      "grad_norm": 2.0400137901306152,
      "learning_rate": 4.9325000000000006e-05,
      "loss": 1.6518,
      "step": 27
    },
    {
      "epoch": 0.035927663498937404,
      "grad_norm": 2.3641414642333984,
      "learning_rate": 4.93e-05,
      "loss": 1.6426,
      "step": 28
    },
    {
      "epoch": 0.03721079433818517,
      "grad_norm": 3.277726888656616,
      "learning_rate": 4.9275000000000005e-05,
      "loss": 1.5592,
      "step": 29
    },
    {
      "epoch": 0.038493925177432935,
      "grad_norm": 2.5180115699768066,
      "learning_rate": 4.9250000000000004e-05,
      "loss": 1.5308,
      "step": 30
    },
    {
      "epoch": 0.039777056016680704,
      "grad_norm": 2.347200393676758,
      "learning_rate": 4.9225000000000004e-05,
      "loss": 1.456,
      "step": 31
    },
    {
      "epoch": 0.041060186855928466,
      "grad_norm": 2.332770347595215,
      "learning_rate": 4.92e-05,
      "loss": 1.3945,
      "step": 32
    },
    {
      "epoch": 0.04234331769517623,
      "grad_norm": 2.4756968021392822,
      "learning_rate": 4.9175e-05,
      "loss": 1.3481,
      "step": 33
    },
    {
      "epoch": 0.043626448534423996,
      "grad_norm": 2.36296010017395,
      "learning_rate": 4.915e-05,
      "loss": 1.3107,
      "step": 34
    },
    {
      "epoch": 0.04490957937367176,
      "grad_norm": 1.7880232334136963,
      "learning_rate": 4.9125e-05,
      "loss": 1.1661,
      "step": 35
    },
    {
      "epoch": 0.04619271021291952,
      "grad_norm": 1.9397306442260742,
      "learning_rate": 4.91e-05,
      "loss": 1.2593,
      "step": 36
    },
    {
      "epoch": 0.04747584105216729,
      "grad_norm": 1.4015451669692993,
      "learning_rate": 4.907500000000001e-05,
      "loss": 1.2378,
      "step": 37
    },
    {
      "epoch": 0.04875897189141505,
      "grad_norm": 2.613532304763794,
      "learning_rate": 4.905e-05,
      "loss": 1.2115,
      "step": 38
    },
    {
      "epoch": 0.05004210273066282,
      "grad_norm": 1.6681703329086304,
      "learning_rate": 4.9025000000000006e-05,
      "loss": 1.0648,
      "step": 39
    },
    {
      "epoch": 0.05132523356991058,
      "grad_norm": 1.9118123054504395,
      "learning_rate": 4.9e-05,
      "loss": 1.0889,
      "step": 40
    },
    {
      "epoch": 0.052608364409158344,
      "grad_norm": 1.2817410230636597,
      "learning_rate": 4.8975000000000005e-05,
      "loss": 1.0944,
      "step": 41
    },
    {
      "epoch": 0.05389149524840611,
      "grad_norm": 1.6214549541473389,
      "learning_rate": 4.8950000000000004e-05,
      "loss": 1.1167,
      "step": 42
    },
    {
      "epoch": 0.055174626087653875,
      "grad_norm": 2.9864516258239746,
      "learning_rate": 4.8925e-05,
      "loss": 0.9986,
      "step": 43
    },
    {
      "epoch": 0.05645775692690164,
      "grad_norm": 0.871046245098114,
      "learning_rate": 4.89e-05,
      "loss": 1.1744,
      "step": 44
    },
    {
      "epoch": 0.057740887766149405,
      "grad_norm": 0.9927793741226196,
      "learning_rate": 4.8875e-05,
      "loss": 1.0843,
      "step": 45
    },
    {
      "epoch": 0.05902401860539717,
      "grad_norm": 1.110379695892334,
      "learning_rate": 4.885e-05,
      "loss": 1.0892,
      "step": 46
    },
    {
      "epoch": 0.060307149444644936,
      "grad_norm": 0.7730011343955994,
      "learning_rate": 4.8825e-05,
      "loss": 1.0035,
      "step": 47
    },
    {
      "epoch": 0.0615902802838927,
      "grad_norm": 0.6672346591949463,
      "learning_rate": 4.88e-05,
      "loss": 1.009,
      "step": 48
    },
    {
      "epoch": 0.06287341112314046,
      "grad_norm": 0.7218219637870789,
      "learning_rate": 4.8775000000000007e-05,
      "loss": 1.0318,
      "step": 49
    },
    {
      "epoch": 0.06415654196238822,
      "grad_norm": 0.5907741189002991,
      "learning_rate": 4.875e-05,
      "loss": 1.0117,
      "step": 50
    },
    {
      "epoch": 0.065439672801636,
      "grad_norm": 1.6326884031295776,
      "learning_rate": 4.8725000000000005e-05,
      "loss": 0.9467,
      "step": 51
    },
    {
      "epoch": 0.06672280364088376,
      "grad_norm": 0.5119956135749817,
      "learning_rate": 4.87e-05,
      "loss": 0.9758,
      "step": 52
    },
    {
      "epoch": 0.06800593448013152,
      "grad_norm": 0.5629539489746094,
      "learning_rate": 4.8675000000000004e-05,
      "loss": 1.053,
      "step": 53
    },
    {
      "epoch": 0.06928906531937928,
      "grad_norm": 0.4981907606124878,
      "learning_rate": 4.8650000000000003e-05,
      "loss": 1.01,
      "step": 54
    },
    {
      "epoch": 0.07057219615862705,
      "grad_norm": 0.40380367636680603,
      "learning_rate": 4.8625e-05,
      "loss": 0.9324,
      "step": 55
    },
    {
      "epoch": 0.07185532699787481,
      "grad_norm": 0.5838307738304138,
      "learning_rate": 4.86e-05,
      "loss": 0.9434,
      "step": 56
    },
    {
      "epoch": 0.07313845783712258,
      "grad_norm": 0.36131080985069275,
      "learning_rate": 4.8575e-05,
      "loss": 0.9273,
      "step": 57
    },
    {
      "epoch": 0.07442158867637035,
      "grad_norm": 0.9754893779754639,
      "learning_rate": 4.855e-05,
      "loss": 0.9915,
      "step": 58
    },
    {
      "epoch": 0.07570471951561811,
      "grad_norm": 0.3703399896621704,
      "learning_rate": 4.8525e-05,
      "loss": 0.8683,
      "step": 59
    },
    {
      "epoch": 0.07698785035486587,
      "grad_norm": 2.2767622470855713,
      "learning_rate": 4.85e-05,
      "loss": 0.9292,
      "step": 60
    },
    {
      "epoch": 0.07827098119411363,
      "grad_norm": 0.3819205164909363,
      "learning_rate": 4.8475000000000006e-05,
      "loss": 0.8336,
      "step": 61
    },
    {
      "epoch": 0.07955411203336141,
      "grad_norm": 0.4169008731842041,
      "learning_rate": 4.845e-05,
      "loss": 0.9473,
      "step": 62
    },
    {
      "epoch": 0.08083724287260917,
      "grad_norm": 0.2896365225315094,
      "learning_rate": 4.8425000000000005e-05,
      "loss": 0.8677,
      "step": 63
    },
    {
      "epoch": 0.08212037371185693,
      "grad_norm": 0.35320723056793213,
      "learning_rate": 4.8400000000000004e-05,
      "loss": 0.8455,
      "step": 64
    },
    {
      "epoch": 0.08340350455110469,
      "grad_norm": 0.29852598905563354,
      "learning_rate": 4.8375000000000004e-05,
      "loss": 0.8759,
      "step": 65
    },
    {
      "epoch": 0.08468663539035245,
      "grad_norm": 0.30656442046165466,
      "learning_rate": 4.835e-05,
      "loss": 0.9847,
      "step": 66
    },
    {
      "epoch": 0.08596976622960023,
      "grad_norm": 0.29515132308006287,
      "learning_rate": 4.8325e-05,
      "loss": 0.8363,
      "step": 67
    },
    {
      "epoch": 0.08725289706884799,
      "grad_norm": 0.2757239043712616,
      "learning_rate": 4.83e-05,
      "loss": 0.7905,
      "step": 68
    },
    {
      "epoch": 0.08853602790809575,
      "grad_norm": 0.23530316352844238,
      "learning_rate": 4.8275e-05,
      "loss": 0.8416,
      "step": 69
    },
    {
      "epoch": 0.08981915874734352,
      "grad_norm": 0.2849583923816681,
      "learning_rate": 4.825e-05,
      "loss": 0.8048,
      "step": 70
    },
    {
      "epoch": 0.09110228958659128,
      "grad_norm": 0.2504192590713501,
      "learning_rate": 4.822500000000001e-05,
      "loss": 0.8059,
      "step": 71
    },
    {
      "epoch": 0.09238542042583904,
      "grad_norm": 0.20606613159179688,
      "learning_rate": 4.82e-05,
      "loss": 0.8948,
      "step": 72
    },
    {
      "epoch": 0.09366855126508682,
      "grad_norm": 0.20878218114376068,
      "learning_rate": 4.8175000000000005e-05,
      "loss": 0.8016,
      "step": 73
    },
    {
      "epoch": 0.09495168210433458,
      "grad_norm": 0.22858284413814545,
      "learning_rate": 4.815e-05,
      "loss": 0.9773,
      "step": 74
    },
    {
      "epoch": 0.09623481294358234,
      "grad_norm": 0.21903325617313385,
      "learning_rate": 4.8125000000000004e-05,
      "loss": 0.9133,
      "step": 75
    },
    {
      "epoch": 0.0975179437828301,
      "grad_norm": 0.6681258082389832,
      "learning_rate": 4.8100000000000004e-05,
      "loss": 0.8091,
      "step": 76
    },
    {
      "epoch": 0.09880107462207786,
      "grad_norm": 0.5215646624565125,
      "learning_rate": 4.8075e-05,
      "loss": 0.8939,
      "step": 77
    },
    {
      "epoch": 0.10008420546132564,
      "grad_norm": 0.2597210705280304,
      "learning_rate": 4.805e-05,
      "loss": 0.7251,
      "step": 78
    },
    {
      "epoch": 0.1013673363005734,
      "grad_norm": 0.17470119893550873,
      "learning_rate": 4.8025e-05,
      "loss": 0.9407,
      "step": 79
    },
    {
      "epoch": 0.10265046713982116,
      "grad_norm": 0.2565261721611023,
      "learning_rate": 4.8e-05,
      "loss": 0.7954,
      "step": 80
    },
    {
      "epoch": 0.10393359797906893,
      "grad_norm": 0.23160725831985474,
      "learning_rate": 4.7975e-05,
      "loss": 0.8817,
      "step": 81
    },
    {
      "epoch": 0.10521672881831669,
      "grad_norm": 0.17381028831005096,
      "learning_rate": 4.795e-05,
      "loss": 0.7779,
      "step": 82
    },
    {
      "epoch": 0.10649985965756446,
      "grad_norm": 0.27167728543281555,
      "learning_rate": 4.7925000000000006e-05,
      "loss": 0.8613,
      "step": 83
    },
    {
      "epoch": 0.10778299049681223,
      "grad_norm": 0.22407206892967224,
      "learning_rate": 4.79e-05,
      "loss": 0.7454,
      "step": 84
    },
    {
      "epoch": 0.10906612133605999,
      "grad_norm": 0.20605726540088654,
      "learning_rate": 4.7875000000000005e-05,
      "loss": 0.7522,
      "step": 85
    },
    {
      "epoch": 0.11034925217530775,
      "grad_norm": 0.19919969141483307,
      "learning_rate": 4.785e-05,
      "loss": 0.8746,
      "step": 86
    },
    {
      "epoch": 0.11163238301455551,
      "grad_norm": 0.2132246494293213,
      "learning_rate": 4.7825000000000004e-05,
      "loss": 0.8168,
      "step": 87
    },
    {
      "epoch": 0.11291551385380327,
      "grad_norm": 0.20548345148563385,
      "learning_rate": 4.78e-05,
      "loss": 0.7897,
      "step": 88
    },
    {
      "epoch": 0.11419864469305105,
      "grad_norm": 0.1737872213125229,
      "learning_rate": 4.7775e-05,
      "loss": 0.6711,
      "step": 89
    },
    {
      "epoch": 0.11548177553229881,
      "grad_norm": 0.1772836446762085,
      "learning_rate": 4.775e-05,
      "loss": 0.7469,
      "step": 90
    },
    {
      "epoch": 0.11676490637154657,
      "grad_norm": 0.20224907994270325,
      "learning_rate": 4.7725e-05,
      "loss": 0.8637,
      "step": 91
    },
    {
      "epoch": 0.11804803721079433,
      "grad_norm": 0.27053549885749817,
      "learning_rate": 4.77e-05,
      "loss": 0.8397,
      "step": 92
    },
    {
      "epoch": 0.1193311680500421,
      "grad_norm": 0.19204196333885193,
      "learning_rate": 4.7675e-05,
      "loss": 0.8484,
      "step": 93
    },
    {
      "epoch": 0.12061429888928987,
      "grad_norm": 0.39770379662513733,
      "learning_rate": 4.765e-05,
      "loss": 0.7054,
      "step": 94
    },
    {
      "epoch": 0.12189742972853763,
      "grad_norm": 0.23019109666347504,
      "learning_rate": 4.7625000000000006e-05,
      "loss": 0.6887,
      "step": 95
    },
    {
      "epoch": 0.1231805605677854,
      "grad_norm": 0.26943087577819824,
      "learning_rate": 4.76e-05,
      "loss": 0.8154,
      "step": 96
    },
    {
      "epoch": 0.12446369140703316,
      "grad_norm": 0.1778383105993271,
      "learning_rate": 4.7575000000000004e-05,
      "loss": 0.8407,
      "step": 97
    },
    {
      "epoch": 0.12574682224628092,
      "grad_norm": 0.18584376573562622,
      "learning_rate": 4.755e-05,
      "loss": 0.7518,
      "step": 98
    },
    {
      "epoch": 0.1270299530855287,
      "grad_norm": 0.15931403636932373,
      "learning_rate": 4.7525e-05,
      "loss": 0.7887,
      "step": 99
    },
    {
      "epoch": 0.12831308392477644,
      "grad_norm": 0.22521662712097168,
      "learning_rate": 4.75e-05,
      "loss": 0.8666,
      "step": 100
    },
    {
      "epoch": 0.12959621476402422,
      "grad_norm": 0.1844562143087387,
      "learning_rate": 4.7475e-05,
      "loss": 0.8231,
      "step": 101
    },
    {
      "epoch": 0.130879345603272,
      "grad_norm": 0.21024508774280548,
      "learning_rate": 4.745e-05,
      "loss": 0.7319,
      "step": 102
    },
    {
      "epoch": 0.13216247644251974,
      "grad_norm": 0.1971369832754135,
      "learning_rate": 4.7425e-05,
      "loss": 0.8952,
      "step": 103
    },
    {
      "epoch": 0.13344560728176752,
      "grad_norm": 1.137277364730835,
      "learning_rate": 4.74e-05,
      "loss": 0.8453,
      "step": 104
    },
    {
      "epoch": 0.13472873812101527,
      "grad_norm": 0.15226082503795624,
      "learning_rate": 4.7375e-05,
      "loss": 0.7656,
      "step": 105
    },
    {
      "epoch": 0.13601186896026304,
      "grad_norm": 1.3863773345947266,
      "learning_rate": 4.735e-05,
      "loss": 0.8475,
      "step": 106
    },
    {
      "epoch": 0.13729499979951082,
      "grad_norm": 0.26768067479133606,
      "learning_rate": 4.7325000000000005e-05,
      "loss": 0.852,
      "step": 107
    },
    {
      "epoch": 0.13857813063875857,
      "grad_norm": 0.24242669343948364,
      "learning_rate": 4.73e-05,
      "loss": 0.8338,
      "step": 108
    },
    {
      "epoch": 0.13986126147800634,
      "grad_norm": 0.1744064837694168,
      "learning_rate": 4.7275000000000004e-05,
      "loss": 0.8767,
      "step": 109
    },
    {
      "epoch": 0.1411443923172541,
      "grad_norm": 0.17411988973617554,
      "learning_rate": 4.7249999999999997e-05,
      "loss": 0.6993,
      "step": 110
    },
    {
      "epoch": 0.14242752315650187,
      "grad_norm": 0.16757948696613312,
      "learning_rate": 4.7225e-05,
      "loss": 0.711,
      "step": 111
    },
    {
      "epoch": 0.14371065399574962,
      "grad_norm": 0.1696353703737259,
      "learning_rate": 4.72e-05,
      "loss": 0.7874,
      "step": 112
    },
    {
      "epoch": 0.1449937848349974,
      "grad_norm": 0.14376036822795868,
      "learning_rate": 4.7175e-05,
      "loss": 0.7958,
      "step": 113
    },
    {
      "epoch": 0.14627691567424517,
      "grad_norm": 0.15580502152442932,
      "learning_rate": 4.715e-05,
      "loss": 0.9563,
      "step": 114
    },
    {
      "epoch": 0.14756004651349292,
      "grad_norm": 0.14952343702316284,
      "learning_rate": 4.7125e-05,
      "loss": 0.7872,
      "step": 115
    },
    {
      "epoch": 0.1488431773527407,
      "grad_norm": 0.510981559753418,
      "learning_rate": 4.71e-05,
      "loss": 0.8497,
      "step": 116
    },
    {
      "epoch": 0.15012630819198844,
      "grad_norm": 0.19700995087623596,
      "learning_rate": 4.7075e-05,
      "loss": 0.8317,
      "step": 117
    },
    {
      "epoch": 0.15140943903123621,
      "grad_norm": 0.12396173179149628,
      "learning_rate": 4.705e-05,
      "loss": 0.841,
      "step": 118
    },
    {
      "epoch": 0.152692569870484,
      "grad_norm": 0.15683265030384064,
      "learning_rate": 4.7025000000000005e-05,
      "loss": 0.8015,
      "step": 119
    },
    {
      "epoch": 0.15397570070973174,
      "grad_norm": 0.3126041889190674,
      "learning_rate": 4.7e-05,
      "loss": 0.7729,
      "step": 120
    },
    {
      "epoch": 0.15525883154897951,
      "grad_norm": 0.23476779460906982,
      "learning_rate": 4.6975000000000003e-05,
      "loss": 0.7395,
      "step": 121
    },
    {
      "epoch": 0.15654196238822726,
      "grad_norm": 0.12463822215795517,
      "learning_rate": 4.695e-05,
      "loss": 0.8292,
      "step": 122
    },
    {
      "epoch": 0.15782509322747504,
      "grad_norm": 0.11740697920322418,
      "learning_rate": 4.6925e-05,
      "loss": 0.7641,
      "step": 123
    },
    {
      "epoch": 0.15910822406672281,
      "grad_norm": 0.1403815895318985,
      "learning_rate": 4.69e-05,
      "loss": 0.6957,
      "step": 124
    },
    {
      "epoch": 0.16039135490597056,
      "grad_norm": 0.3695715367794037,
      "learning_rate": 4.6875e-05,
      "loss": 0.7657,
      "step": 125
    },
    {
      "epoch": 0.16167448574521834,
      "grad_norm": 0.12423618882894516,
      "learning_rate": 4.685000000000001e-05,
      "loss": 0.6624,
      "step": 126
    },
    {
      "epoch": 0.1629576165844661,
      "grad_norm": 0.1719760000705719,
      "learning_rate": 4.6825e-05,
      "loss": 0.7968,
      "step": 127
    },
    {
      "epoch": 0.16424074742371386,
      "grad_norm": 0.14952242374420166,
      "learning_rate": 4.6800000000000006e-05,
      "loss": 0.7519,
      "step": 128
    },
    {
      "epoch": 0.16552387826296164,
      "grad_norm": 0.23585611581802368,
      "learning_rate": 4.6775000000000005e-05,
      "loss": 0.6761,
      "step": 129
    },
    {
      "epoch": 0.16680700910220939,
      "grad_norm": 0.13714896142482758,
      "learning_rate": 4.6750000000000005e-05,
      "loss": 0.7623,
      "step": 130
    },
    {
      "epoch": 0.16809013994145716,
      "grad_norm": 0.8348445892333984,
      "learning_rate": 4.6725000000000004e-05,
      "loss": 0.7666,
      "step": 131
    },
    {
      "epoch": 0.1693732707807049,
      "grad_norm": 0.4564683735370636,
      "learning_rate": 4.6700000000000003e-05,
      "loss": 0.7857,
      "step": 132
    },
    {
      "epoch": 0.17065640161995269,
      "grad_norm": 0.15979893505573273,
      "learning_rate": 4.6675e-05,
      "loss": 0.6463,
      "step": 133
    },
    {
      "epoch": 0.17193953245920046,
      "grad_norm": 0.23555421829223633,
      "learning_rate": 4.665e-05,
      "loss": 0.8773,
      "step": 134
    },
    {
      "epoch": 0.1732226632984482,
      "grad_norm": 0.2651432156562805,
      "learning_rate": 4.6625e-05,
      "loss": 0.7965,
      "step": 135
    },
    {
      "epoch": 0.17450579413769599,
      "grad_norm": 0.17633482813835144,
      "learning_rate": 4.660000000000001e-05,
      "loss": 0.6069,
      "step": 136
    },
    {
      "epoch": 0.17578892497694373,
      "grad_norm": 0.16391968727111816,
      "learning_rate": 4.6575e-05,
      "loss": 0.7548,
      "step": 137
    },
    {
      "epoch": 0.1770720558161915,
      "grad_norm": 0.1614704579114914,
      "learning_rate": 4.655000000000001e-05,
      "loss": 0.7524,
      "step": 138
    },
    {
      "epoch": 0.17835518665543929,
      "grad_norm": 0.17900526523590088,
      "learning_rate": 4.6525e-05,
      "loss": 0.7822,
      "step": 139
    },
    {
      "epoch": 0.17963831749468703,
      "grad_norm": 0.1562780886888504,
      "learning_rate": 4.6500000000000005e-05,
      "loss": 0.7525,
      "step": 140
    },
    {
      "epoch": 0.1809214483339348,
      "grad_norm": 0.18637755513191223,
      "learning_rate": 4.6475000000000005e-05,
      "loss": 0.9685,
      "step": 141
    },
    {
      "epoch": 0.18220457917318256,
      "grad_norm": 0.15222413837909698,
      "learning_rate": 4.6450000000000004e-05,
      "loss": 0.7117,
      "step": 142
    },
    {
      "epoch": 0.18348771001243033,
      "grad_norm": 0.28989797830581665,
      "learning_rate": 4.6425000000000004e-05,
      "loss": 0.7748,
      "step": 143
    },
    {
      "epoch": 0.18477084085167808,
      "grad_norm": 0.158959299325943,
      "learning_rate": 4.64e-05,
      "loss": 0.801,
      "step": 144
    },
    {
      "epoch": 0.18605397169092586,
      "grad_norm": 0.213655024766922,
      "learning_rate": 4.6375e-05,
      "loss": 0.8263,
      "step": 145
    },
    {
      "epoch": 0.18733710253017363,
      "grad_norm": 0.15139731764793396,
      "learning_rate": 4.635e-05,
      "loss": 0.7891,
      "step": 146
    },
    {
      "epoch": 0.18862023336942138,
      "grad_norm": 0.18600833415985107,
      "learning_rate": 4.6325e-05,
      "loss": 0.7138,
      "step": 147
    },
    {
      "epoch": 0.18990336420866916,
      "grad_norm": 0.18099449574947357,
      "learning_rate": 4.630000000000001e-05,
      "loss": 0.6801,
      "step": 148
    },
    {
      "epoch": 0.1911864950479169,
      "grad_norm": 0.1817125678062439,
      "learning_rate": 4.6275e-05,
      "loss": 0.7756,
      "step": 149
    },
    {
      "epoch": 0.19246962588716468,
      "grad_norm": 0.15464788675308228,
      "learning_rate": 4.6250000000000006e-05,
      "loss": 0.8141,
      "step": 150
    },
    {
      "epoch": 0.19375275672641246,
      "grad_norm": 0.1502385139465332,
      "learning_rate": 4.6225e-05,
      "loss": 0.7853,
      "step": 151
    },
    {
      "epoch": 0.1950358875656602,
      "grad_norm": 0.15769711136817932,
      "learning_rate": 4.6200000000000005e-05,
      "loss": 0.7538,
      "step": 152
    },
    {
      "epoch": 0.19631901840490798,
      "grad_norm": 0.17884378135204315,
      "learning_rate": 4.6175000000000004e-05,
      "loss": 0.7834,
      "step": 153
    },
    {
      "epoch": 0.19760214924415573,
      "grad_norm": 0.2618480324745178,
      "learning_rate": 4.6150000000000004e-05,
      "loss": 0.7247,
      "step": 154
    },
    {
      "epoch": 0.1988852800834035,
      "grad_norm": 0.16772985458374023,
      "learning_rate": 4.6125e-05,
      "loss": 0.7222,
      "step": 155
    },
    {
      "epoch": 0.20016841092265128,
      "grad_norm": 0.5297219753265381,
      "learning_rate": 4.61e-05,
      "loss": 0.623,
      "step": 156
    },
    {
      "epoch": 0.20145154176189903,
      "grad_norm": 0.1848820298910141,
      "learning_rate": 4.6075e-05,
      "loss": 0.867,
      "step": 157
    },
    {
      "epoch": 0.2027346726011468,
      "grad_norm": 0.18218453228473663,
      "learning_rate": 4.605e-05,
      "loss": 0.7626,
      "step": 158
    },
    {
      "epoch": 0.20401780344039455,
      "grad_norm": 0.15847554802894592,
      "learning_rate": 4.6025e-05,
      "loss": 0.7309,
      "step": 159
    },
    {
      "epoch": 0.20530093427964233,
      "grad_norm": 0.17456156015396118,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.8075,
      "step": 160
    },
    {
      "epoch": 0.2065840651188901,
      "grad_norm": 0.22012141346931458,
      "learning_rate": 4.5975e-05,
      "loss": 0.7555,
      "step": 161
    },
    {
      "epoch": 0.20786719595813785,
      "grad_norm": 0.1616412252187729,
      "learning_rate": 4.5950000000000006e-05,
      "loss": 0.7499,
      "step": 162
    },
    {
      "epoch": 0.20915032679738563,
      "grad_norm": 0.17827025055885315,
      "learning_rate": 4.5925e-05,
      "loss": 0.825,
      "step": 163
    },
    {
      "epoch": 0.21043345763663338,
      "grad_norm": 0.16361252963542938,
      "learning_rate": 4.5900000000000004e-05,
      "loss": 0.812,
      "step": 164
    },
    {
      "epoch": 0.21171658847588115,
      "grad_norm": 0.15956725180149078,
      "learning_rate": 4.5875000000000004e-05,
      "loss": 0.8105,
      "step": 165
    },
    {
      "epoch": 0.21299971931512893,
      "grad_norm": 0.142374649643898,
      "learning_rate": 4.585e-05,
      "loss": 0.6533,
      "step": 166
    },
    {
      "epoch": 0.21428285015437668,
      "grad_norm": 0.21484073996543884,
      "learning_rate": 4.5825e-05,
      "loss": 0.7558,
      "step": 167
    },
    {
      "epoch": 0.21556598099362445,
      "grad_norm": 0.19737152755260468,
      "learning_rate": 4.58e-05,
      "loss": 0.7859,
      "step": 168
    },
    {
      "epoch": 0.2168491118328722,
      "grad_norm": 0.2588985860347748,
      "learning_rate": 4.5775e-05,
      "loss": 0.7326,
      "step": 169
    },
    {
      "epoch": 0.21813224267211997,
      "grad_norm": 0.25438037514686584,
      "learning_rate": 4.575e-05,
      "loss": 0.7487,
      "step": 170
    },
    {
      "epoch": 0.21941537351136775,
      "grad_norm": 0.18834321200847626,
      "learning_rate": 4.5725e-05,
      "loss": 0.7649,
      "step": 171
    },
    {
      "epoch": 0.2206985043506155,
      "grad_norm": 0.1718263030052185,
      "learning_rate": 4.5700000000000006e-05,
      "loss": 0.7104,
      "step": 172
    },
    {
      "epoch": 0.22198163518986327,
      "grad_norm": 0.3754954934120178,
      "learning_rate": 4.5675e-05,
      "loss": 0.7586,
      "step": 173
    },
    {
      "epoch": 0.22326476602911102,
      "grad_norm": 0.21970051527023315,
      "learning_rate": 4.5650000000000005e-05,
      "loss": 0.8078,
      "step": 174
    },
    {
      "epoch": 0.2245478968683588,
      "grad_norm": 0.1525285243988037,
      "learning_rate": 4.5625e-05,
      "loss": 0.7373,
      "step": 175
    },
    {
      "epoch": 0.22583102770760655,
      "grad_norm": 0.24669454991817474,
      "learning_rate": 4.5600000000000004e-05,
      "loss": 0.7786,
      "step": 176
    },
    {
      "epoch": 0.22711415854685432,
      "grad_norm": 0.17058874666690826,
      "learning_rate": 4.5575e-05,
      "loss": 0.7569,
      "step": 177
    },
    {
      "epoch": 0.2283972893861021,
      "grad_norm": 0.16309134662151337,
      "learning_rate": 4.555e-05,
      "loss": 0.7962,
      "step": 178
    },
    {
      "epoch": 0.22968042022534985,
      "grad_norm": 0.23535701632499695,
      "learning_rate": 4.5525e-05,
      "loss": 0.7849,
      "step": 179
    },
    {
      "epoch": 0.23096355106459762,
      "grad_norm": 0.23046502470970154,
      "learning_rate": 4.55e-05,
      "loss": 0.7611,
      "step": 180
    },
    {
      "epoch": 0.23224668190384537,
      "grad_norm": 0.20204584300518036,
      "learning_rate": 4.5475e-05,
      "loss": 0.7258,
      "step": 181
    },
    {
      "epoch": 0.23352981274309315,
      "grad_norm": 0.14607994258403778,
      "learning_rate": 4.545000000000001e-05,
      "loss": 0.7558,
      "step": 182
    },
    {
      "epoch": 0.23481294358234092,
      "grad_norm": 0.2063671052455902,
      "learning_rate": 4.5425e-05,
      "loss": 0.7373,
      "step": 183
    },
    {
      "epoch": 0.23609607442158867,
      "grad_norm": 0.1957271248102188,
      "learning_rate": 4.5400000000000006e-05,
      "loss": 0.7724,
      "step": 184
    },
    {
      "epoch": 0.23737920526083645,
      "grad_norm": 0.2711730897426605,
      "learning_rate": 4.5375e-05,
      "loss": 0.8121,
      "step": 185
    },
    {
      "epoch": 0.2386623361000842,
      "grad_norm": 0.26164382696151733,
      "learning_rate": 4.5350000000000005e-05,
      "loss": 0.7644,
      "step": 186
    },
    {
      "epoch": 0.23994546693933197,
      "grad_norm": 0.17849335074424744,
      "learning_rate": 4.5325000000000004e-05,
      "loss": 0.8668,
      "step": 187
    },
    {
      "epoch": 0.24122859777857975,
      "grad_norm": 0.1632041186094284,
      "learning_rate": 4.53e-05,
      "loss": 0.7589,
      "step": 188
    },
    {
      "epoch": 0.2425117286178275,
      "grad_norm": 0.23931750655174255,
      "learning_rate": 4.5275e-05,
      "loss": 0.7503,
      "step": 189
    },
    {
      "epoch": 0.24379485945707527,
      "grad_norm": 0.19124819338321686,
      "learning_rate": 4.525e-05,
      "loss": 0.6548,
      "step": 190
    },
    {
      "epoch": 0.24507799029632302,
      "grad_norm": 0.3940922021865845,
      "learning_rate": 4.5225e-05,
      "loss": 0.7103,
      "step": 191
    },
    {
      "epoch": 0.2463611211355708,
      "grad_norm": 0.3621397912502289,
      "learning_rate": 4.52e-05,
      "loss": 0.8107,
      "step": 192
    },
    {
      "epoch": 0.24764425197481857,
      "grad_norm": 0.32553207874298096,
      "learning_rate": 4.5175e-05,
      "loss": 0.563,
      "step": 193
    },
    {
      "epoch": 0.24892738281406632,
      "grad_norm": 0.2594768702983856,
      "learning_rate": 4.5150000000000006e-05,
      "loss": 0.7747,
      "step": 194
    },
    {
      "epoch": 0.2502105136533141,
      "grad_norm": 0.22826555371284485,
      "learning_rate": 4.5125e-05,
      "loss": 0.6699,
      "step": 195
    },
    {
      "epoch": 0.25149364449256184,
      "grad_norm": 0.2798546850681305,
      "learning_rate": 4.5100000000000005e-05,
      "loss": 0.7919,
      "step": 196
    },
    {
      "epoch": 0.2527767753318096,
      "grad_norm": 0.21005459129810333,
      "learning_rate": 4.5075e-05,
      "loss": 0.7997,
      "step": 197
    },
    {
      "epoch": 0.2540599061710574,
      "grad_norm": 0.21063022315502167,
      "learning_rate": 4.5050000000000004e-05,
      "loss": 0.7078,
      "step": 198
    },
    {
      "epoch": 0.25534303701030514,
      "grad_norm": 0.22296057641506195,
      "learning_rate": 4.5025000000000003e-05,
      "loss": 0.6468,
      "step": 199
    },
    {
      "epoch": 0.2566261678495529,
      "grad_norm": 0.22070306539535522,
      "learning_rate": 4.5e-05,
      "loss": 0.7068,
      "step": 200
    },
    {
      "epoch": 0.2579092986888007,
      "grad_norm": 0.20852509140968323,
      "learning_rate": 4.4975e-05,
      "loss": 0.8155,
      "step": 201
    },
    {
      "epoch": 0.25919242952804844,
      "grad_norm": 0.2545461058616638,
      "learning_rate": 4.495e-05,
      "loss": 0.6878,
      "step": 202
    },
    {
      "epoch": 0.2604755603672962,
      "grad_norm": 0.24751250445842743,
      "learning_rate": 4.4925e-05,
      "loss": 0.6558,
      "step": 203
    },
    {
      "epoch": 0.261758691206544,
      "grad_norm": 0.20747962594032288,
      "learning_rate": 4.49e-05,
      "loss": 0.7053,
      "step": 204
    },
    {
      "epoch": 0.26304182204579174,
      "grad_norm": 0.27158552408218384,
      "learning_rate": 4.4875e-05,
      "loss": 0.7433,
      "step": 205
    },
    {
      "epoch": 0.2643249528850395,
      "grad_norm": 0.19930243492126465,
      "learning_rate": 4.4850000000000006e-05,
      "loss": 0.7224,
      "step": 206
    },
    {
      "epoch": 0.26560808372428724,
      "grad_norm": 0.1569337397813797,
      "learning_rate": 4.4825e-05,
      "loss": 0.6318,
      "step": 207
    },
    {
      "epoch": 0.26689121456353504,
      "grad_norm": 0.2139304280281067,
      "learning_rate": 4.4800000000000005e-05,
      "loss": 0.6774,
      "step": 208
    },
    {
      "epoch": 0.2681743454027828,
      "grad_norm": 0.38392388820648193,
      "learning_rate": 4.4775e-05,
      "loss": 0.701,
      "step": 209
    },
    {
      "epoch": 0.26945747624203054,
      "grad_norm": 0.18242083489894867,
      "learning_rate": 4.4750000000000004e-05,
      "loss": 0.7547,
      "step": 210
    },
    {
      "epoch": 0.27074060708127834,
      "grad_norm": 0.15443281829357147,
      "learning_rate": 4.4725e-05,
      "loss": 0.8482,
      "step": 211
    },
    {
      "epoch": 0.2720237379205261,
      "grad_norm": 0.4224720299243927,
      "learning_rate": 4.47e-05,
      "loss": 0.7356,
      "step": 212
    },
    {
      "epoch": 0.27330686875977384,
      "grad_norm": 0.6908435225486755,
      "learning_rate": 4.4675e-05,
      "loss": 0.7431,
      "step": 213
    },
    {
      "epoch": 0.27458999959902164,
      "grad_norm": 0.15524673461914062,
      "learning_rate": 4.465e-05,
      "loss": 0.7457,
      "step": 214
    },
    {
      "epoch": 0.2758731304382694,
      "grad_norm": 0.2470109611749649,
      "learning_rate": 4.4625e-05,
      "loss": 0.7559,
      "step": 215
    },
    {
      "epoch": 0.27715626127751714,
      "grad_norm": 0.23502202332019806,
      "learning_rate": 4.46e-05,
      "loss": 0.7246,
      "step": 216
    },
    {
      "epoch": 0.2784393921167649,
      "grad_norm": 0.10532274097204208,
      "learning_rate": 4.4575e-05,
      "loss": 0.6755,
      "step": 217
    },
    {
      "epoch": 0.2797225229560127,
      "grad_norm": 0.2412102222442627,
      "learning_rate": 4.4550000000000005e-05,
      "loss": 0.779,
      "step": 218
    },
    {
      "epoch": 0.28100565379526043,
      "grad_norm": 0.18583190441131592,
      "learning_rate": 4.4525e-05,
      "loss": 0.6665,
      "step": 219
    },
    {
      "epoch": 0.2822887846345082,
      "grad_norm": 0.43246787786483765,
      "learning_rate": 4.4500000000000004e-05,
      "loss": 0.7053,
      "step": 220
    },
    {
      "epoch": 0.283571915473756,
      "grad_norm": 0.16642989218235016,
      "learning_rate": 4.4475e-05,
      "loss": 0.7227,
      "step": 221
    },
    {
      "epoch": 0.28485504631300373,
      "grad_norm": 0.22487099468708038,
      "learning_rate": 4.445e-05,
      "loss": 0.6774,
      "step": 222
    },
    {
      "epoch": 0.2861381771522515,
      "grad_norm": 0.174720898270607,
      "learning_rate": 4.4425e-05,
      "loss": 0.6099,
      "step": 223
    },
    {
      "epoch": 0.28742130799149923,
      "grad_norm": 0.27207937836647034,
      "learning_rate": 4.44e-05,
      "loss": 0.6901,
      "step": 224
    },
    {
      "epoch": 0.28870443883074703,
      "grad_norm": 0.22528418898582458,
      "learning_rate": 4.4375e-05,
      "loss": 0.6917,
      "step": 225
    },
    {
      "epoch": 0.2899875696699948,
      "grad_norm": 0.19728706777095795,
      "learning_rate": 4.435e-05,
      "loss": 0.7252,
      "step": 226
    },
    {
      "epoch": 0.29127070050924253,
      "grad_norm": 0.5100868344306946,
      "learning_rate": 4.4325e-05,
      "loss": 0.7132,
      "step": 227
    },
    {
      "epoch": 0.29255383134849033,
      "grad_norm": 0.1497899442911148,
      "learning_rate": 4.43e-05,
      "loss": 0.7294,
      "step": 228
    },
    {
      "epoch": 0.2938369621877381,
      "grad_norm": 0.12238690257072449,
      "learning_rate": 4.4275e-05,
      "loss": 0.7316,
      "step": 229
    },
    {
      "epoch": 0.29512009302698583,
      "grad_norm": 0.13851596415042877,
      "learning_rate": 4.4250000000000005e-05,
      "loss": 0.717,
      "step": 230
    },
    {
      "epoch": 0.29640322386623363,
      "grad_norm": 0.12655270099639893,
      "learning_rate": 4.4225e-05,
      "loss": 0.7203,
      "step": 231
    },
    {
      "epoch": 0.2976863547054814,
      "grad_norm": 0.3238154649734497,
      "learning_rate": 4.4200000000000004e-05,
      "loss": 0.866,
      "step": 232
    },
    {
      "epoch": 0.29896948554472913,
      "grad_norm": 0.2502175271511078,
      "learning_rate": 4.4174999999999996e-05,
      "loss": 0.7352,
      "step": 233
    },
    {
      "epoch": 0.3002526163839769,
      "grad_norm": 0.2580633759498596,
      "learning_rate": 4.415e-05,
      "loss": 0.7971,
      "step": 234
    },
    {
      "epoch": 0.3015357472232247,
      "grad_norm": 0.174432635307312,
      "learning_rate": 4.4125e-05,
      "loss": 0.6533,
      "step": 235
    },
    {
      "epoch": 0.30281887806247243,
      "grad_norm": 0.12130558490753174,
      "learning_rate": 4.41e-05,
      "loss": 0.6119,
      "step": 236
    },
    {
      "epoch": 0.3041020089017202,
      "grad_norm": 0.1885250359773636,
      "learning_rate": 4.4075e-05,
      "loss": 0.6652,
      "step": 237
    },
    {
      "epoch": 0.305385139740968,
      "grad_norm": 0.3832365870475769,
      "learning_rate": 4.405e-05,
      "loss": 0.7246,
      "step": 238
    },
    {
      "epoch": 0.30666827058021573,
      "grad_norm": 0.15098263323307037,
      "learning_rate": 4.4025e-05,
      "loss": 0.7255,
      "step": 239
    },
    {
      "epoch": 0.3079514014194635,
      "grad_norm": 0.22306117415428162,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.6433,
      "step": 240
    },
    {
      "epoch": 0.3092345322587113,
      "grad_norm": 0.29963889718055725,
      "learning_rate": 4.3975e-05,
      "loss": 0.7924,
      "step": 241
    },
    {
      "epoch": 0.31051766309795903,
      "grad_norm": 0.2793080806732178,
      "learning_rate": 4.3950000000000004e-05,
      "loss": 0.6911,
      "step": 242
    },
    {
      "epoch": 0.3118007939372068,
      "grad_norm": 0.1941698044538498,
      "learning_rate": 4.3925e-05,
      "loss": 0.6497,
      "step": 243
    },
    {
      "epoch": 0.3130839247764545,
      "grad_norm": 0.16749228537082672,
      "learning_rate": 4.39e-05,
      "loss": 0.7619,
      "step": 244
    },
    {
      "epoch": 0.31436705561570233,
      "grad_norm": 0.22296510636806488,
      "learning_rate": 4.3875e-05,
      "loss": 0.7228,
      "step": 245
    },
    {
      "epoch": 0.3156501864549501,
      "grad_norm": 0.10758569836616516,
      "learning_rate": 4.385e-05,
      "loss": 0.7063,
      "step": 246
    },
    {
      "epoch": 0.3169333172941978,
      "grad_norm": 0.11221723258495331,
      "learning_rate": 4.3825e-05,
      "loss": 0.7788,
      "step": 247
    },
    {
      "epoch": 0.31821644813344563,
      "grad_norm": 0.1285574585199356,
      "learning_rate": 4.38e-05,
      "loss": 0.7237,
      "step": 248
    },
    {
      "epoch": 0.3194995789726934,
      "grad_norm": 0.17621482908725739,
      "learning_rate": 4.3775e-05,
      "loss": 0.776,
      "step": 249
    },
    {
      "epoch": 0.3207827098119411,
      "grad_norm": 0.23940083384513855,
      "learning_rate": 4.375e-05,
      "loss": 0.6544,
      "step": 250
    },
    {
      "epoch": 0.32206584065118893,
      "grad_norm": 0.18219539523124695,
      "learning_rate": 4.3725000000000006e-05,
      "loss": 0.7052,
      "step": 251
    },
    {
      "epoch": 0.3233489714904367,
      "grad_norm": 0.2235192358493805,
      "learning_rate": 4.3700000000000005e-05,
      "loss": 0.7341,
      "step": 252
    },
    {
      "epoch": 0.3246321023296844,
      "grad_norm": 0.12257345765829086,
      "learning_rate": 4.3675000000000005e-05,
      "loss": 0.6985,
      "step": 253
    },
    {
      "epoch": 0.3259152331689322,
      "grad_norm": 0.13306491076946259,
      "learning_rate": 4.3650000000000004e-05,
      "loss": 0.6349,
      "step": 254
    },
    {
      "epoch": 0.32719836400818,
      "grad_norm": 0.14606040716171265,
      "learning_rate": 4.3625e-05,
      "loss": 0.6852,
      "step": 255
    },
    {
      "epoch": 0.3284814948474277,
      "grad_norm": 0.15417885780334473,
      "learning_rate": 4.36e-05,
      "loss": 0.7388,
      "step": 256
    },
    {
      "epoch": 0.32976462568667547,
      "grad_norm": 0.14008395373821259,
      "learning_rate": 4.3575e-05,
      "loss": 0.7375,
      "step": 257
    },
    {
      "epoch": 0.3310477565259233,
      "grad_norm": 0.09606937319040298,
      "learning_rate": 4.355e-05,
      "loss": 0.6241,
      "step": 258
    },
    {
      "epoch": 0.332330887365171,
      "grad_norm": 0.11622468382120132,
      "learning_rate": 4.352500000000001e-05,
      "loss": 0.6454,
      "step": 259
    },
    {
      "epoch": 0.33361401820441877,
      "grad_norm": 0.17620514333248138,
      "learning_rate": 4.35e-05,
      "loss": 0.7258,
      "step": 260
    },
    {
      "epoch": 0.3348971490436665,
      "grad_norm": 0.124411441385746,
      "learning_rate": 4.3475000000000006e-05,
      "loss": 0.8191,
      "step": 261
    },
    {
      "epoch": 0.3361802798829143,
      "grad_norm": 0.12266983836889267,
      "learning_rate": 4.345e-05,
      "loss": 0.6128,
      "step": 262
    },
    {
      "epoch": 0.33746341072216207,
      "grad_norm": 0.10693590342998505,
      "learning_rate": 4.3425000000000005e-05,
      "loss": 0.754,
      "step": 263
    },
    {
      "epoch": 0.3387465415614098,
      "grad_norm": 0.13327822089195251,
      "learning_rate": 4.3400000000000005e-05,
      "loss": 0.7114,
      "step": 264
    },
    {
      "epoch": 0.3400296724006576,
      "grad_norm": 0.12884409725666046,
      "learning_rate": 4.3375000000000004e-05,
      "loss": 0.6084,
      "step": 265
    },
    {
      "epoch": 0.34131280323990537,
      "grad_norm": 0.12179514020681381,
      "learning_rate": 4.335e-05,
      "loss": 0.6761,
      "step": 266
    },
    {
      "epoch": 0.3425959340791531,
      "grad_norm": 0.12201149761676788,
      "learning_rate": 4.3325e-05,
      "loss": 0.7349,
      "step": 267
    },
    {
      "epoch": 0.3438790649184009,
      "grad_norm": 0.13721981644630432,
      "learning_rate": 4.33e-05,
      "loss": 0.7206,
      "step": 268
    },
    {
      "epoch": 0.34516219575764867,
      "grad_norm": 0.13685637712478638,
      "learning_rate": 4.3275e-05,
      "loss": 0.729,
      "step": 269
    },
    {
      "epoch": 0.3464453265968964,
      "grad_norm": 0.13976940512657166,
      "learning_rate": 4.325e-05,
      "loss": 0.6914,
      "step": 270
    },
    {
      "epoch": 0.34772845743614417,
      "grad_norm": 0.11607019603252411,
      "learning_rate": 4.322500000000001e-05,
      "loss": 0.7433,
      "step": 271
    },
    {
      "epoch": 0.34901158827539197,
      "grad_norm": 0.09183208644390106,
      "learning_rate": 4.32e-05,
      "loss": 0.7288,
      "step": 272
    },
    {
      "epoch": 0.3502947191146397,
      "grad_norm": 0.15023820102214813,
      "learning_rate": 4.3175000000000006e-05,
      "loss": 0.8108,
      "step": 273
    },
    {
      "epoch": 0.35157784995388747,
      "grad_norm": 0.09322658181190491,
      "learning_rate": 4.315e-05,
      "loss": 0.5655,
      "step": 274
    },
    {
      "epoch": 0.35286098079313527,
      "grad_norm": 0.1667962372303009,
      "learning_rate": 4.3125000000000005e-05,
      "loss": 0.788,
      "step": 275
    },
    {
      "epoch": 0.354144111632383,
      "grad_norm": 0.10862076282501221,
      "learning_rate": 4.3100000000000004e-05,
      "loss": 0.6886,
      "step": 276
    },
    {
      "epoch": 0.35542724247163077,
      "grad_norm": 0.13779810070991516,
      "learning_rate": 4.3075000000000003e-05,
      "loss": 0.7352,
      "step": 277
    },
    {
      "epoch": 0.35671037331087857,
      "grad_norm": 0.14231543242931366,
      "learning_rate": 4.305e-05,
      "loss": 0.684,
      "step": 278
    },
    {
      "epoch": 0.3579935041501263,
      "grad_norm": 0.12743398547172546,
      "learning_rate": 4.3025e-05,
      "loss": 0.7913,
      "step": 279
    },
    {
      "epoch": 0.35927663498937407,
      "grad_norm": 0.16025429964065552,
      "learning_rate": 4.3e-05,
      "loss": 0.8285,
      "step": 280
    },
    {
      "epoch": 0.3605597658286218,
      "grad_norm": 0.09905673563480377,
      "learning_rate": 4.2975e-05,
      "loss": 0.6566,
      "step": 281
    },
    {
      "epoch": 0.3618428966678696,
      "grad_norm": 0.14399418234825134,
      "learning_rate": 4.295e-05,
      "loss": 0.7017,
      "step": 282
    },
    {
      "epoch": 0.36312602750711737,
      "grad_norm": 0.14603625237941742,
      "learning_rate": 4.2925000000000007e-05,
      "loss": 0.8207,
      "step": 283
    },
    {
      "epoch": 0.3644091583463651,
      "grad_norm": 0.10519102215766907,
      "learning_rate": 4.29e-05,
      "loss": 0.7816,
      "step": 284
    },
    {
      "epoch": 0.3656922891856129,
      "grad_norm": 0.16130349040031433,
      "learning_rate": 4.2875000000000005e-05,
      "loss": 0.8411,
      "step": 285
    },
    {
      "epoch": 0.36697542002486067,
      "grad_norm": 0.12829618155956268,
      "learning_rate": 4.285e-05,
      "loss": 0.7287,
      "step": 286
    },
    {
      "epoch": 0.3682585508641084,
      "grad_norm": 0.13783273100852966,
      "learning_rate": 4.2825000000000004e-05,
      "loss": 0.6557,
      "step": 287
    },
    {
      "epoch": 0.36954168170335616,
      "grad_norm": 0.2266722321510315,
      "learning_rate": 4.2800000000000004e-05,
      "loss": 0.7901,
      "step": 288
    },
    {
      "epoch": 0.37082481254260397,
      "grad_norm": 0.20219266414642334,
      "learning_rate": 4.2775e-05,
      "loss": 0.8339,
      "step": 289
    },
    {
      "epoch": 0.3721079433818517,
      "grad_norm": 0.14629176259040833,
      "learning_rate": 4.275e-05,
      "loss": 0.6839,
      "step": 290
    },
    {
      "epoch": 0.37339107422109946,
      "grad_norm": 0.13952559232711792,
      "learning_rate": 4.2725e-05,
      "loss": 0.7236,
      "step": 291
    },
    {
      "epoch": 0.37467420506034727,
      "grad_norm": 0.09744761884212494,
      "learning_rate": 4.27e-05,
      "loss": 0.6416,
      "step": 292
    },
    {
      "epoch": 0.375957335899595,
      "grad_norm": 0.13428550958633423,
      "learning_rate": 4.2675e-05,
      "loss": 0.7371,
      "step": 293
    },
    {
      "epoch": 0.37724046673884276,
      "grad_norm": 0.10251361131668091,
      "learning_rate": 4.265e-05,
      "loss": 0.7959,
      "step": 294
    },
    {
      "epoch": 0.37852359757809056,
      "grad_norm": 0.10133925825357437,
      "learning_rate": 4.2625000000000006e-05,
      "loss": 0.6989,
      "step": 295
    },
    {
      "epoch": 0.3798067284173383,
      "grad_norm": 0.13307690620422363,
      "learning_rate": 4.26e-05,
      "loss": 0.7686,
      "step": 296
    },
    {
      "epoch": 0.38108985925658606,
      "grad_norm": 0.09745471924543381,
      "learning_rate": 4.2575000000000005e-05,
      "loss": 0.7217,
      "step": 297
    },
    {
      "epoch": 0.3823729900958338,
      "grad_norm": 0.1382995992898941,
      "learning_rate": 4.2550000000000004e-05,
      "loss": 0.7169,
      "step": 298
    },
    {
      "epoch": 0.3836561209350816,
      "grad_norm": 0.10830988734960556,
      "learning_rate": 4.2525000000000004e-05,
      "loss": 0.743,
      "step": 299
    },
    {
      "epoch": 0.38493925177432936,
      "grad_norm": 0.12490814924240112,
      "learning_rate": 4.25e-05,
      "loss": 0.6781,
      "step": 300
    },
    {
      "epoch": 0.3862223826135771,
      "grad_norm": 0.14296282827854156,
      "learning_rate": 4.2475e-05,
      "loss": 0.6795,
      "step": 301
    },
    {
      "epoch": 0.3875055134528249,
      "grad_norm": 1.190176010131836,
      "learning_rate": 4.245e-05,
      "loss": 0.6589,
      "step": 302
    },
    {
      "epoch": 0.38878864429207266,
      "grad_norm": 0.1819627583026886,
      "learning_rate": 4.2425e-05,
      "loss": 0.7022,
      "step": 303
    },
    {
      "epoch": 0.3900717751313204,
      "grad_norm": 0.10687839984893799,
      "learning_rate": 4.24e-05,
      "loss": 0.7268,
      "step": 304
    },
    {
      "epoch": 0.3913549059705682,
      "grad_norm": 0.21718841791152954,
      "learning_rate": 4.237500000000001e-05,
      "loss": 0.6079,
      "step": 305
    },
    {
      "epoch": 0.39263803680981596,
      "grad_norm": 0.16769003868103027,
      "learning_rate": 4.235e-05,
      "loss": 0.6945,
      "step": 306
    },
    {
      "epoch": 0.3939211676490637,
      "grad_norm": 0.12068881094455719,
      "learning_rate": 4.2325000000000006e-05,
      "loss": 0.6757,
      "step": 307
    },
    {
      "epoch": 0.39520429848831146,
      "grad_norm": 0.10158878564834595,
      "learning_rate": 4.23e-05,
      "loss": 0.637,
      "step": 308
    },
    {
      "epoch": 0.39648742932755926,
      "grad_norm": 0.13318391144275665,
      "learning_rate": 4.2275000000000004e-05,
      "loss": 0.7597,
      "step": 309
    },
    {
      "epoch": 0.397770560166807,
      "grad_norm": 0.10585636645555496,
      "learning_rate": 4.2250000000000004e-05,
      "loss": 0.6586,
      "step": 310
    },
    {
      "epoch": 0.39905369100605476,
      "grad_norm": 0.13218839466571808,
      "learning_rate": 4.2225e-05,
      "loss": 0.7596,
      "step": 311
    },
    {
      "epoch": 0.40033682184530256,
      "grad_norm": 0.16066090762615204,
      "learning_rate": 4.22e-05,
      "loss": 0.7112,
      "step": 312
    },
    {
      "epoch": 0.4016199526845503,
      "grad_norm": 0.12629477679729462,
      "learning_rate": 4.2175e-05,
      "loss": 0.7511,
      "step": 313
    },
    {
      "epoch": 0.40290308352379806,
      "grad_norm": 0.11707097291946411,
      "learning_rate": 4.215e-05,
      "loss": 0.6702,
      "step": 314
    },
    {
      "epoch": 0.40418621436304586,
      "grad_norm": 0.09740698337554932,
      "learning_rate": 4.2125e-05,
      "loss": 0.6316,
      "step": 315
    },
    {
      "epoch": 0.4054693452022936,
      "grad_norm": 0.44215667247772217,
      "learning_rate": 4.21e-05,
      "loss": 0.7108,
      "step": 316
    },
    {
      "epoch": 0.40675247604154136,
      "grad_norm": 0.1296062022447586,
      "learning_rate": 4.2075000000000006e-05,
      "loss": 0.7188,
      "step": 317
    },
    {
      "epoch": 0.4080356068807891,
      "grad_norm": 0.12833233177661896,
      "learning_rate": 4.205e-05,
      "loss": 0.6451,
      "step": 318
    },
    {
      "epoch": 0.4093187377200369,
      "grad_norm": 0.1068439707159996,
      "learning_rate": 4.2025000000000005e-05,
      "loss": 0.6595,
      "step": 319
    },
    {
      "epoch": 0.41060186855928466,
      "grad_norm": 0.3968654274940491,
      "learning_rate": 4.2e-05,
      "loss": 0.7609,
      "step": 320
    },
    {
      "epoch": 0.4118849993985324,
      "grad_norm": 0.10848594456911087,
      "learning_rate": 4.1975000000000004e-05,
      "loss": 0.738,
      "step": 321
    },
    {
      "epoch": 0.4131681302377802,
      "grad_norm": 0.2635965049266815,
      "learning_rate": 4.195e-05,
      "loss": 0.8273,
      "step": 322
    },
    {
      "epoch": 0.41445126107702795,
      "grad_norm": 0.11041995882987976,
      "learning_rate": 4.1925e-05,
      "loss": 0.7257,
      "step": 323
    },
    {
      "epoch": 0.4157343919162757,
      "grad_norm": 0.11466779559850693,
      "learning_rate": 4.19e-05,
      "loss": 0.7222,
      "step": 324
    },
    {
      "epoch": 0.41701752275552345,
      "grad_norm": 0.12307105213403702,
      "learning_rate": 4.1875e-05,
      "loss": 0.7541,
      "step": 325
    },
    {
      "epoch": 0.41830065359477125,
      "grad_norm": 0.1996818482875824,
      "learning_rate": 4.185e-05,
      "loss": 0.6577,
      "step": 326
    },
    {
      "epoch": 0.419583784434019,
      "grad_norm": 0.13063374161720276,
      "learning_rate": 4.1825e-05,
      "loss": 0.6566,
      "step": 327
    },
    {
      "epoch": 0.42086691527326675,
      "grad_norm": 0.1197972446680069,
      "learning_rate": 4.18e-05,
      "loss": 0.783,
      "step": 328
    },
    {
      "epoch": 0.42215004611251455,
      "grad_norm": 0.1051694005727768,
      "learning_rate": 4.1775000000000006e-05,
      "loss": 0.737,
      "step": 329
    },
    {
      "epoch": 0.4234331769517623,
      "grad_norm": 0.26117968559265137,
      "learning_rate": 4.175e-05,
      "loss": 0.6978,
      "step": 330
    },
    {
      "epoch": 0.42471630779101005,
      "grad_norm": 0.13131479918956757,
      "learning_rate": 4.1725000000000005e-05,
      "loss": 0.6617,
      "step": 331
    },
    {
      "epoch": 0.42599943863025785,
      "grad_norm": 0.12850546836853027,
      "learning_rate": 4.17e-05,
      "loss": 0.718,
      "step": 332
    },
    {
      "epoch": 0.4272825694695056,
      "grad_norm": 0.228835329413414,
      "learning_rate": 4.1675e-05,
      "loss": 0.7535,
      "step": 333
    },
    {
      "epoch": 0.42856570030875335,
      "grad_norm": 0.10166200995445251,
      "learning_rate": 4.165e-05,
      "loss": 0.732,
      "step": 334
    },
    {
      "epoch": 0.4298488311480011,
      "grad_norm": 0.09312116354703903,
      "learning_rate": 4.1625e-05,
      "loss": 0.6502,
      "step": 335
    },
    {
      "epoch": 0.4311319619872489,
      "grad_norm": 0.2288721799850464,
      "learning_rate": 4.16e-05,
      "loss": 0.7085,
      "step": 336
    },
    {
      "epoch": 0.43241509282649665,
      "grad_norm": 0.09574178606271744,
      "learning_rate": 4.1575e-05,
      "loss": 0.6957,
      "step": 337
    },
    {
      "epoch": 0.4336982236657444,
      "grad_norm": 0.12139047682285309,
      "learning_rate": 4.155e-05,
      "loss": 0.7138,
      "step": 338
    },
    {
      "epoch": 0.4349813545049922,
      "grad_norm": 0.15936976671218872,
      "learning_rate": 4.1525e-05,
      "loss": 0.6967,
      "step": 339
    },
    {
      "epoch": 0.43626448534423995,
      "grad_norm": 0.09337805211544037,
      "learning_rate": 4.15e-05,
      "loss": 0.6865,
      "step": 340
    },
    {
      "epoch": 0.4375476161834877,
      "grad_norm": 0.1254083812236786,
      "learning_rate": 4.1475000000000005e-05,
      "loss": 0.6319,
      "step": 341
    },
    {
      "epoch": 0.4388307470227355,
      "grad_norm": 0.15749913454055786,
      "learning_rate": 4.145e-05,
      "loss": 0.7653,
      "step": 342
    },
    {
      "epoch": 0.44011387786198325,
      "grad_norm": 0.13738861680030823,
      "learning_rate": 4.1425000000000004e-05,
      "loss": 0.687,
      "step": 343
    },
    {
      "epoch": 0.441397008701231,
      "grad_norm": 0.11583111435174942,
      "learning_rate": 4.14e-05,
      "loss": 0.5889,
      "step": 344
    },
    {
      "epoch": 0.44268013954047875,
      "grad_norm": 0.13357993960380554,
      "learning_rate": 4.1375e-05,
      "loss": 0.6581,
      "step": 345
    },
    {
      "epoch": 0.44396327037972655,
      "grad_norm": 0.364935964345932,
      "learning_rate": 4.135e-05,
      "loss": 0.6958,
      "step": 346
    },
    {
      "epoch": 0.4452464012189743,
      "grad_norm": 0.10726168006658554,
      "learning_rate": 4.1325e-05,
      "loss": 0.7682,
      "step": 347
    },
    {
      "epoch": 0.44652953205822205,
      "grad_norm": 0.10372709482908249,
      "learning_rate": 4.13e-05,
      "loss": 0.7375,
      "step": 348
    },
    {
      "epoch": 0.44781266289746985,
      "grad_norm": 0.6129089593887329,
      "learning_rate": 4.1275e-05,
      "loss": 0.6867,
      "step": 349
    },
    {
      "epoch": 0.4490957937367176,
      "grad_norm": 0.09958565980195999,
      "learning_rate": 4.125e-05,
      "loss": 0.5806,
      "step": 350
    },
    {
      "epoch": 0.45037892457596534,
      "grad_norm": 0.15060438215732574,
      "learning_rate": 4.1225e-05,
      "loss": 0.5737,
      "step": 351
    },
    {
      "epoch": 0.4516620554152131,
      "grad_norm": 0.11395902186632156,
      "learning_rate": 4.12e-05,
      "loss": 0.7295,
      "step": 352
    },
    {
      "epoch": 0.4529451862544609,
      "grad_norm": 0.10930227488279343,
      "learning_rate": 4.1175000000000005e-05,
      "loss": 0.6333,
      "step": 353
    },
    {
      "epoch": 0.45422831709370864,
      "grad_norm": 0.11299290508031845,
      "learning_rate": 4.115e-05,
      "loss": 0.7293,
      "step": 354
    },
    {
      "epoch": 0.4555114479329564,
      "grad_norm": 0.09619927406311035,
      "learning_rate": 4.1125000000000004e-05,
      "loss": 0.6951,
      "step": 355
    },
    {
      "epoch": 0.4567945787722042,
      "grad_norm": 0.12364543974399567,
      "learning_rate": 4.11e-05,
      "loss": 0.6802,
      "step": 356
    },
    {
      "epoch": 0.45807770961145194,
      "grad_norm": 0.10955870151519775,
      "learning_rate": 4.1075e-05,
      "loss": 0.7237,
      "step": 357
    },
    {
      "epoch": 0.4593608404506997,
      "grad_norm": 0.10276415199041367,
      "learning_rate": 4.105e-05,
      "loss": 0.8011,
      "step": 358
    },
    {
      "epoch": 0.4606439712899475,
      "grad_norm": 0.10726416856050491,
      "learning_rate": 4.1025e-05,
      "loss": 0.6751,
      "step": 359
    },
    {
      "epoch": 0.46192710212919524,
      "grad_norm": 0.15864935517311096,
      "learning_rate": 4.1e-05,
      "loss": 0.695,
      "step": 360
    },
    {
      "epoch": 0.463210232968443,
      "grad_norm": 0.10666662454605103,
      "learning_rate": 4.0975e-05,
      "loss": 0.7144,
      "step": 361
    },
    {
      "epoch": 0.46449336380769074,
      "grad_norm": 0.11262679100036621,
      "learning_rate": 4.095e-05,
      "loss": 0.7272,
      "step": 362
    },
    {
      "epoch": 0.46577649464693854,
      "grad_norm": 0.11993991583585739,
      "learning_rate": 4.0925000000000005e-05,
      "loss": 0.7247,
      "step": 363
    },
    {
      "epoch": 0.4670596254861863,
      "grad_norm": 0.13503335416316986,
      "learning_rate": 4.09e-05,
      "loss": 0.7171,
      "step": 364
    },
    {
      "epoch": 0.46834275632543404,
      "grad_norm": 0.0916631668806076,
      "learning_rate": 4.0875000000000004e-05,
      "loss": 0.692,
      "step": 365
    },
    {
      "epoch": 0.46962588716468184,
      "grad_norm": 0.10188502818346024,
      "learning_rate": 4.085e-05,
      "loss": 0.5619,
      "step": 366
    },
    {
      "epoch": 0.4709090180039296,
      "grad_norm": 0.1331438422203064,
      "learning_rate": 4.0825e-05,
      "loss": 0.7865,
      "step": 367
    },
    {
      "epoch": 0.47219214884317734,
      "grad_norm": 0.10209344327449799,
      "learning_rate": 4.08e-05,
      "loss": 0.723,
      "step": 368
    },
    {
      "epoch": 0.47347527968242514,
      "grad_norm": 0.12400554865598679,
      "learning_rate": 4.0775e-05,
      "loss": 0.6532,
      "step": 369
    },
    {
      "epoch": 0.4747584105216729,
      "grad_norm": 0.10274975746870041,
      "learning_rate": 4.075e-05,
      "loss": 0.6542,
      "step": 370
    },
    {
      "epoch": 0.47604154136092064,
      "grad_norm": 0.11357192695140839,
      "learning_rate": 4.0725e-05,
      "loss": 0.7189,
      "step": 371
    },
    {
      "epoch": 0.4773246722001684,
      "grad_norm": 0.11296133697032928,
      "learning_rate": 4.07e-05,
      "loss": 0.6672,
      "step": 372
    },
    {
      "epoch": 0.4786078030394162,
      "grad_norm": 0.0833190530538559,
      "learning_rate": 4.0675e-05,
      "loss": 0.6506,
      "step": 373
    },
    {
      "epoch": 0.47989093387866394,
      "grad_norm": 0.0922357589006424,
      "learning_rate": 4.065e-05,
      "loss": 0.5853,
      "step": 374
    },
    {
      "epoch": 0.4811740647179117,
      "grad_norm": 0.10162580013275146,
      "learning_rate": 4.0625000000000005e-05,
      "loss": 0.6347,
      "step": 375
    },
    {
      "epoch": 0.4824571955571595,
      "grad_norm": 0.2291833907365799,
      "learning_rate": 4.0600000000000004e-05,
      "loss": 0.6597,
      "step": 376
    },
    {
      "epoch": 0.48374032639640724,
      "grad_norm": 0.14269761741161346,
      "learning_rate": 4.0575000000000004e-05,
      "loss": 0.7278,
      "step": 377
    },
    {
      "epoch": 0.485023457235655,
      "grad_norm": 0.1148926168680191,
      "learning_rate": 4.055e-05,
      "loss": 0.6746,
      "step": 378
    },
    {
      "epoch": 0.4863065880749028,
      "grad_norm": 0.13270127773284912,
      "learning_rate": 4.0525e-05,
      "loss": 0.7616,
      "step": 379
    },
    {
      "epoch": 0.48758971891415054,
      "grad_norm": 0.10678492486476898,
      "learning_rate": 4.05e-05,
      "loss": 0.5618,
      "step": 380
    },
    {
      "epoch": 0.4888728497533983,
      "grad_norm": 0.10953043401241302,
      "learning_rate": 4.0475e-05,
      "loss": 0.715,
      "step": 381
    },
    {
      "epoch": 0.49015598059264603,
      "grad_norm": 0.18458960950374603,
      "learning_rate": 4.045000000000001e-05,
      "loss": 0.732,
      "step": 382
    },
    {
      "epoch": 0.49143911143189384,
      "grad_norm": 0.16866900026798248,
      "learning_rate": 4.0425e-05,
      "loss": 0.6466,
      "step": 383
    },
    {
      "epoch": 0.4927222422711416,
      "grad_norm": 0.12769177556037903,
      "learning_rate": 4.0400000000000006e-05,
      "loss": 0.7613,
      "step": 384
    },
    {
      "epoch": 0.49400537311038933,
      "grad_norm": 0.1298220455646515,
      "learning_rate": 4.0375e-05,
      "loss": 0.8113,
      "step": 385
    },
    {
      "epoch": 0.49528850394963714,
      "grad_norm": 0.1262449324131012,
      "learning_rate": 4.0350000000000005e-05,
      "loss": 0.6595,
      "step": 386
    },
    {
      "epoch": 0.4965716347888849,
      "grad_norm": 0.13865424692630768,
      "learning_rate": 4.0325000000000004e-05,
      "loss": 0.8089,
      "step": 387
    },
    {
      "epoch": 0.49785476562813263,
      "grad_norm": 0.11206932365894318,
      "learning_rate": 4.0300000000000004e-05,
      "loss": 0.7074,
      "step": 388
    },
    {
      "epoch": 0.4991378964673804,
      "grad_norm": 0.3256579339504242,
      "learning_rate": 4.0275e-05,
      "loss": 0.7009,
      "step": 389
    },
    {
      "epoch": 0.5004210273066282,
      "grad_norm": 0.11995213478803635,
      "learning_rate": 4.025e-05,
      "loss": 0.6509,
      "step": 390
    },
    {
      "epoch": 0.501704158145876,
      "grad_norm": 0.16177313029766083,
      "learning_rate": 4.0225e-05,
      "loss": 0.6049,
      "step": 391
    },
    {
      "epoch": 0.5029872889851237,
      "grad_norm": 0.10474003106355667,
      "learning_rate": 4.02e-05,
      "loss": 0.7406,
      "step": 392
    },
    {
      "epoch": 0.5042704198243715,
      "grad_norm": 0.10574903339147568,
      "learning_rate": 4.0175e-05,
      "loss": 0.6689,
      "step": 393
    },
    {
      "epoch": 0.5055535506636192,
      "grad_norm": 0.09745536744594574,
      "learning_rate": 4.015000000000001e-05,
      "loss": 0.6656,
      "step": 394
    },
    {
      "epoch": 0.506836681502867,
      "grad_norm": 0.10704592615365982,
      "learning_rate": 4.0125e-05,
      "loss": 0.6878,
      "step": 395
    },
    {
      "epoch": 0.5081198123421148,
      "grad_norm": 0.11414460837841034,
      "learning_rate": 4.0100000000000006e-05,
      "loss": 0.7297,
      "step": 396
    },
    {
      "epoch": 0.5094029431813625,
      "grad_norm": 0.1069076731801033,
      "learning_rate": 4.0075e-05,
      "loss": 0.7759,
      "step": 397
    },
    {
      "epoch": 0.5106860740206103,
      "grad_norm": 0.11225546151399612,
      "learning_rate": 4.0050000000000004e-05,
      "loss": 0.6959,
      "step": 398
    },
    {
      "epoch": 0.5119692048598581,
      "grad_norm": 0.10544376075267792,
      "learning_rate": 4.0025000000000004e-05,
      "loss": 0.6525,
      "step": 399
    },
    {
      "epoch": 0.5132523356991058,
      "grad_norm": 0.09464980661869049,
      "learning_rate": 4e-05,
      "loss": 0.6935,
      "step": 400
    },
    {
      "epoch": 0.5145354665383536,
      "grad_norm": 0.10676474869251251,
      "learning_rate": 3.9975e-05,
      "loss": 0.6782,
      "step": 401
    },
    {
      "epoch": 0.5158185973776014,
      "grad_norm": 0.7250363826751709,
      "learning_rate": 3.995e-05,
      "loss": 0.6993,
      "step": 402
    },
    {
      "epoch": 0.5171017282168491,
      "grad_norm": 0.11197555065155029,
      "learning_rate": 3.9925e-05,
      "loss": 0.7675,
      "step": 403
    },
    {
      "epoch": 0.5183848590560969,
      "grad_norm": 0.09306884557008743,
      "learning_rate": 3.99e-05,
      "loss": 0.6595,
      "step": 404
    },
    {
      "epoch": 0.5196679898953447,
      "grad_norm": 0.09707053005695343,
      "learning_rate": 3.9875e-05,
      "loss": 0.6583,
      "step": 405
    },
    {
      "epoch": 0.5209511207345924,
      "grad_norm": 0.12495215982198715,
      "learning_rate": 3.9850000000000006e-05,
      "loss": 0.7006,
      "step": 406
    },
    {
      "epoch": 0.5222342515738402,
      "grad_norm": 0.13880349695682526,
      "learning_rate": 3.9825e-05,
      "loss": 0.5929,
      "step": 407
    },
    {
      "epoch": 0.523517382413088,
      "grad_norm": 0.11584901064634323,
      "learning_rate": 3.9800000000000005e-05,
      "loss": 0.7288,
      "step": 408
    },
    {
      "epoch": 0.5248005132523357,
      "grad_norm": 0.12831082940101624,
      "learning_rate": 3.9775e-05,
      "loss": 0.6866,
      "step": 409
    },
    {
      "epoch": 0.5260836440915835,
      "grad_norm": 0.14364874362945557,
      "learning_rate": 3.9750000000000004e-05,
      "loss": 0.6733,
      "step": 410
    },
    {
      "epoch": 0.5273667749308313,
      "grad_norm": 0.11207055300474167,
      "learning_rate": 3.9725e-05,
      "loss": 0.7886,
      "step": 411
    },
    {
      "epoch": 0.528649905770079,
      "grad_norm": 0.12298090755939484,
      "learning_rate": 3.97e-05,
      "loss": 0.7573,
      "step": 412
    },
    {
      "epoch": 0.5299330366093268,
      "grad_norm": 0.12522552907466888,
      "learning_rate": 3.9675e-05,
      "loss": 0.6978,
      "step": 413
    },
    {
      "epoch": 0.5312161674485745,
      "grad_norm": 0.09684982895851135,
      "learning_rate": 3.965e-05,
      "loss": 0.6688,
      "step": 414
    },
    {
      "epoch": 0.5324992982878223,
      "grad_norm": 0.1034354642033577,
      "learning_rate": 3.9625e-05,
      "loss": 0.622,
      "step": 415
    },
    {
      "epoch": 0.5337824291270701,
      "grad_norm": 0.1495935320854187,
      "learning_rate": 3.960000000000001e-05,
      "loss": 0.714,
      "step": 416
    },
    {
      "epoch": 0.5350655599663178,
      "grad_norm": 0.11201269179582596,
      "learning_rate": 3.9575e-05,
      "loss": 0.6568,
      "step": 417
    },
    {
      "epoch": 0.5363486908055656,
      "grad_norm": 0.13177452981472015,
      "learning_rate": 3.9550000000000006e-05,
      "loss": 0.7075,
      "step": 418
    },
    {
      "epoch": 0.5376318216448134,
      "grad_norm": 0.09106247127056122,
      "learning_rate": 3.9525e-05,
      "loss": 0.6503,
      "step": 419
    },
    {
      "epoch": 0.5389149524840611,
      "grad_norm": 0.13225482404232025,
      "learning_rate": 3.9500000000000005e-05,
      "loss": 0.7562,
      "step": 420
    },
    {
      "epoch": 0.5401980833233089,
      "grad_norm": 0.13026607036590576,
      "learning_rate": 3.9475000000000004e-05,
      "loss": 0.5717,
      "step": 421
    },
    {
      "epoch": 0.5414812141625567,
      "grad_norm": 0.1064525842666626,
      "learning_rate": 3.9450000000000003e-05,
      "loss": 0.7271,
      "step": 422
    },
    {
      "epoch": 0.5427643450018044,
      "grad_norm": 0.08889277279376984,
      "learning_rate": 3.9425e-05,
      "loss": 0.6634,
      "step": 423
    },
    {
      "epoch": 0.5440474758410522,
      "grad_norm": 0.11130121350288391,
      "learning_rate": 3.94e-05,
      "loss": 0.613,
      "step": 424
    },
    {
      "epoch": 0.5453306066803,
      "grad_norm": 0.08470254391431808,
      "learning_rate": 3.9375e-05,
      "loss": 0.7527,
      "step": 425
    },
    {
      "epoch": 0.5466137375195477,
      "grad_norm": 0.09825795143842697,
      "learning_rate": 3.935e-05,
      "loss": 0.6724,
      "step": 426
    },
    {
      "epoch": 0.5478968683587955,
      "grad_norm": 0.133823961019516,
      "learning_rate": 3.9325e-05,
      "loss": 0.6495,
      "step": 427
    },
    {
      "epoch": 0.5491799991980433,
      "grad_norm": 0.14606650173664093,
      "learning_rate": 3.9300000000000007e-05,
      "loss": 0.7166,
      "step": 428
    },
    {
      "epoch": 0.550463130037291,
      "grad_norm": 1.0690324306488037,
      "learning_rate": 3.9275e-05,
      "loss": 0.748,
      "step": 429
    },
    {
      "epoch": 0.5517462608765388,
      "grad_norm": 0.1060820147395134,
      "learning_rate": 3.9250000000000005e-05,
      "loss": 0.7421,
      "step": 430
    },
    {
      "epoch": 0.5530293917157865,
      "grad_norm": 0.1108517199754715,
      "learning_rate": 3.9225e-05,
      "loss": 0.6704,
      "step": 431
    },
    {
      "epoch": 0.5543125225550343,
      "grad_norm": 0.12280447781085968,
      "learning_rate": 3.9200000000000004e-05,
      "loss": 0.7649,
      "step": 432
    },
    {
      "epoch": 0.5555956533942821,
      "grad_norm": 0.08868840336799622,
      "learning_rate": 3.9175000000000004e-05,
      "loss": 0.6241,
      "step": 433
    },
    {
      "epoch": 0.5568787842335298,
      "grad_norm": 0.16192899644374847,
      "learning_rate": 3.915e-05,
      "loss": 0.7131,
      "step": 434
    },
    {
      "epoch": 0.5581619150727776,
      "grad_norm": 0.10645978897809982,
      "learning_rate": 3.9125e-05,
      "loss": 0.7346,
      "step": 435
    },
    {
      "epoch": 0.5594450459120254,
      "grad_norm": 0.1020224466919899,
      "learning_rate": 3.91e-05,
      "loss": 0.6264,
      "step": 436
    },
    {
      "epoch": 0.5607281767512731,
      "grad_norm": 0.11578159779310226,
      "learning_rate": 3.9075e-05,
      "loss": 0.7315,
      "step": 437
    },
    {
      "epoch": 0.5620113075905209,
      "grad_norm": 0.09932518005371094,
      "learning_rate": 3.905e-05,
      "loss": 0.6656,
      "step": 438
    },
    {
      "epoch": 0.5632944384297687,
      "grad_norm": 0.12751439213752747,
      "learning_rate": 3.9025e-05,
      "loss": 0.7505,
      "step": 439
    },
    {
      "epoch": 0.5645775692690164,
      "grad_norm": 0.09480132907629013,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 0.759,
      "step": 440
    },
    {
      "epoch": 0.5658607001082642,
      "grad_norm": 0.1168651357293129,
      "learning_rate": 3.8975e-05,
      "loss": 0.7229,
      "step": 441
    },
    {
      "epoch": 0.567143830947512,
      "grad_norm": 0.27809938788414,
      "learning_rate": 3.8950000000000005e-05,
      "loss": 0.6962,
      "step": 442
    },
    {
      "epoch": 0.5684269617867597,
      "grad_norm": 0.11947769671678543,
      "learning_rate": 3.8925e-05,
      "loss": 0.6451,
      "step": 443
    },
    {
      "epoch": 0.5697100926260075,
      "grad_norm": 0.1496148705482483,
      "learning_rate": 3.8900000000000004e-05,
      "loss": 0.6345,
      "step": 444
    },
    {
      "epoch": 0.5709932234652553,
      "grad_norm": 0.12359619140625,
      "learning_rate": 3.8875e-05,
      "loss": 0.708,
      "step": 445
    },
    {
      "epoch": 0.572276354304503,
      "grad_norm": 0.11766932159662247,
      "learning_rate": 3.885e-05,
      "loss": 0.7756,
      "step": 446
    },
    {
      "epoch": 0.5735594851437508,
      "grad_norm": 0.1767989844083786,
      "learning_rate": 3.8825e-05,
      "loss": 0.7411,
      "step": 447
    },
    {
      "epoch": 0.5748426159829985,
      "grad_norm": 0.10538271069526672,
      "learning_rate": 3.88e-05,
      "loss": 0.7055,
      "step": 448
    },
    {
      "epoch": 0.5761257468222463,
      "grad_norm": 0.11386648565530777,
      "learning_rate": 3.8775e-05,
      "loss": 0.7333,
      "step": 449
    },
    {
      "epoch": 0.5774088776614941,
      "grad_norm": 0.10873893648386002,
      "learning_rate": 3.875e-05,
      "loss": 0.6584,
      "step": 450
    },
    {
      "epoch": 0.5786920085007418,
      "grad_norm": 0.11098892986774445,
      "learning_rate": 3.8725e-05,
      "loss": 0.6268,
      "step": 451
    },
    {
      "epoch": 0.5799751393399896,
      "grad_norm": 0.14231470227241516,
      "learning_rate": 3.8700000000000006e-05,
      "loss": 0.7082,
      "step": 452
    },
    {
      "epoch": 0.5812582701792374,
      "grad_norm": 0.09992900490760803,
      "learning_rate": 3.8675e-05,
      "loss": 0.7379,
      "step": 453
    },
    {
      "epoch": 0.5825414010184851,
      "grad_norm": 0.10456696152687073,
      "learning_rate": 3.8650000000000004e-05,
      "loss": 0.6899,
      "step": 454
    },
    {
      "epoch": 0.5838245318577329,
      "grad_norm": 0.14194917678833008,
      "learning_rate": 3.8625e-05,
      "loss": 0.6468,
      "step": 455
    },
    {
      "epoch": 0.5851076626969807,
      "grad_norm": 0.10396645218133926,
      "learning_rate": 3.86e-05,
      "loss": 0.718,
      "step": 456
    },
    {
      "epoch": 0.5863907935362284,
      "grad_norm": 0.13729867339134216,
      "learning_rate": 3.8575e-05,
      "loss": 0.6819,
      "step": 457
    },
    {
      "epoch": 0.5876739243754762,
      "grad_norm": 0.12848074734210968,
      "learning_rate": 3.855e-05,
      "loss": 0.6606,
      "step": 458
    },
    {
      "epoch": 0.588957055214724,
      "grad_norm": 0.13237309455871582,
      "learning_rate": 3.8525e-05,
      "loss": 0.7332,
      "step": 459
    },
    {
      "epoch": 0.5902401860539717,
      "grad_norm": 0.12253362685441971,
      "learning_rate": 3.85e-05,
      "loss": 0.694,
      "step": 460
    },
    {
      "epoch": 0.5915233168932195,
      "grad_norm": 0.09407396614551544,
      "learning_rate": 3.8475e-05,
      "loss": 0.7031,
      "step": 461
    },
    {
      "epoch": 0.5928064477324673,
      "grad_norm": 0.10881764441728592,
      "learning_rate": 3.845e-05,
      "loss": 0.667,
      "step": 462
    },
    {
      "epoch": 0.594089578571715,
      "grad_norm": 0.10429378598928452,
      "learning_rate": 3.8425e-05,
      "loss": 0.5487,
      "step": 463
    },
    {
      "epoch": 0.5953727094109628,
      "grad_norm": 0.10591240972280502,
      "learning_rate": 3.8400000000000005e-05,
      "loss": 0.5753,
      "step": 464
    },
    {
      "epoch": 0.5966558402502106,
      "grad_norm": 0.10311070084571838,
      "learning_rate": 3.8375e-05,
      "loss": 0.7177,
      "step": 465
    },
    {
      "epoch": 0.5979389710894583,
      "grad_norm": 0.10071313381195068,
      "learning_rate": 3.8350000000000004e-05,
      "loss": 0.5977,
      "step": 466
    },
    {
      "epoch": 0.5992221019287061,
      "grad_norm": 0.12940941751003265,
      "learning_rate": 3.8324999999999996e-05,
      "loss": 0.7215,
      "step": 467
    },
    {
      "epoch": 0.6005052327679538,
      "grad_norm": 0.14956092834472656,
      "learning_rate": 3.83e-05,
      "loss": 0.6522,
      "step": 468
    },
    {
      "epoch": 0.6017883636072016,
      "grad_norm": 0.1391070932149887,
      "learning_rate": 3.8275e-05,
      "loss": 0.7422,
      "step": 469
    },
    {
      "epoch": 0.6030714944464494,
      "grad_norm": 0.10113101452589035,
      "learning_rate": 3.825e-05,
      "loss": 0.7616,
      "step": 470
    },
    {
      "epoch": 0.6043546252856971,
      "grad_norm": 0.11136680841445923,
      "learning_rate": 3.8225e-05,
      "loss": 0.687,
      "step": 471
    },
    {
      "epoch": 0.6056377561249449,
      "grad_norm": 0.17701995372772217,
      "learning_rate": 3.82e-05,
      "loss": 0.826,
      "step": 472
    },
    {
      "epoch": 0.6069208869641927,
      "grad_norm": 0.9745544791221619,
      "learning_rate": 3.8175e-05,
      "loss": 0.6419,
      "step": 473
    },
    {
      "epoch": 0.6082040178034404,
      "grad_norm": 0.13341066241264343,
      "learning_rate": 3.8150000000000006e-05,
      "loss": 0.6596,
      "step": 474
    },
    {
      "epoch": 0.6094871486426882,
      "grad_norm": 0.12377429753541946,
      "learning_rate": 3.8125e-05,
      "loss": 0.7391,
      "step": 475
    },
    {
      "epoch": 0.610770279481936,
      "grad_norm": 0.13578259944915771,
      "learning_rate": 3.8100000000000005e-05,
      "loss": 0.6208,
      "step": 476
    },
    {
      "epoch": 0.6120534103211837,
      "grad_norm": 0.10899040102958679,
      "learning_rate": 3.8075e-05,
      "loss": 0.5854,
      "step": 477
    },
    {
      "epoch": 0.6133365411604315,
      "grad_norm": 0.1064559668302536,
      "learning_rate": 3.805e-05,
      "loss": 0.7712,
      "step": 478
    },
    {
      "epoch": 0.6146196719996793,
      "grad_norm": 5.155559539794922,
      "learning_rate": 3.8025e-05,
      "loss": 0.7801,
      "step": 479
    },
    {
      "epoch": 0.615902802838927,
      "grad_norm": 0.1567067801952362,
      "learning_rate": 3.8e-05,
      "loss": 0.7311,
      "step": 480
    },
    {
      "epoch": 0.6171859336781748,
      "grad_norm": 0.12479838728904724,
      "learning_rate": 3.7975e-05,
      "loss": 0.7153,
      "step": 481
    },
    {
      "epoch": 0.6184690645174226,
      "grad_norm": 0.09556035697460175,
      "learning_rate": 3.795e-05,
      "loss": 0.5879,
      "step": 482
    },
    {
      "epoch": 0.6197521953566703,
      "grad_norm": 0.13914047181606293,
      "learning_rate": 3.7925e-05,
      "loss": 0.7255,
      "step": 483
    },
    {
      "epoch": 0.6210353261959181,
      "grad_norm": 0.14421620965003967,
      "learning_rate": 3.79e-05,
      "loss": 0.6898,
      "step": 484
    },
    {
      "epoch": 0.6223184570351658,
      "grad_norm": 0.13707441091537476,
      "learning_rate": 3.7875e-05,
      "loss": 0.7546,
      "step": 485
    },
    {
      "epoch": 0.6236015878744136,
      "grad_norm": 0.11180365085601807,
      "learning_rate": 3.7850000000000005e-05,
      "loss": 0.7406,
      "step": 486
    },
    {
      "epoch": 0.6248847187136614,
      "grad_norm": 0.13789613544940948,
      "learning_rate": 3.7825e-05,
      "loss": 0.75,
      "step": 487
    },
    {
      "epoch": 0.626167849552909,
      "grad_norm": 0.26896464824676514,
      "learning_rate": 3.7800000000000004e-05,
      "loss": 0.6984,
      "step": 488
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 0.1413726955652237,
      "learning_rate": 3.7775e-05,
      "loss": 0.6909,
      "step": 489
    },
    {
      "epoch": 0.6287341112314047,
      "grad_norm": 0.11270018666982651,
      "learning_rate": 3.775e-05,
      "loss": 0.6581,
      "step": 490
    },
    {
      "epoch": 0.6300172420706524,
      "grad_norm": 0.11073976010084152,
      "learning_rate": 3.7725e-05,
      "loss": 0.6281,
      "step": 491
    },
    {
      "epoch": 0.6313003729099002,
      "grad_norm": 0.11867275089025497,
      "learning_rate": 3.77e-05,
      "loss": 0.6931,
      "step": 492
    },
    {
      "epoch": 0.632583503749148,
      "grad_norm": 0.3392511308193207,
      "learning_rate": 3.7675e-05,
      "loss": 0.7221,
      "step": 493
    },
    {
      "epoch": 0.6338666345883956,
      "grad_norm": 0.11137609928846359,
      "learning_rate": 3.765e-05,
      "loss": 0.6256,
      "step": 494
    },
    {
      "epoch": 0.6351497654276435,
      "grad_norm": 0.116745725274086,
      "learning_rate": 3.7625e-05,
      "loss": 0.6181,
      "step": 495
    },
    {
      "epoch": 0.6364328962668913,
      "grad_norm": 0.09434323757886887,
      "learning_rate": 3.76e-05,
      "loss": 0.741,
      "step": 496
    },
    {
      "epoch": 0.637716027106139,
      "grad_norm": 0.1807403713464737,
      "learning_rate": 3.7575e-05,
      "loss": 0.7614,
      "step": 497
    },
    {
      "epoch": 0.6389991579453868,
      "grad_norm": 0.12564918398857117,
      "learning_rate": 3.7550000000000005e-05,
      "loss": 0.6461,
      "step": 498
    },
    {
      "epoch": 0.6402822887846346,
      "grad_norm": 0.09859195351600647,
      "learning_rate": 3.7525e-05,
      "loss": 0.6982,
      "step": 499
    },
    {
      "epoch": 0.6415654196238822,
      "grad_norm": 0.09869594126939774,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.7154,
      "step": 500
    },
    {
      "epoch": 0.64284855046313,
      "grad_norm": 0.11536712199449539,
      "learning_rate": 3.7475e-05,
      "loss": 0.7432,
      "step": 501
    },
    {
      "epoch": 0.6441316813023779,
      "grad_norm": 0.15481530129909515,
      "learning_rate": 3.745e-05,
      "loss": 0.6877,
      "step": 502
    },
    {
      "epoch": 0.6454148121416255,
      "grad_norm": 0.09947146475315094,
      "learning_rate": 3.7425e-05,
      "loss": 0.7033,
      "step": 503
    },
    {
      "epoch": 0.6466979429808734,
      "grad_norm": 0.1348600536584854,
      "learning_rate": 3.74e-05,
      "loss": 0.7586,
      "step": 504
    },
    {
      "epoch": 0.647981073820121,
      "grad_norm": 0.09145575016736984,
      "learning_rate": 3.737500000000001e-05,
      "loss": 0.6825,
      "step": 505
    },
    {
      "epoch": 0.6492642046593688,
      "grad_norm": 0.09988348931074142,
      "learning_rate": 3.735e-05,
      "loss": 0.6468,
      "step": 506
    },
    {
      "epoch": 0.6505473354986167,
      "grad_norm": 0.10711393505334854,
      "learning_rate": 3.7325000000000006e-05,
      "loss": 0.8177,
      "step": 507
    },
    {
      "epoch": 0.6518304663378643,
      "grad_norm": 0.29017120599746704,
      "learning_rate": 3.73e-05,
      "loss": 0.7262,
      "step": 508
    },
    {
      "epoch": 0.6531135971771121,
      "grad_norm": 0.09933415055274963,
      "learning_rate": 3.7275000000000005e-05,
      "loss": 0.7006,
      "step": 509
    },
    {
      "epoch": 0.65439672801636,
      "grad_norm": 0.09891393035650253,
      "learning_rate": 3.7250000000000004e-05,
      "loss": 0.5933,
      "step": 510
    },
    {
      "epoch": 0.6556798588556076,
      "grad_norm": 0.12664063274860382,
      "learning_rate": 3.7225000000000004e-05,
      "loss": 0.7408,
      "step": 511
    },
    {
      "epoch": 0.6569629896948554,
      "grad_norm": 0.08862703293561935,
      "learning_rate": 3.72e-05,
      "loss": 0.6918,
      "step": 512
    },
    {
      "epoch": 0.6582461205341033,
      "grad_norm": 0.1242414340376854,
      "learning_rate": 3.7175e-05,
      "loss": 0.7847,
      "step": 513
    },
    {
      "epoch": 0.6595292513733509,
      "grad_norm": 0.09848427027463913,
      "learning_rate": 3.715e-05,
      "loss": 0.6721,
      "step": 514
    },
    {
      "epoch": 0.6608123822125987,
      "grad_norm": 0.12347912788391113,
      "learning_rate": 3.7125e-05,
      "loss": 0.7302,
      "step": 515
    },
    {
      "epoch": 0.6620955130518466,
      "grad_norm": 0.19923174381256104,
      "learning_rate": 3.71e-05,
      "loss": 0.6485,
      "step": 516
    },
    {
      "epoch": 0.6633786438910942,
      "grad_norm": 0.1401410847902298,
      "learning_rate": 3.707500000000001e-05,
      "loss": 0.732,
      "step": 517
    },
    {
      "epoch": 0.664661774730342,
      "grad_norm": 0.12570728361606598,
      "learning_rate": 3.705e-05,
      "loss": 0.7602,
      "step": 518
    },
    {
      "epoch": 0.6659449055695899,
      "grad_norm": 0.10133013874292374,
      "learning_rate": 3.7025000000000005e-05,
      "loss": 0.5781,
      "step": 519
    },
    {
      "epoch": 0.6672280364088375,
      "grad_norm": 0.15065886080265045,
      "learning_rate": 3.7e-05,
      "loss": 0.7046,
      "step": 520
    },
    {
      "epoch": 0.6685111672480853,
      "grad_norm": 0.09465158730745316,
      "learning_rate": 3.6975000000000004e-05,
      "loss": 0.7745,
      "step": 521
    },
    {
      "epoch": 0.669794298087333,
      "grad_norm": 0.17046134173870087,
      "learning_rate": 3.6950000000000004e-05,
      "loss": 0.6858,
      "step": 522
    },
    {
      "epoch": 0.6710774289265808,
      "grad_norm": 0.15582051873207092,
      "learning_rate": 3.6925e-05,
      "loss": 0.6547,
      "step": 523
    },
    {
      "epoch": 0.6723605597658286,
      "grad_norm": 0.09487300366163254,
      "learning_rate": 3.69e-05,
      "loss": 0.6562,
      "step": 524
    },
    {
      "epoch": 0.6736436906050763,
      "grad_norm": 7.371533393859863,
      "learning_rate": 3.6875e-05,
      "loss": 0.7456,
      "step": 525
    },
    {
      "epoch": 0.6749268214443241,
      "grad_norm": 0.18365183472633362,
      "learning_rate": 3.685e-05,
      "loss": 0.7767,
      "step": 526
    },
    {
      "epoch": 0.676209952283572,
      "grad_norm": 0.14761795103549957,
      "learning_rate": 3.6825e-05,
      "loss": 0.78,
      "step": 527
    },
    {
      "epoch": 0.6774930831228196,
      "grad_norm": 0.10602157562971115,
      "learning_rate": 3.68e-05,
      "loss": 0.6765,
      "step": 528
    },
    {
      "epoch": 0.6787762139620674,
      "grad_norm": 0.18120723962783813,
      "learning_rate": 3.6775000000000006e-05,
      "loss": 0.7096,
      "step": 529
    },
    {
      "epoch": 0.6800593448013152,
      "grad_norm": 0.12192177027463913,
      "learning_rate": 3.675e-05,
      "loss": 0.698,
      "step": 530
    },
    {
      "epoch": 0.6813424756405629,
      "grad_norm": 0.10490310192108154,
      "learning_rate": 3.6725000000000005e-05,
      "loss": 0.7023,
      "step": 531
    },
    {
      "epoch": 0.6826256064798107,
      "grad_norm": 0.09801571816205978,
      "learning_rate": 3.6700000000000004e-05,
      "loss": 0.6708,
      "step": 532
    },
    {
      "epoch": 0.6839087373190585,
      "grad_norm": 0.7583180069923401,
      "learning_rate": 3.6675000000000004e-05,
      "loss": 0.7597,
      "step": 533
    },
    {
      "epoch": 0.6851918681583062,
      "grad_norm": 0.10812049359083176,
      "learning_rate": 3.665e-05,
      "loss": 0.6392,
      "step": 534
    },
    {
      "epoch": 0.686474998997554,
      "grad_norm": 0.10230198502540588,
      "learning_rate": 3.6625e-05,
      "loss": 0.6675,
      "step": 535
    },
    {
      "epoch": 0.6877581298368018,
      "grad_norm": 0.08567744493484497,
      "learning_rate": 3.66e-05,
      "loss": 0.6303,
      "step": 536
    },
    {
      "epoch": 0.6890412606760495,
      "grad_norm": 0.09311284869909286,
      "learning_rate": 3.6575e-05,
      "loss": 0.6512,
      "step": 537
    },
    {
      "epoch": 0.6903243915152973,
      "grad_norm": 0.10817094147205353,
      "learning_rate": 3.655e-05,
      "loss": 0.7085,
      "step": 538
    },
    {
      "epoch": 0.691607522354545,
      "grad_norm": 0.3169589638710022,
      "learning_rate": 3.652500000000001e-05,
      "loss": 0.6198,
      "step": 539
    },
    {
      "epoch": 0.6928906531937928,
      "grad_norm": 0.09287841618061066,
      "learning_rate": 3.65e-05,
      "loss": 0.6398,
      "step": 540
    },
    {
      "epoch": 0.6941737840330406,
      "grad_norm": 0.10117990523576736,
      "learning_rate": 3.6475000000000006e-05,
      "loss": 0.715,
      "step": 541
    },
    {
      "epoch": 0.6954569148722883,
      "grad_norm": 0.14389462769031525,
      "learning_rate": 3.645e-05,
      "loss": 0.7736,
      "step": 542
    },
    {
      "epoch": 0.6967400457115361,
      "grad_norm": 0.20831343531608582,
      "learning_rate": 3.6425000000000004e-05,
      "loss": 0.8113,
      "step": 543
    },
    {
      "epoch": 0.6980231765507839,
      "grad_norm": 0.12163504213094711,
      "learning_rate": 3.6400000000000004e-05,
      "loss": 0.7315,
      "step": 544
    },
    {
      "epoch": 0.6993063073900316,
      "grad_norm": 0.1033041849732399,
      "learning_rate": 3.6375e-05,
      "loss": 0.4909,
      "step": 545
    },
    {
      "epoch": 0.7005894382292794,
      "grad_norm": 0.1105664074420929,
      "learning_rate": 3.635e-05,
      "loss": 0.7465,
      "step": 546
    },
    {
      "epoch": 0.7018725690685272,
      "grad_norm": 0.10103859752416611,
      "learning_rate": 3.6325e-05,
      "loss": 0.7244,
      "step": 547
    },
    {
      "epoch": 0.7031556999077749,
      "grad_norm": 0.09831926971673965,
      "learning_rate": 3.63e-05,
      "loss": 0.692,
      "step": 548
    },
    {
      "epoch": 0.7044388307470227,
      "grad_norm": 0.43204283714294434,
      "learning_rate": 3.6275e-05,
      "loss": 0.7081,
      "step": 549
    },
    {
      "epoch": 0.7057219615862705,
      "grad_norm": 0.10138247162103653,
      "learning_rate": 3.625e-05,
      "loss": 0.7177,
      "step": 550
    },
    {
      "epoch": 0.7070050924255182,
      "grad_norm": 0.11076109111309052,
      "learning_rate": 3.6225000000000006e-05,
      "loss": 0.7668,
      "step": 551
    },
    {
      "epoch": 0.708288223264766,
      "grad_norm": 0.08695856481790543,
      "learning_rate": 3.62e-05,
      "loss": 0.5628,
      "step": 552
    },
    {
      "epoch": 0.7095713541040138,
      "grad_norm": 0.10035163909196854,
      "learning_rate": 3.6175000000000005e-05,
      "loss": 0.6186,
      "step": 553
    },
    {
      "epoch": 0.7108544849432615,
      "grad_norm": 0.2083156704902649,
      "learning_rate": 3.615e-05,
      "loss": 0.6441,
      "step": 554
    },
    {
      "epoch": 0.7121376157825093,
      "grad_norm": 0.10638521611690521,
      "learning_rate": 3.6125000000000004e-05,
      "loss": 0.7388,
      "step": 555
    },
    {
      "epoch": 0.7134207466217571,
      "grad_norm": 0.1127423495054245,
      "learning_rate": 3.61e-05,
      "loss": 0.7282,
      "step": 556
    },
    {
      "epoch": 0.7147038774610048,
      "grad_norm": 0.1786346137523651,
      "learning_rate": 3.6075e-05,
      "loss": 0.6751,
      "step": 557
    },
    {
      "epoch": 0.7159870083002526,
      "grad_norm": 0.11509651690721512,
      "learning_rate": 3.605e-05,
      "loss": 0.692,
      "step": 558
    },
    {
      "epoch": 0.7172701391395003,
      "grad_norm": 0.12053357809782028,
      "learning_rate": 3.6025e-05,
      "loss": 0.675,
      "step": 559
    },
    {
      "epoch": 0.7185532699787481,
      "grad_norm": 0.12872034311294556,
      "learning_rate": 3.6e-05,
      "loss": 0.5682,
      "step": 560
    },
    {
      "epoch": 0.7198364008179959,
      "grad_norm": 0.1278202384710312,
      "learning_rate": 3.5975e-05,
      "loss": 0.6938,
      "step": 561
    },
    {
      "epoch": 0.7211195316572436,
      "grad_norm": 0.12903237342834473,
      "learning_rate": 3.595e-05,
      "loss": 0.7333,
      "step": 562
    },
    {
      "epoch": 0.7224026624964914,
      "grad_norm": 0.21061474084854126,
      "learning_rate": 3.5925000000000006e-05,
      "loss": 0.7054,
      "step": 563
    },
    {
      "epoch": 0.7236857933357392,
      "grad_norm": 0.10763483494520187,
      "learning_rate": 3.59e-05,
      "loss": 0.712,
      "step": 564
    },
    {
      "epoch": 0.7249689241749869,
      "grad_norm": 0.12677288055419922,
      "learning_rate": 3.5875000000000005e-05,
      "loss": 0.6685,
      "step": 565
    },
    {
      "epoch": 0.7262520550142347,
      "grad_norm": 0.2623748481273651,
      "learning_rate": 3.585e-05,
      "loss": 0.7501,
      "step": 566
    },
    {
      "epoch": 0.7275351858534825,
      "grad_norm": 0.09604787826538086,
      "learning_rate": 3.5825000000000003e-05,
      "loss": 0.6238,
      "step": 567
    },
    {
      "epoch": 0.7288183166927302,
      "grad_norm": 0.14227741956710815,
      "learning_rate": 3.58e-05,
      "loss": 0.6405,
      "step": 568
    },
    {
      "epoch": 0.730101447531978,
      "grad_norm": 0.1902012974023819,
      "learning_rate": 3.5775e-05,
      "loss": 0.6313,
      "step": 569
    },
    {
      "epoch": 0.7313845783712258,
      "grad_norm": 0.11100679636001587,
      "learning_rate": 3.575e-05,
      "loss": 0.6618,
      "step": 570
    },
    {
      "epoch": 0.7326677092104735,
      "grad_norm": 0.13485203683376312,
      "learning_rate": 3.5725e-05,
      "loss": 0.6441,
      "step": 571
    },
    {
      "epoch": 0.7339508400497213,
      "grad_norm": 0.12201449275016785,
      "learning_rate": 3.57e-05,
      "loss": 0.6563,
      "step": 572
    },
    {
      "epoch": 0.7352339708889691,
      "grad_norm": 0.1061677560210228,
      "learning_rate": 3.5675e-05,
      "loss": 0.5762,
      "step": 573
    },
    {
      "epoch": 0.7365171017282168,
      "grad_norm": 0.26799824833869934,
      "learning_rate": 3.565e-05,
      "loss": 0.7065,
      "step": 574
    },
    {
      "epoch": 0.7378002325674646,
      "grad_norm": 0.09463923424482346,
      "learning_rate": 3.5625000000000005e-05,
      "loss": 0.6106,
      "step": 575
    },
    {
      "epoch": 0.7390833634067123,
      "grad_norm": 0.1303613930940628,
      "learning_rate": 3.56e-05,
      "loss": 0.7319,
      "step": 576
    },
    {
      "epoch": 0.7403664942459601,
      "grad_norm": 0.12025018781423569,
      "learning_rate": 3.5575000000000004e-05,
      "loss": 0.7125,
      "step": 577
    },
    {
      "epoch": 0.7416496250852079,
      "grad_norm": 0.11420907080173492,
      "learning_rate": 3.555e-05,
      "loss": 0.7094,
      "step": 578
    },
    {
      "epoch": 0.7429327559244556,
      "grad_norm": 0.1070111095905304,
      "learning_rate": 3.5525e-05,
      "loss": 0.6751,
      "step": 579
    },
    {
      "epoch": 0.7442158867637034,
      "grad_norm": 0.12131603062152863,
      "learning_rate": 3.55e-05,
      "loss": 0.7196,
      "step": 580
    },
    {
      "epoch": 0.7454990176029512,
      "grad_norm": 0.11366055905818939,
      "learning_rate": 3.5475e-05,
      "loss": 0.7131,
      "step": 581
    },
    {
      "epoch": 0.7467821484421989,
      "grad_norm": 0.10522059351205826,
      "learning_rate": 3.545e-05,
      "loss": 0.6632,
      "step": 582
    },
    {
      "epoch": 0.7480652792814467,
      "grad_norm": 0.10321354866027832,
      "learning_rate": 3.5425e-05,
      "loss": 0.6377,
      "step": 583
    },
    {
      "epoch": 0.7493484101206945,
      "grad_norm": 0.11817213892936707,
      "learning_rate": 3.54e-05,
      "loss": 0.7257,
      "step": 584
    },
    {
      "epoch": 0.7506315409599422,
      "grad_norm": 0.12378465384244919,
      "learning_rate": 3.5375e-05,
      "loss": 0.6693,
      "step": 585
    },
    {
      "epoch": 0.75191467179919,
      "grad_norm": 0.4140489995479584,
      "learning_rate": 3.535e-05,
      "loss": 0.7795,
      "step": 586
    },
    {
      "epoch": 0.7531978026384378,
      "grad_norm": 0.10037937760353088,
      "learning_rate": 3.5325000000000005e-05,
      "loss": 0.6737,
      "step": 587
    },
    {
      "epoch": 0.7544809334776855,
      "grad_norm": 0.10465231537818909,
      "learning_rate": 3.53e-05,
      "loss": 0.5987,
      "step": 588
    },
    {
      "epoch": 0.7557640643169333,
      "grad_norm": 0.11430302262306213,
      "learning_rate": 3.5275000000000004e-05,
      "loss": 0.7008,
      "step": 589
    },
    {
      "epoch": 0.7570471951561811,
      "grad_norm": 0.13027602434158325,
      "learning_rate": 3.525e-05,
      "loss": 0.6337,
      "step": 590
    },
    {
      "epoch": 0.7583303259954288,
      "grad_norm": 0.11717857420444489,
      "learning_rate": 3.5225e-05,
      "loss": 0.6423,
      "step": 591
    },
    {
      "epoch": 0.7596134568346766,
      "grad_norm": 0.1307988315820694,
      "learning_rate": 3.52e-05,
      "loss": 0.5831,
      "step": 592
    },
    {
      "epoch": 0.7608965876739244,
      "grad_norm": 0.11526813358068466,
      "learning_rate": 3.5175e-05,
      "loss": 0.7257,
      "step": 593
    },
    {
      "epoch": 0.7621797185131721,
      "grad_norm": 0.10773120820522308,
      "learning_rate": 3.515e-05,
      "loss": 0.6475,
      "step": 594
    },
    {
      "epoch": 0.7634628493524199,
      "grad_norm": 0.1046852096915245,
      "learning_rate": 3.5125e-05,
      "loss": 0.6936,
      "step": 595
    },
    {
      "epoch": 0.7647459801916676,
      "grad_norm": 0.1332172155380249,
      "learning_rate": 3.51e-05,
      "loss": 0.6364,
      "step": 596
    },
    {
      "epoch": 0.7660291110309154,
      "grad_norm": 0.13384002447128296,
      "learning_rate": 3.5075000000000006e-05,
      "loss": 0.8112,
      "step": 597
    },
    {
      "epoch": 0.7673122418701632,
      "grad_norm": 0.1073940098285675,
      "learning_rate": 3.505e-05,
      "loss": 0.7684,
      "step": 598
    },
    {
      "epoch": 0.7685953727094109,
      "grad_norm": 0.10215780884027481,
      "learning_rate": 3.5025000000000004e-05,
      "loss": 0.6891,
      "step": 599
    },
    {
      "epoch": 0.7698785035486587,
      "grad_norm": 0.10806446522474289,
      "learning_rate": 3.5e-05,
      "loss": 0.7169,
      "step": 600
    },
    {
      "epoch": 0.7711616343879065,
      "grad_norm": 0.2700684070587158,
      "learning_rate": 3.4975e-05,
      "loss": 0.8259,
      "step": 601
    },
    {
      "epoch": 0.7724447652271542,
      "grad_norm": 0.1045248731970787,
      "learning_rate": 3.495e-05,
      "loss": 0.7346,
      "step": 602
    },
    {
      "epoch": 0.773727896066402,
      "grad_norm": 0.12629562616348267,
      "learning_rate": 3.4925e-05,
      "loss": 0.7416,
      "step": 603
    },
    {
      "epoch": 0.7750110269056498,
      "grad_norm": 0.09575358778238297,
      "learning_rate": 3.49e-05,
      "loss": 0.6455,
      "step": 604
    },
    {
      "epoch": 0.7762941577448975,
      "grad_norm": 0.11784084141254425,
      "learning_rate": 3.4875e-05,
      "loss": 0.7002,
      "step": 605
    },
    {
      "epoch": 0.7775772885841453,
      "grad_norm": 0.264914870262146,
      "learning_rate": 3.485e-05,
      "loss": 0.7674,
      "step": 606
    },
    {
      "epoch": 0.7788604194233931,
      "grad_norm": 0.10858652740716934,
      "learning_rate": 3.4825e-05,
      "loss": 0.8125,
      "step": 607
    },
    {
      "epoch": 0.7801435502626408,
      "grad_norm": 0.11780703067779541,
      "learning_rate": 3.48e-05,
      "loss": 0.6987,
      "step": 608
    },
    {
      "epoch": 0.7814266811018886,
      "grad_norm": 0.09321476519107819,
      "learning_rate": 3.4775000000000005e-05,
      "loss": 0.5926,
      "step": 609
    },
    {
      "epoch": 0.7827098119411364,
      "grad_norm": 0.12450412660837173,
      "learning_rate": 3.475e-05,
      "loss": 0.6771,
      "step": 610
    },
    {
      "epoch": 0.7839929427803841,
      "grad_norm": 0.09587979316711426,
      "learning_rate": 3.4725000000000004e-05,
      "loss": 0.6467,
      "step": 611
    },
    {
      "epoch": 0.7852760736196319,
      "grad_norm": 0.1411789357662201,
      "learning_rate": 3.4699999999999996e-05,
      "loss": 0.72,
      "step": 612
    },
    {
      "epoch": 0.7865592044588796,
      "grad_norm": 0.11389713734388351,
      "learning_rate": 3.4675e-05,
      "loss": 0.6113,
      "step": 613
    },
    {
      "epoch": 0.7878423352981274,
      "grad_norm": 0.17298798263072968,
      "learning_rate": 3.465e-05,
      "loss": 0.5873,
      "step": 614
    },
    {
      "epoch": 0.7891254661373752,
      "grad_norm": 0.14938412606716156,
      "learning_rate": 3.4625e-05,
      "loss": 0.6788,
      "step": 615
    },
    {
      "epoch": 0.7904085969766229,
      "grad_norm": 0.10452687740325928,
      "learning_rate": 3.46e-05,
      "loss": 0.7003,
      "step": 616
    },
    {
      "epoch": 0.7916917278158707,
      "grad_norm": 0.16350068151950836,
      "learning_rate": 3.4575e-05,
      "loss": 0.7899,
      "step": 617
    },
    {
      "epoch": 0.7929748586551185,
      "grad_norm": 0.20277175307273865,
      "learning_rate": 3.455e-05,
      "loss": 0.7788,
      "step": 618
    },
    {
      "epoch": 0.7942579894943662,
      "grad_norm": 0.13396036624908447,
      "learning_rate": 3.4525e-05,
      "loss": 0.6683,
      "step": 619
    },
    {
      "epoch": 0.795541120333614,
      "grad_norm": 0.0961032286286354,
      "learning_rate": 3.45e-05,
      "loss": 0.7018,
      "step": 620
    },
    {
      "epoch": 0.7968242511728618,
      "grad_norm": 0.1324867457151413,
      "learning_rate": 3.4475000000000005e-05,
      "loss": 0.695,
      "step": 621
    },
    {
      "epoch": 0.7981073820121095,
      "grad_norm": 0.2975425720214844,
      "learning_rate": 3.445e-05,
      "loss": 0.6641,
      "step": 622
    },
    {
      "epoch": 0.7993905128513573,
      "grad_norm": 0.11577823013067245,
      "learning_rate": 3.4425e-05,
      "loss": 0.64,
      "step": 623
    },
    {
      "epoch": 0.8006736436906051,
      "grad_norm": 0.12378153204917908,
      "learning_rate": 3.4399999999999996e-05,
      "loss": 0.7557,
      "step": 624
    },
    {
      "epoch": 0.8019567745298528,
      "grad_norm": 0.11317790299654007,
      "learning_rate": 3.4375e-05,
      "loss": 0.6021,
      "step": 625
    },
    {
      "epoch": 0.8032399053691006,
      "grad_norm": 0.12973865866661072,
      "learning_rate": 3.435e-05,
      "loss": 0.6593,
      "step": 626
    },
    {
      "epoch": 0.8045230362083484,
      "grad_norm": 0.15441685914993286,
      "learning_rate": 3.4325e-05,
      "loss": 0.707,
      "step": 627
    },
    {
      "epoch": 0.8058061670475961,
      "grad_norm": 0.10903256386518478,
      "learning_rate": 3.430000000000001e-05,
      "loss": 0.578,
      "step": 628
    },
    {
      "epoch": 0.8070892978868439,
      "grad_norm": 0.10009898990392685,
      "learning_rate": 3.4275e-05,
      "loss": 0.7718,
      "step": 629
    },
    {
      "epoch": 0.8083724287260917,
      "grad_norm": 0.11873394250869751,
      "learning_rate": 3.4250000000000006e-05,
      "loss": 0.724,
      "step": 630
    },
    {
      "epoch": 0.8096555595653394,
      "grad_norm": 0.1902954876422882,
      "learning_rate": 3.4225e-05,
      "loss": 0.6973,
      "step": 631
    },
    {
      "epoch": 0.8109386904045872,
      "grad_norm": 0.1514977216720581,
      "learning_rate": 3.4200000000000005e-05,
      "loss": 0.6036,
      "step": 632
    },
    {
      "epoch": 0.8122218212438349,
      "grad_norm": 0.10894272476434708,
      "learning_rate": 3.4175000000000004e-05,
      "loss": 0.7341,
      "step": 633
    },
    {
      "epoch": 0.8135049520830827,
      "grad_norm": 0.10203227400779724,
      "learning_rate": 3.415e-05,
      "loss": 0.6946,
      "step": 634
    },
    {
      "epoch": 0.8147880829223305,
      "grad_norm": 0.13857731223106384,
      "learning_rate": 3.4125e-05,
      "loss": 0.6224,
      "step": 635
    },
    {
      "epoch": 0.8160712137615782,
      "grad_norm": 0.15909959375858307,
      "learning_rate": 3.41e-05,
      "loss": 0.6847,
      "step": 636
    },
    {
      "epoch": 0.817354344600826,
      "grad_norm": 0.10778973251581192,
      "learning_rate": 3.4075e-05,
      "loss": 0.6689,
      "step": 637
    },
    {
      "epoch": 0.8186374754400738,
      "grad_norm": 0.12273471057415009,
      "learning_rate": 3.405e-05,
      "loss": 0.6911,
      "step": 638
    },
    {
      "epoch": 0.8199206062793215,
      "grad_norm": 0.1394212245941162,
      "learning_rate": 3.4025e-05,
      "loss": 0.7075,
      "step": 639
    },
    {
      "epoch": 0.8212037371185693,
      "grad_norm": 0.12034229934215546,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.7439,
      "step": 640
    },
    {
      "epoch": 0.8224868679578171,
      "grad_norm": 0.11535851657390594,
      "learning_rate": 3.3975e-05,
      "loss": 0.6799,
      "step": 641
    },
    {
      "epoch": 0.8237699987970648,
      "grad_norm": 0.11069920659065247,
      "learning_rate": 3.3950000000000005e-05,
      "loss": 0.8017,
      "step": 642
    },
    {
      "epoch": 0.8250531296363126,
      "grad_norm": 0.11882181465625763,
      "learning_rate": 3.3925e-05,
      "loss": 0.7136,
      "step": 643
    },
    {
      "epoch": 0.8263362604755604,
      "grad_norm": 0.10768967866897583,
      "learning_rate": 3.3900000000000004e-05,
      "loss": 0.7086,
      "step": 644
    },
    {
      "epoch": 0.8276193913148081,
      "grad_norm": 0.13374952971935272,
      "learning_rate": 3.3875000000000003e-05,
      "loss": 0.7783,
      "step": 645
    },
    {
      "epoch": 0.8289025221540559,
      "grad_norm": 0.19134868681430817,
      "learning_rate": 3.385e-05,
      "loss": 0.6951,
      "step": 646
    },
    {
      "epoch": 0.8301856529933037,
      "grad_norm": 0.08837475627660751,
      "learning_rate": 3.3825e-05,
      "loss": 0.7214,
      "step": 647
    },
    {
      "epoch": 0.8314687838325514,
      "grad_norm": 0.1260792315006256,
      "learning_rate": 3.38e-05,
      "loss": 0.6693,
      "step": 648
    },
    {
      "epoch": 0.8327519146717992,
      "grad_norm": 0.1632176637649536,
      "learning_rate": 3.3775e-05,
      "loss": 0.784,
      "step": 649
    },
    {
      "epoch": 0.8340350455110469,
      "grad_norm": 0.0972166433930397,
      "learning_rate": 3.375000000000001e-05,
      "loss": 0.7175,
      "step": 650
    },
    {
      "epoch": 0.8353181763502947,
      "grad_norm": 0.12688030302524567,
      "learning_rate": 3.3725e-05,
      "loss": 0.783,
      "step": 651
    },
    {
      "epoch": 0.8366013071895425,
      "grad_norm": 0.15024475753307343,
      "learning_rate": 3.3700000000000006e-05,
      "loss": 0.7094,
      "step": 652
    },
    {
      "epoch": 0.8378844380287902,
      "grad_norm": 0.14747752249240875,
      "learning_rate": 3.3675e-05,
      "loss": 0.6857,
      "step": 653
    },
    {
      "epoch": 0.839167568868038,
      "grad_norm": 0.1281762421131134,
      "learning_rate": 3.3650000000000005e-05,
      "loss": 0.738,
      "step": 654
    },
    {
      "epoch": 0.8404506997072858,
      "grad_norm": 0.13929852843284607,
      "learning_rate": 3.3625000000000004e-05,
      "loss": 0.6748,
      "step": 655
    },
    {
      "epoch": 0.8417338305465335,
      "grad_norm": 0.15822292864322662,
      "learning_rate": 3.3600000000000004e-05,
      "loss": 0.5666,
      "step": 656
    },
    {
      "epoch": 0.8430169613857813,
      "grad_norm": 0.3289751410484314,
      "learning_rate": 3.3575e-05,
      "loss": 0.653,
      "step": 657
    },
    {
      "epoch": 0.8443000922250291,
      "grad_norm": 0.09926114231348038,
      "learning_rate": 3.355e-05,
      "loss": 0.6958,
      "step": 658
    },
    {
      "epoch": 0.8455832230642768,
      "grad_norm": 0.10496655851602554,
      "learning_rate": 3.3525e-05,
      "loss": 0.7561,
      "step": 659
    },
    {
      "epoch": 0.8468663539035246,
      "grad_norm": 0.11823925375938416,
      "learning_rate": 3.35e-05,
      "loss": 0.6348,
      "step": 660
    },
    {
      "epoch": 0.8481494847427724,
      "grad_norm": 0.1079612448811531,
      "learning_rate": 3.3475e-05,
      "loss": 0.6669,
      "step": 661
    },
    {
      "epoch": 0.8494326155820201,
      "grad_norm": 0.16647039353847504,
      "learning_rate": 3.345000000000001e-05,
      "loss": 0.7158,
      "step": 662
    },
    {
      "epoch": 0.8507157464212679,
      "grad_norm": 0.11895445734262466,
      "learning_rate": 3.3425e-05,
      "loss": 0.7091,
      "step": 663
    },
    {
      "epoch": 0.8519988772605157,
      "grad_norm": 3.7275030612945557,
      "learning_rate": 3.3400000000000005e-05,
      "loss": 0.7115,
      "step": 664
    },
    {
      "epoch": 0.8532820080997634,
      "grad_norm": 0.140670508146286,
      "learning_rate": 3.3375e-05,
      "loss": 0.7133,
      "step": 665
    },
    {
      "epoch": 0.8545651389390112,
      "grad_norm": 0.12398913502693176,
      "learning_rate": 3.3350000000000004e-05,
      "loss": 0.7288,
      "step": 666
    },
    {
      "epoch": 0.8558482697782589,
      "grad_norm": 0.3376699388027191,
      "learning_rate": 3.3325000000000004e-05,
      "loss": 0.6146,
      "step": 667
    },
    {
      "epoch": 0.8571314006175067,
      "grad_norm": 0.1278139054775238,
      "learning_rate": 3.33e-05,
      "loss": 0.6748,
      "step": 668
    },
    {
      "epoch": 0.8584145314567545,
      "grad_norm": 0.2500833570957184,
      "learning_rate": 3.3275e-05,
      "loss": 0.803,
      "step": 669
    },
    {
      "epoch": 0.8596976622960022,
      "grad_norm": 0.12750349938869476,
      "learning_rate": 3.325e-05,
      "loss": 0.7418,
      "step": 670
    },
    {
      "epoch": 0.86098079313525,
      "grad_norm": 0.10624389350414276,
      "learning_rate": 3.3225e-05,
      "loss": 0.5929,
      "step": 671
    },
    {
      "epoch": 0.8622639239744978,
      "grad_norm": 0.15755298733711243,
      "learning_rate": 3.32e-05,
      "loss": 0.6289,
      "step": 672
    },
    {
      "epoch": 0.8635470548137455,
      "grad_norm": 0.150392547249794,
      "learning_rate": 3.3175e-05,
      "loss": 0.7058,
      "step": 673
    },
    {
      "epoch": 0.8648301856529933,
      "grad_norm": 0.11339676380157471,
      "learning_rate": 3.3150000000000006e-05,
      "loss": 0.6321,
      "step": 674
    },
    {
      "epoch": 0.8661133164922411,
      "grad_norm": 0.10511218756437302,
      "learning_rate": 3.3125e-05,
      "loss": 0.6401,
      "step": 675
    },
    {
      "epoch": 0.8673964473314888,
      "grad_norm": 0.12629434466362,
      "learning_rate": 3.3100000000000005e-05,
      "loss": 0.6545,
      "step": 676
    },
    {
      "epoch": 0.8686795781707366,
      "grad_norm": 0.09492748975753784,
      "learning_rate": 3.3075e-05,
      "loss": 0.662,
      "step": 677
    },
    {
      "epoch": 0.8699627090099844,
      "grad_norm": 0.10670129209756851,
      "learning_rate": 3.3050000000000004e-05,
      "loss": 0.6515,
      "step": 678
    },
    {
      "epoch": 0.8712458398492321,
      "grad_norm": 0.11453550308942795,
      "learning_rate": 3.3025e-05,
      "loss": 0.8207,
      "step": 679
    },
    {
      "epoch": 0.8725289706884799,
      "grad_norm": 0.17096608877182007,
      "learning_rate": 3.3e-05,
      "loss": 0.6742,
      "step": 680
    },
    {
      "epoch": 0.8738121015277277,
      "grad_norm": 0.14091692864894867,
      "learning_rate": 3.2975e-05,
      "loss": 0.6032,
      "step": 681
    },
    {
      "epoch": 0.8750952323669754,
      "grad_norm": 0.10900155454874039,
      "learning_rate": 3.295e-05,
      "loss": 0.7213,
      "step": 682
    },
    {
      "epoch": 0.8763783632062232,
      "grad_norm": 0.12387724220752716,
      "learning_rate": 3.2925e-05,
      "loss": 0.6665,
      "step": 683
    },
    {
      "epoch": 0.877661494045471,
      "grad_norm": 0.15852327644824982,
      "learning_rate": 3.29e-05,
      "loss": 0.6255,
      "step": 684
    },
    {
      "epoch": 0.8789446248847187,
      "grad_norm": 0.11493482440710068,
      "learning_rate": 3.2875e-05,
      "loss": 0.6141,
      "step": 685
    },
    {
      "epoch": 0.8802277557239665,
      "grad_norm": 0.09226331859827042,
      "learning_rate": 3.2850000000000006e-05,
      "loss": 0.7189,
      "step": 686
    },
    {
      "epoch": 0.8815108865632142,
      "grad_norm": 0.14584526419639587,
      "learning_rate": 3.2825e-05,
      "loss": 0.6564,
      "step": 687
    },
    {
      "epoch": 0.882794017402462,
      "grad_norm": 0.19676053524017334,
      "learning_rate": 3.2800000000000004e-05,
      "loss": 0.6532,
      "step": 688
    },
    {
      "epoch": 0.8840771482417098,
      "grad_norm": 0.0988076776266098,
      "learning_rate": 3.2775e-05,
      "loss": 0.6978,
      "step": 689
    },
    {
      "epoch": 0.8853602790809575,
      "grad_norm": 0.12281625717878342,
      "learning_rate": 3.275e-05,
      "loss": 0.7577,
      "step": 690
    },
    {
      "epoch": 0.8866434099202053,
      "grad_norm": 0.11525499820709229,
      "learning_rate": 3.2725e-05,
      "loss": 0.7294,
      "step": 691
    },
    {
      "epoch": 0.8879265407594531,
      "grad_norm": 0.1129651740193367,
      "learning_rate": 3.27e-05,
      "loss": 0.6003,
      "step": 692
    },
    {
      "epoch": 0.8892096715987008,
      "grad_norm": 0.11046162992715836,
      "learning_rate": 3.2675e-05,
      "loss": 0.7011,
      "step": 693
    },
    {
      "epoch": 0.8904928024379486,
      "grad_norm": 0.12570635974407196,
      "learning_rate": 3.265e-05,
      "loss": 0.6912,
      "step": 694
    },
    {
      "epoch": 0.8917759332771964,
      "grad_norm": 0.1159944236278534,
      "learning_rate": 3.2625e-05,
      "loss": 0.7992,
      "step": 695
    },
    {
      "epoch": 0.8930590641164441,
      "grad_norm": 0.12372931838035583,
      "learning_rate": 3.26e-05,
      "loss": 0.6827,
      "step": 696
    },
    {
      "epoch": 0.8943421949556919,
      "grad_norm": 0.34311729669570923,
      "learning_rate": 3.2575e-05,
      "loss": 0.7244,
      "step": 697
    },
    {
      "epoch": 0.8956253257949397,
      "grad_norm": 0.10930059105157852,
      "learning_rate": 3.2550000000000005e-05,
      "loss": 0.7029,
      "step": 698
    },
    {
      "epoch": 0.8969084566341874,
      "grad_norm": 0.17305094003677368,
      "learning_rate": 3.2525e-05,
      "loss": 0.7334,
      "step": 699
    },
    {
      "epoch": 0.8981915874734352,
      "grad_norm": 0.09384559839963913,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.7054,
      "step": 700
    },
    {
      "epoch": 0.899474718312683,
      "grad_norm": 0.11595708131790161,
      "learning_rate": 3.2474999999999997e-05,
      "loss": 0.8539,
      "step": 701
    },
    {
      "epoch": 0.9007578491519307,
      "grad_norm": 0.1440788060426712,
      "learning_rate": 3.245e-05,
      "loss": 0.7168,
      "step": 702
    },
    {
      "epoch": 0.9020409799911785,
      "grad_norm": 0.13530150055885315,
      "learning_rate": 3.2425e-05,
      "loss": 0.7139,
      "step": 703
    },
    {
      "epoch": 0.9033241108304262,
      "grad_norm": 0.09742891788482666,
      "learning_rate": 3.24e-05,
      "loss": 0.668,
      "step": 704
    },
    {
      "epoch": 0.904607241669674,
      "grad_norm": 0.11917275190353394,
      "learning_rate": 3.2375e-05,
      "loss": 0.646,
      "step": 705
    },
    {
      "epoch": 0.9058903725089218,
      "grad_norm": 0.11326149106025696,
      "learning_rate": 3.235e-05,
      "loss": 0.8007,
      "step": 706
    },
    {
      "epoch": 0.9071735033481695,
      "grad_norm": 0.10032694786787033,
      "learning_rate": 3.2325e-05,
      "loss": 0.6208,
      "step": 707
    },
    {
      "epoch": 0.9084566341874173,
      "grad_norm": 0.12406820803880692,
      "learning_rate": 3.2300000000000006e-05,
      "loss": 0.6081,
      "step": 708
    },
    {
      "epoch": 0.9097397650266651,
      "grad_norm": 0.20673808455467224,
      "learning_rate": 3.2275e-05,
      "loss": 0.6455,
      "step": 709
    },
    {
      "epoch": 0.9110228958659128,
      "grad_norm": 0.09357975423336029,
      "learning_rate": 3.2250000000000005e-05,
      "loss": 0.5994,
      "step": 710
    },
    {
      "epoch": 0.9123060267051606,
      "grad_norm": 0.216871440410614,
      "learning_rate": 3.2225e-05,
      "loss": 0.6575,
      "step": 711
    },
    {
      "epoch": 0.9135891575444084,
      "grad_norm": 0.12477193027734756,
      "learning_rate": 3.2200000000000003e-05,
      "loss": 0.7463,
      "step": 712
    },
    {
      "epoch": 0.9148722883836561,
      "grad_norm": 0.1255732625722885,
      "learning_rate": 3.2175e-05,
      "loss": 0.7634,
      "step": 713
    },
    {
      "epoch": 0.9161554192229039,
      "grad_norm": 0.11574459075927734,
      "learning_rate": 3.215e-05,
      "loss": 0.7417,
      "step": 714
    },
    {
      "epoch": 0.9174385500621517,
      "grad_norm": 0.11036047339439392,
      "learning_rate": 3.2125e-05,
      "loss": 0.7138,
      "step": 715
    },
    {
      "epoch": 0.9187216809013994,
      "grad_norm": 0.13630536198616028,
      "learning_rate": 3.21e-05,
      "loss": 0.7111,
      "step": 716
    },
    {
      "epoch": 0.9200048117406472,
      "grad_norm": 0.11095181107521057,
      "learning_rate": 3.2075e-05,
      "loss": 0.6402,
      "step": 717
    },
    {
      "epoch": 0.921287942579895,
      "grad_norm": 0.11368012428283691,
      "learning_rate": 3.205e-05,
      "loss": 0.6004,
      "step": 718
    },
    {
      "epoch": 0.9225710734191427,
      "grad_norm": 0.11912807077169418,
      "learning_rate": 3.2025e-05,
      "loss": 0.6057,
      "step": 719
    },
    {
      "epoch": 0.9238542042583905,
      "grad_norm": 0.09563388675451279,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.7242,
      "step": 720
    },
    {
      "epoch": 0.9251373350976383,
      "grad_norm": 0.12385242432355881,
      "learning_rate": 3.1975e-05,
      "loss": 0.7972,
      "step": 721
    },
    {
      "epoch": 0.926420465936886,
      "grad_norm": 0.12675169110298157,
      "learning_rate": 3.1950000000000004e-05,
      "loss": 0.7287,
      "step": 722
    },
    {
      "epoch": 0.9277035967761338,
      "grad_norm": 0.09524949640035629,
      "learning_rate": 3.1925e-05,
      "loss": 0.7318,
      "step": 723
    },
    {
      "epoch": 0.9289867276153815,
      "grad_norm": 0.11698839068412781,
      "learning_rate": 3.19e-05,
      "loss": 0.6346,
      "step": 724
    },
    {
      "epoch": 0.9302698584546293,
      "grad_norm": 0.13197773694992065,
      "learning_rate": 3.1875e-05,
      "loss": 0.6024,
      "step": 725
    },
    {
      "epoch": 0.9315529892938771,
      "grad_norm": 0.11549938470125198,
      "learning_rate": 3.185e-05,
      "loss": 0.6216,
      "step": 726
    },
    {
      "epoch": 0.9328361201331248,
      "grad_norm": 0.09477383643388748,
      "learning_rate": 3.1825e-05,
      "loss": 0.7455,
      "step": 727
    },
    {
      "epoch": 0.9341192509723726,
      "grad_norm": 0.15136121213436127,
      "learning_rate": 3.18e-05,
      "loss": 0.7844,
      "step": 728
    },
    {
      "epoch": 0.9354023818116204,
      "grad_norm": 0.14280231297016144,
      "learning_rate": 3.1775e-05,
      "loss": 0.6697,
      "step": 729
    },
    {
      "epoch": 0.9366855126508681,
      "grad_norm": 0.14406636357307434,
      "learning_rate": 3.175e-05,
      "loss": 0.6741,
      "step": 730
    },
    {
      "epoch": 0.9379686434901159,
      "grad_norm": 0.11636316031217575,
      "learning_rate": 3.1725e-05,
      "loss": 0.7399,
      "step": 731
    },
    {
      "epoch": 0.9392517743293637,
      "grad_norm": 0.13145846128463745,
      "learning_rate": 3.1700000000000005e-05,
      "loss": 0.6877,
      "step": 732
    },
    {
      "epoch": 0.9405349051686114,
      "grad_norm": 0.10198771953582764,
      "learning_rate": 3.1675e-05,
      "loss": 0.604,
      "step": 733
    },
    {
      "epoch": 0.9418180360078592,
      "grad_norm": 0.09470993280410767,
      "learning_rate": 3.1650000000000004e-05,
      "loss": 0.6379,
      "step": 734
    },
    {
      "epoch": 0.943101166847107,
      "grad_norm": 0.08992660045623779,
      "learning_rate": 3.1624999999999996e-05,
      "loss": 0.738,
      "step": 735
    },
    {
      "epoch": 0.9443842976863547,
      "grad_norm": 0.14292825758457184,
      "learning_rate": 3.16e-05,
      "loss": 0.7405,
      "step": 736
    },
    {
      "epoch": 0.9456674285256025,
      "grad_norm": 0.15853001177310944,
      "learning_rate": 3.1575e-05,
      "loss": 0.6144,
      "step": 737
    },
    {
      "epoch": 0.9469505593648503,
      "grad_norm": 0.11740943789482117,
      "learning_rate": 3.155e-05,
      "loss": 0.6516,
      "step": 738
    },
    {
      "epoch": 0.948233690204098,
      "grad_norm": 0.10960552841424942,
      "learning_rate": 3.1525e-05,
      "loss": 0.6779,
      "step": 739
    },
    {
      "epoch": 0.9495168210433458,
      "grad_norm": 0.1766705960035324,
      "learning_rate": 3.15e-05,
      "loss": 0.7002,
      "step": 740
    },
    {
      "epoch": 0.9507999518825935,
      "grad_norm": 0.13118472695350647,
      "learning_rate": 3.1475e-05,
      "loss": 0.7683,
      "step": 741
    },
    {
      "epoch": 0.9520830827218413,
      "grad_norm": 0.10828334093093872,
      "learning_rate": 3.145e-05,
      "loss": 0.7017,
      "step": 742
    },
    {
      "epoch": 0.9533662135610891,
      "grad_norm": 0.1170574277639389,
      "learning_rate": 3.1425e-05,
      "loss": 0.7475,
      "step": 743
    },
    {
      "epoch": 0.9546493444003368,
      "grad_norm": 0.2421141266822815,
      "learning_rate": 3.1400000000000004e-05,
      "loss": 0.7273,
      "step": 744
    },
    {
      "epoch": 0.9559324752395846,
      "grad_norm": 0.11599459499120712,
      "learning_rate": 3.1375e-05,
      "loss": 0.663,
      "step": 745
    },
    {
      "epoch": 0.9572156060788324,
      "grad_norm": 0.13600559532642365,
      "learning_rate": 3.135e-05,
      "loss": 0.799,
      "step": 746
    },
    {
      "epoch": 0.9584987369180801,
      "grad_norm": 0.12247034162282944,
      "learning_rate": 3.1324999999999996e-05,
      "loss": 0.7035,
      "step": 747
    },
    {
      "epoch": 0.9597818677573279,
      "grad_norm": 0.1288401037454605,
      "learning_rate": 3.13e-05,
      "loss": 0.6652,
      "step": 748
    },
    {
      "epoch": 0.9610649985965757,
      "grad_norm": 0.11469082534313202,
      "learning_rate": 3.1275e-05,
      "loss": 0.7364,
      "step": 749
    },
    {
      "epoch": 0.9623481294358234,
      "grad_norm": 0.1188461035490036,
      "learning_rate": 3.125e-05,
      "loss": 0.7132,
      "step": 750
    },
    {
      "epoch": 0.9636312602750712,
      "grad_norm": 0.17554962635040283,
      "learning_rate": 3.122500000000001e-05,
      "loss": 0.6914,
      "step": 751
    },
    {
      "epoch": 0.964914391114319,
      "grad_norm": 0.13335861265659332,
      "learning_rate": 3.12e-05,
      "loss": 0.7287,
      "step": 752
    },
    {
      "epoch": 0.9661975219535667,
      "grad_norm": 0.08735929429531097,
      "learning_rate": 3.1175000000000006e-05,
      "loss": 0.6148,
      "step": 753
    },
    {
      "epoch": 0.9674806527928145,
      "grad_norm": 0.12466836720705032,
      "learning_rate": 3.115e-05,
      "loss": 0.7693,
      "step": 754
    },
    {
      "epoch": 0.9687637836320623,
      "grad_norm": 0.17256440222263336,
      "learning_rate": 3.1125000000000004e-05,
      "loss": 0.7873,
      "step": 755
    },
    {
      "epoch": 0.97004691447131,
      "grad_norm": 0.11574805527925491,
      "learning_rate": 3.1100000000000004e-05,
      "loss": 0.6424,
      "step": 756
    },
    {
      "epoch": 0.9713300453105578,
      "grad_norm": 0.16383589804172516,
      "learning_rate": 3.1075e-05,
      "loss": 0.6657,
      "step": 757
    },
    {
      "epoch": 0.9726131761498056,
      "grad_norm": 0.13190145790576935,
      "learning_rate": 3.105e-05,
      "loss": 0.6204,
      "step": 758
    },
    {
      "epoch": 0.9738963069890533,
      "grad_norm": 0.13317909836769104,
      "learning_rate": 3.1025e-05,
      "loss": 0.7545,
      "step": 759
    },
    {
      "epoch": 0.9751794378283011,
      "grad_norm": 0.1382766216993332,
      "learning_rate": 3.1e-05,
      "loss": 0.6019,
      "step": 760
    },
    {
      "epoch": 0.9764625686675488,
      "grad_norm": 0.1393786072731018,
      "learning_rate": 3.0975e-05,
      "loss": 0.692,
      "step": 761
    },
    {
      "epoch": 0.9777456995067966,
      "grad_norm": 0.12596078217029572,
      "learning_rate": 3.095e-05,
      "loss": 0.6738,
      "step": 762
    },
    {
      "epoch": 0.9790288303460444,
      "grad_norm": 0.10756366699934006,
      "learning_rate": 3.0925000000000006e-05,
      "loss": 0.668,
      "step": 763
    },
    {
      "epoch": 0.9803119611852921,
      "grad_norm": 0.09744535386562347,
      "learning_rate": 3.09e-05,
      "loss": 0.7354,
      "step": 764
    },
    {
      "epoch": 0.9815950920245399,
      "grad_norm": 0.14871183037757874,
      "learning_rate": 3.0875000000000005e-05,
      "loss": 0.6644,
      "step": 765
    },
    {
      "epoch": 0.9828782228637877,
      "grad_norm": 0.11440213024616241,
      "learning_rate": 3.0850000000000004e-05,
      "loss": 0.6229,
      "step": 766
    },
    {
      "epoch": 0.9841613537030354,
      "grad_norm": 0.11519957333803177,
      "learning_rate": 3.0825000000000004e-05,
      "loss": 0.6858,
      "step": 767
    },
    {
      "epoch": 0.9854444845422832,
      "grad_norm": 0.13082237541675568,
      "learning_rate": 3.08e-05,
      "loss": 0.7405,
      "step": 768
    },
    {
      "epoch": 0.986727615381531,
      "grad_norm": 0.12260410189628601,
      "learning_rate": 3.0775e-05,
      "loss": 0.6935,
      "step": 769
    },
    {
      "epoch": 0.9880107462207787,
      "grad_norm": 0.15404881536960602,
      "learning_rate": 3.075e-05,
      "loss": 0.7526,
      "step": 770
    },
    {
      "epoch": 0.9892938770600265,
      "grad_norm": 0.1284703016281128,
      "learning_rate": 3.0725e-05,
      "loss": 0.5874,
      "step": 771
    },
    {
      "epoch": 0.9905770078992743,
      "grad_norm": 0.30294910073280334,
      "learning_rate": 3.07e-05,
      "loss": 0.7329,
      "step": 772
    },
    {
      "epoch": 0.991860138738522,
      "grad_norm": 0.10839234292507172,
      "learning_rate": 3.067500000000001e-05,
      "loss": 0.7128,
      "step": 773
    },
    {
      "epoch": 0.9931432695777698,
      "grad_norm": 0.1409377157688141,
      "learning_rate": 3.065e-05,
      "loss": 0.6618,
      "step": 774
    },
    {
      "epoch": 0.9944264004170176,
      "grad_norm": 0.17123432457447052,
      "learning_rate": 3.0625000000000006e-05,
      "loss": 0.7855,
      "step": 775
    },
    {
      "epoch": 0.9957095312562653,
      "grad_norm": 0.1716230809688568,
      "learning_rate": 3.06e-05,
      "loss": 0.6405,
      "step": 776
    },
    {
      "epoch": 0.9969926620955131,
      "grad_norm": 0.14809831976890564,
      "learning_rate": 3.0575000000000005e-05,
      "loss": 0.6401,
      "step": 777
    },
    {
      "epoch": 0.9982757929347608,
      "grad_norm": 0.1278572529554367,
      "learning_rate": 3.0550000000000004e-05,
      "loss": 0.7392,
      "step": 778
    },
    {
      "epoch": 0.9995589237740086,
      "grad_norm": 0.09439048916101456,
      "learning_rate": 3.0525e-05,
      "loss": 0.622,
      "step": 779
    },
    {
      "epoch": 1.0008420546132564,
      "grad_norm": 0.12026798725128174,
      "learning_rate": 3.05e-05,
      "loss": 0.7348,
      "step": 780
    },
    {
      "epoch": 1.0021251854525042,
      "grad_norm": 0.10444330424070358,
      "learning_rate": 3.0475000000000002e-05,
      "loss": 0.5259,
      "step": 781
    },
    {
      "epoch": 1.003408316291752,
      "grad_norm": 0.08929070830345154,
      "learning_rate": 3.045e-05,
      "loss": 0.6791,
      "step": 782
    },
    {
      "epoch": 1.0046914471309996,
      "grad_norm": 0.1265541911125183,
      "learning_rate": 3.0425000000000004e-05,
      "loss": 0.7023,
      "step": 783
    },
    {
      "epoch": 1.0059745779702474,
      "grad_norm": 0.12845323979854584,
      "learning_rate": 3.04e-05,
      "loss": 0.6593,
      "step": 784
    },
    {
      "epoch": 1.0072577088094952,
      "grad_norm": 0.11886178702116013,
      "learning_rate": 3.0375000000000003e-05,
      "loss": 0.7327,
      "step": 785
    },
    {
      "epoch": 1.008540839648743,
      "grad_norm": 0.12317848205566406,
      "learning_rate": 3.035e-05,
      "loss": 0.7881,
      "step": 786
    },
    {
      "epoch": 1.0098239704879908,
      "grad_norm": 0.1767103374004364,
      "learning_rate": 3.0325000000000002e-05,
      "loss": 0.62,
      "step": 787
    },
    {
      "epoch": 1.0111071013272384,
      "grad_norm": 0.15777979791164398,
      "learning_rate": 3.03e-05,
      "loss": 0.5671,
      "step": 788
    },
    {
      "epoch": 1.0123902321664862,
      "grad_norm": 0.11238143593072891,
      "learning_rate": 3.0275000000000004e-05,
      "loss": 0.6874,
      "step": 789
    },
    {
      "epoch": 1.013673363005734,
      "grad_norm": 0.12935291230678558,
      "learning_rate": 3.025e-05,
      "loss": 0.6349,
      "step": 790
    },
    {
      "epoch": 1.0149564938449818,
      "grad_norm": 0.15927043557167053,
      "learning_rate": 3.0225000000000003e-05,
      "loss": 0.6922,
      "step": 791
    },
    {
      "epoch": 1.0162396246842296,
      "grad_norm": 0.11516556888818741,
      "learning_rate": 3.02e-05,
      "loss": 0.6669,
      "step": 792
    },
    {
      "epoch": 1.0175227555234774,
      "grad_norm": 0.16156385838985443,
      "learning_rate": 3.0175e-05,
      "loss": 0.7768,
      "step": 793
    },
    {
      "epoch": 1.018805886362725,
      "grad_norm": 0.1712319403886795,
      "learning_rate": 3.015e-05,
      "loss": 0.7598,
      "step": 794
    },
    {
      "epoch": 1.0200890172019728,
      "grad_norm": 0.11005910485982895,
      "learning_rate": 3.0125000000000004e-05,
      "loss": 0.6247,
      "step": 795
    },
    {
      "epoch": 1.0213721480412206,
      "grad_norm": 0.12428899109363556,
      "learning_rate": 3.01e-05,
      "loss": 0.7613,
      "step": 796
    },
    {
      "epoch": 1.0226552788804684,
      "grad_norm": 0.11932521313428879,
      "learning_rate": 3.0075000000000003e-05,
      "loss": 0.67,
      "step": 797
    },
    {
      "epoch": 1.0239384097197162,
      "grad_norm": 0.11373074352741241,
      "learning_rate": 3.0050000000000002e-05,
      "loss": 0.6214,
      "step": 798
    },
    {
      "epoch": 1.025221540558964,
      "grad_norm": 0.11149106919765472,
      "learning_rate": 3.0025000000000005e-05,
      "loss": 0.7325,
      "step": 799
    },
    {
      "epoch": 1.0265046713982116,
      "grad_norm": 0.11363644152879715,
      "learning_rate": 3e-05,
      "loss": 0.7096,
      "step": 800
    },
    {
      "epoch": 1.0277878022374594,
      "grad_norm": 0.12109331786632538,
      "learning_rate": 2.9975000000000004e-05,
      "loss": 0.7554,
      "step": 801
    },
    {
      "epoch": 1.0290709330767072,
      "grad_norm": 0.10524129867553711,
      "learning_rate": 2.995e-05,
      "loss": 0.6765,
      "step": 802
    },
    {
      "epoch": 1.030354063915955,
      "grad_norm": 0.11807513982057571,
      "learning_rate": 2.9925000000000002e-05,
      "loss": 0.6675,
      "step": 803
    },
    {
      "epoch": 1.0316371947552028,
      "grad_norm": 0.11643943935632706,
      "learning_rate": 2.9900000000000002e-05,
      "loss": 0.6746,
      "step": 804
    },
    {
      "epoch": 1.0329203255944504,
      "grad_norm": 0.11422648280858994,
      "learning_rate": 2.9875000000000004e-05,
      "loss": 0.6584,
      "step": 805
    },
    {
      "epoch": 1.0342034564336982,
      "grad_norm": 0.13378027081489563,
      "learning_rate": 2.985e-05,
      "loss": 0.7075,
      "step": 806
    },
    {
      "epoch": 1.035486587272946,
      "grad_norm": 0.11719894409179688,
      "learning_rate": 2.9825000000000003e-05,
      "loss": 0.7045,
      "step": 807
    },
    {
      "epoch": 1.0367697181121938,
      "grad_norm": 0.09588094800710678,
      "learning_rate": 2.98e-05,
      "loss": 0.7219,
      "step": 808
    },
    {
      "epoch": 1.0380528489514416,
      "grad_norm": 0.1114092692732811,
      "learning_rate": 2.9775000000000002e-05,
      "loss": 0.6583,
      "step": 809
    },
    {
      "epoch": 1.0393359797906894,
      "grad_norm": 0.12139885127544403,
      "learning_rate": 2.975e-05,
      "loss": 0.7356,
      "step": 810
    },
    {
      "epoch": 1.040619110629937,
      "grad_norm": 0.22631877660751343,
      "learning_rate": 2.9725000000000004e-05,
      "loss": 0.7075,
      "step": 811
    },
    {
      "epoch": 1.0419022414691848,
      "grad_norm": 0.13090193271636963,
      "learning_rate": 2.97e-05,
      "loss": 0.7631,
      "step": 812
    },
    {
      "epoch": 1.0431853723084326,
      "grad_norm": 0.16709767282009125,
      "learning_rate": 2.9675000000000003e-05,
      "loss": 0.6808,
      "step": 813
    },
    {
      "epoch": 1.0444685031476804,
      "grad_norm": 0.12479514628648758,
      "learning_rate": 2.965e-05,
      "loss": 0.7689,
      "step": 814
    },
    {
      "epoch": 1.0457516339869282,
      "grad_norm": 0.12561389803886414,
      "learning_rate": 2.9625000000000002e-05,
      "loss": 0.6884,
      "step": 815
    },
    {
      "epoch": 1.047034764826176,
      "grad_norm": 0.1445583701133728,
      "learning_rate": 2.96e-05,
      "loss": 0.6805,
      "step": 816
    },
    {
      "epoch": 1.0483178956654235,
      "grad_norm": 0.18690825998783112,
      "learning_rate": 2.9575000000000004e-05,
      "loss": 0.6526,
      "step": 817
    },
    {
      "epoch": 1.0496010265046714,
      "grad_norm": 0.10547809302806854,
      "learning_rate": 2.955e-05,
      "loss": 0.5546,
      "step": 818
    },
    {
      "epoch": 1.0508841573439192,
      "grad_norm": 0.12739458680152893,
      "learning_rate": 2.9525000000000003e-05,
      "loss": 0.7023,
      "step": 819
    },
    {
      "epoch": 1.052167288183167,
      "grad_norm": 0.14758267998695374,
      "learning_rate": 2.95e-05,
      "loss": 0.7306,
      "step": 820
    },
    {
      "epoch": 1.0534504190224148,
      "grad_norm": 0.15525870025157928,
      "learning_rate": 2.9475e-05,
      "loss": 0.654,
      "step": 821
    },
    {
      "epoch": 1.0547335498616626,
      "grad_norm": 0.10875718295574188,
      "learning_rate": 2.945e-05,
      "loss": 0.6931,
      "step": 822
    },
    {
      "epoch": 1.0560166807009101,
      "grad_norm": 0.13541480898857117,
      "learning_rate": 2.9425000000000004e-05,
      "loss": 0.7958,
      "step": 823
    },
    {
      "epoch": 1.057299811540158,
      "grad_norm": 0.2795816659927368,
      "learning_rate": 2.94e-05,
      "loss": 0.6568,
      "step": 824
    },
    {
      "epoch": 1.0585829423794058,
      "grad_norm": 0.1110072061419487,
      "learning_rate": 2.9375000000000003e-05,
      "loss": 0.6203,
      "step": 825
    },
    {
      "epoch": 1.0598660732186536,
      "grad_norm": 0.12494420260190964,
      "learning_rate": 2.935e-05,
      "loss": 0.6803,
      "step": 826
    },
    {
      "epoch": 1.0611492040579014,
      "grad_norm": 0.1429266333580017,
      "learning_rate": 2.9325e-05,
      "loss": 0.5458,
      "step": 827
    },
    {
      "epoch": 1.062432334897149,
      "grad_norm": 0.14799568057060242,
      "learning_rate": 2.93e-05,
      "loss": 0.6824,
      "step": 828
    },
    {
      "epoch": 1.0637154657363967,
      "grad_norm": 0.13354504108428955,
      "learning_rate": 2.9275000000000003e-05,
      "loss": 0.6073,
      "step": 829
    },
    {
      "epoch": 1.0649985965756446,
      "grad_norm": 0.11491900682449341,
      "learning_rate": 2.925e-05,
      "loss": 0.6188,
      "step": 830
    },
    {
      "epoch": 1.0662817274148924,
      "grad_norm": 0.23519422113895416,
      "learning_rate": 2.9225000000000002e-05,
      "loss": 0.6745,
      "step": 831
    },
    {
      "epoch": 1.0675648582541402,
      "grad_norm": 0.10423803329467773,
      "learning_rate": 2.9199999999999998e-05,
      "loss": 0.7543,
      "step": 832
    },
    {
      "epoch": 1.068847989093388,
      "grad_norm": 0.0953143835067749,
      "learning_rate": 2.9175e-05,
      "loss": 0.7016,
      "step": 833
    },
    {
      "epoch": 1.0701311199326355,
      "grad_norm": 0.13647860288619995,
      "learning_rate": 2.915e-05,
      "loss": 0.6921,
      "step": 834
    },
    {
      "epoch": 1.0714142507718833,
      "grad_norm": 0.16713190078735352,
      "learning_rate": 2.9125000000000003e-05,
      "loss": 0.6995,
      "step": 835
    },
    {
      "epoch": 1.0726973816111312,
      "grad_norm": 0.10679234564304352,
      "learning_rate": 2.91e-05,
      "loss": 0.6952,
      "step": 836
    },
    {
      "epoch": 1.073980512450379,
      "grad_norm": 0.13314153254032135,
      "learning_rate": 2.9075000000000002e-05,
      "loss": 0.6464,
      "step": 837
    },
    {
      "epoch": 1.0752636432896268,
      "grad_norm": 0.11489319801330566,
      "learning_rate": 2.9049999999999998e-05,
      "loss": 0.6773,
      "step": 838
    },
    {
      "epoch": 1.0765467741288743,
      "grad_norm": 0.12004727125167847,
      "learning_rate": 2.9025e-05,
      "loss": 0.6397,
      "step": 839
    },
    {
      "epoch": 1.0778299049681221,
      "grad_norm": 0.09766285866498947,
      "learning_rate": 2.9e-05,
      "loss": 0.6089,
      "step": 840
    },
    {
      "epoch": 1.07911303580737,
      "grad_norm": 0.21478769183158875,
      "learning_rate": 2.8975000000000003e-05,
      "loss": 0.5598,
      "step": 841
    },
    {
      "epoch": 1.0803961666466178,
      "grad_norm": 0.13094161450862885,
      "learning_rate": 2.895e-05,
      "loss": 0.7221,
      "step": 842
    },
    {
      "epoch": 1.0816792974858656,
      "grad_norm": 0.12187467515468597,
      "learning_rate": 2.8925000000000002e-05,
      "loss": 0.6657,
      "step": 843
    },
    {
      "epoch": 1.0829624283251134,
      "grad_norm": 0.12380946427583694,
      "learning_rate": 2.8899999999999998e-05,
      "loss": 0.7175,
      "step": 844
    },
    {
      "epoch": 1.084245559164361,
      "grad_norm": 0.10810744017362595,
      "learning_rate": 2.8875e-05,
      "loss": 0.6529,
      "step": 845
    },
    {
      "epoch": 1.0855286900036087,
      "grad_norm": 0.12645648419857025,
      "learning_rate": 2.885e-05,
      "loss": 0.6081,
      "step": 846
    },
    {
      "epoch": 1.0868118208428565,
      "grad_norm": 0.18685542047023773,
      "learning_rate": 2.8825000000000003e-05,
      "loss": 0.6164,
      "step": 847
    },
    {
      "epoch": 1.0880949516821043,
      "grad_norm": 0.10530265420675278,
      "learning_rate": 2.88e-05,
      "loss": 0.7204,
      "step": 848
    },
    {
      "epoch": 1.0893780825213522,
      "grad_norm": 0.16235588490962982,
      "learning_rate": 2.8775e-05,
      "loss": 0.6686,
      "step": 849
    },
    {
      "epoch": 1.0906612133606,
      "grad_norm": 0.1767028272151947,
      "learning_rate": 2.8749999999999997e-05,
      "loss": 0.7742,
      "step": 850
    },
    {
      "epoch": 1.0919443441998475,
      "grad_norm": 0.1402859389781952,
      "learning_rate": 2.8725e-05,
      "loss": 0.7348,
      "step": 851
    },
    {
      "epoch": 1.0932274750390953,
      "grad_norm": 0.18379436433315277,
      "learning_rate": 2.87e-05,
      "loss": 0.6384,
      "step": 852
    },
    {
      "epoch": 1.0945106058783431,
      "grad_norm": 0.163492351770401,
      "learning_rate": 2.8675000000000002e-05,
      "loss": 0.6029,
      "step": 853
    },
    {
      "epoch": 1.095793736717591,
      "grad_norm": 0.10406468063592911,
      "learning_rate": 2.865e-05,
      "loss": 0.7308,
      "step": 854
    },
    {
      "epoch": 1.0970768675568388,
      "grad_norm": 0.2764090895652771,
      "learning_rate": 2.8625e-05,
      "loss": 0.6124,
      "step": 855
    },
    {
      "epoch": 1.0983599983960866,
      "grad_norm": 0.14464637637138367,
      "learning_rate": 2.86e-05,
      "loss": 0.6838,
      "step": 856
    },
    {
      "epoch": 1.0996431292353341,
      "grad_norm": 0.13967151939868927,
      "learning_rate": 2.8575000000000003e-05,
      "loss": 0.6928,
      "step": 857
    },
    {
      "epoch": 1.100926260074582,
      "grad_norm": 0.12177060544490814,
      "learning_rate": 2.855e-05,
      "loss": 0.6594,
      "step": 858
    },
    {
      "epoch": 1.1022093909138297,
      "grad_norm": 0.13491985201835632,
      "learning_rate": 2.8525000000000002e-05,
      "loss": 0.7428,
      "step": 859
    },
    {
      "epoch": 1.1034925217530775,
      "grad_norm": 0.12143675982952118,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 0.8762,
      "step": 860
    },
    {
      "epoch": 1.1047756525923254,
      "grad_norm": 0.18120476603507996,
      "learning_rate": 2.8475e-05,
      "loss": 0.5823,
      "step": 861
    },
    {
      "epoch": 1.106058783431573,
      "grad_norm": 0.13680404424667358,
      "learning_rate": 2.845e-05,
      "loss": 0.7047,
      "step": 862
    },
    {
      "epoch": 1.1073419142708207,
      "grad_norm": 0.15481993556022644,
      "learning_rate": 2.8425000000000003e-05,
      "loss": 0.7763,
      "step": 863
    },
    {
      "epoch": 1.1086250451100685,
      "grad_norm": 0.15296027064323425,
      "learning_rate": 2.84e-05,
      "loss": 0.612,
      "step": 864
    },
    {
      "epoch": 1.1099081759493163,
      "grad_norm": 0.11018173396587372,
      "learning_rate": 2.8375000000000002e-05,
      "loss": 0.6218,
      "step": 865
    },
    {
      "epoch": 1.1111913067885641,
      "grad_norm": 0.10221578180789948,
      "learning_rate": 2.8349999999999998e-05,
      "loss": 0.7263,
      "step": 866
    },
    {
      "epoch": 1.112474437627812,
      "grad_norm": 0.12413452565670013,
      "learning_rate": 2.8325e-05,
      "loss": 0.7254,
      "step": 867
    },
    {
      "epoch": 1.1137575684670595,
      "grad_norm": 0.10972278565168381,
      "learning_rate": 2.83e-05,
      "loss": 0.7251,
      "step": 868
    },
    {
      "epoch": 1.1150406993063073,
      "grad_norm": 0.10585711151361465,
      "learning_rate": 2.8275000000000003e-05,
      "loss": 0.75,
      "step": 869
    },
    {
      "epoch": 1.1163238301455551,
      "grad_norm": 0.09238840639591217,
      "learning_rate": 2.825e-05,
      "loss": 0.5907,
      "step": 870
    },
    {
      "epoch": 1.117606960984803,
      "grad_norm": 0.112511545419693,
      "learning_rate": 2.8225e-05,
      "loss": 0.6368,
      "step": 871
    },
    {
      "epoch": 1.1188900918240507,
      "grad_norm": 0.138427272439003,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 0.6889,
      "step": 872
    },
    {
      "epoch": 1.1201732226632986,
      "grad_norm": 0.11633303016424179,
      "learning_rate": 2.8175e-05,
      "loss": 0.719,
      "step": 873
    },
    {
      "epoch": 1.1214563535025461,
      "grad_norm": 0.13872164487838745,
      "learning_rate": 2.815e-05,
      "loss": 0.7132,
      "step": 874
    },
    {
      "epoch": 1.122739484341794,
      "grad_norm": 0.11514315754175186,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 0.699,
      "step": 875
    },
    {
      "epoch": 1.1240226151810417,
      "grad_norm": 0.1311374008655548,
      "learning_rate": 2.8100000000000005e-05,
      "loss": 0.8298,
      "step": 876
    },
    {
      "epoch": 1.1253057460202895,
      "grad_norm": 0.13604219257831573,
      "learning_rate": 2.8075e-05,
      "loss": 0.7054,
      "step": 877
    },
    {
      "epoch": 1.1265888768595373,
      "grad_norm": 10.313078880310059,
      "learning_rate": 2.8050000000000004e-05,
      "loss": 0.5702,
      "step": 878
    },
    {
      "epoch": 1.1278720076987852,
      "grad_norm": 0.15520745515823364,
      "learning_rate": 2.8025e-05,
      "loss": 0.6732,
      "step": 879
    },
    {
      "epoch": 1.1291551385380327,
      "grad_norm": 0.14060474932193756,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.6678,
      "step": 880
    },
    {
      "epoch": 1.1304382693772805,
      "grad_norm": 0.11521503329277039,
      "learning_rate": 2.7975000000000002e-05,
      "loss": 0.6354,
      "step": 881
    },
    {
      "epoch": 1.1317214002165283,
      "grad_norm": 0.11902789771556854,
      "learning_rate": 2.7950000000000005e-05,
      "loss": 0.7709,
      "step": 882
    },
    {
      "epoch": 1.1330045310557761,
      "grad_norm": 0.1732594072818756,
      "learning_rate": 2.7925e-05,
      "loss": 0.594,
      "step": 883
    },
    {
      "epoch": 1.134287661895024,
      "grad_norm": 0.12338356673717499,
      "learning_rate": 2.7900000000000004e-05,
      "loss": 0.7031,
      "step": 884
    },
    {
      "epoch": 1.1355707927342715,
      "grad_norm": 0.09952358901500702,
      "learning_rate": 2.7875e-05,
      "loss": 0.7444,
      "step": 885
    },
    {
      "epoch": 1.1368539235735193,
      "grad_norm": 0.15083059668540955,
      "learning_rate": 2.7850000000000003e-05,
      "loss": 0.8254,
      "step": 886
    },
    {
      "epoch": 1.1381370544127671,
      "grad_norm": 0.1394781470298767,
      "learning_rate": 2.7825000000000002e-05,
      "loss": 0.7361,
      "step": 887
    },
    {
      "epoch": 1.139420185252015,
      "grad_norm": 0.11669556796550751,
      "learning_rate": 2.7800000000000005e-05,
      "loss": 0.6562,
      "step": 888
    },
    {
      "epoch": 1.1407033160912627,
      "grad_norm": 0.1053437888622284,
      "learning_rate": 2.7775e-05,
      "loss": 0.685,
      "step": 889
    },
    {
      "epoch": 1.1419864469305105,
      "grad_norm": 0.11455594748258591,
      "learning_rate": 2.7750000000000004e-05,
      "loss": 0.623,
      "step": 890
    },
    {
      "epoch": 1.1432695777697581,
      "grad_norm": 0.14740270376205444,
      "learning_rate": 2.7725e-05,
      "loss": 0.6943,
      "step": 891
    },
    {
      "epoch": 1.144552708609006,
      "grad_norm": 0.10675308108329773,
      "learning_rate": 2.7700000000000002e-05,
      "loss": 0.6738,
      "step": 892
    },
    {
      "epoch": 1.1458358394482537,
      "grad_norm": 0.1463991105556488,
      "learning_rate": 2.7675000000000002e-05,
      "loss": 0.5661,
      "step": 893
    },
    {
      "epoch": 1.1471189702875015,
      "grad_norm": 0.17814530432224274,
      "learning_rate": 2.7650000000000005e-05,
      "loss": 0.6642,
      "step": 894
    },
    {
      "epoch": 1.1484021011267493,
      "grad_norm": 0.11428478360176086,
      "learning_rate": 2.7625e-05,
      "loss": 0.6403,
      "step": 895
    },
    {
      "epoch": 1.149685231965997,
      "grad_norm": 0.1191159039735794,
      "learning_rate": 2.7600000000000003e-05,
      "loss": 0.7357,
      "step": 896
    },
    {
      "epoch": 1.1509683628052447,
      "grad_norm": 0.13085156679153442,
      "learning_rate": 2.7575e-05,
      "loss": 0.7411,
      "step": 897
    },
    {
      "epoch": 1.1522514936444925,
      "grad_norm": 0.21236440539360046,
      "learning_rate": 2.7550000000000002e-05,
      "loss": 0.7155,
      "step": 898
    },
    {
      "epoch": 1.1535346244837403,
      "grad_norm": 0.11280061304569244,
      "learning_rate": 2.7525e-05,
      "loss": 0.7569,
      "step": 899
    },
    {
      "epoch": 1.1548177553229881,
      "grad_norm": 0.13036152720451355,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 0.7439,
      "step": 900
    },
    {
      "epoch": 1.156100886162236,
      "grad_norm": 0.11562920361757278,
      "learning_rate": 2.7475e-05,
      "loss": 0.7294,
      "step": 901
    },
    {
      "epoch": 1.1573840170014837,
      "grad_norm": 0.1192454844713211,
      "learning_rate": 2.7450000000000003e-05,
      "loss": 0.6651,
      "step": 902
    },
    {
      "epoch": 1.1586671478407313,
      "grad_norm": 0.1082087904214859,
      "learning_rate": 2.7425e-05,
      "loss": 0.6442,
      "step": 903
    },
    {
      "epoch": 1.1599502786799791,
      "grad_norm": 0.1306343376636505,
      "learning_rate": 2.7400000000000002e-05,
      "loss": 0.743,
      "step": 904
    },
    {
      "epoch": 1.161233409519227,
      "grad_norm": 0.15252164006233215,
      "learning_rate": 2.7375e-05,
      "loss": 0.7173,
      "step": 905
    },
    {
      "epoch": 1.1625165403584747,
      "grad_norm": 0.12856437265872955,
      "learning_rate": 2.7350000000000004e-05,
      "loss": 0.6934,
      "step": 906
    },
    {
      "epoch": 1.1637996711977223,
      "grad_norm": 0.15587373077869415,
      "learning_rate": 2.7325e-05,
      "loss": 0.6987,
      "step": 907
    },
    {
      "epoch": 1.1650828020369701,
      "grad_norm": 0.1130947545170784,
      "learning_rate": 2.7300000000000003e-05,
      "loss": 0.7228,
      "step": 908
    },
    {
      "epoch": 1.166365932876218,
      "grad_norm": 0.1268240511417389,
      "learning_rate": 2.7275e-05,
      "loss": 0.7641,
      "step": 909
    },
    {
      "epoch": 1.1676490637154657,
      "grad_norm": 0.21334180235862732,
      "learning_rate": 2.725e-05,
      "loss": 0.7868,
      "step": 910
    },
    {
      "epoch": 1.1689321945547135,
      "grad_norm": 0.1154848113656044,
      "learning_rate": 2.7225e-05,
      "loss": 0.6623,
      "step": 911
    },
    {
      "epoch": 1.1702153253939613,
      "grad_norm": 0.11536215245723724,
      "learning_rate": 2.7200000000000004e-05,
      "loss": 0.6398,
      "step": 912
    },
    {
      "epoch": 1.1714984562332091,
      "grad_norm": 4.682082176208496,
      "learning_rate": 2.7175e-05,
      "loss": 0.6452,
      "step": 913
    },
    {
      "epoch": 1.1727815870724567,
      "grad_norm": 0.1865924894809723,
      "learning_rate": 2.7150000000000003e-05,
      "loss": 0.7821,
      "step": 914
    },
    {
      "epoch": 1.1740647179117045,
      "grad_norm": 0.14829710125923157,
      "learning_rate": 2.7125000000000002e-05,
      "loss": 0.6608,
      "step": 915
    },
    {
      "epoch": 1.1753478487509523,
      "grad_norm": 0.18191851675510406,
      "learning_rate": 2.7100000000000005e-05,
      "loss": 0.6625,
      "step": 916
    },
    {
      "epoch": 1.1766309795902001,
      "grad_norm": 0.4607589542865753,
      "learning_rate": 2.7075e-05,
      "loss": 0.7108,
      "step": 917
    },
    {
      "epoch": 1.177914110429448,
      "grad_norm": 0.15805894136428833,
      "learning_rate": 2.7050000000000004e-05,
      "loss": 0.595,
      "step": 918
    },
    {
      "epoch": 1.1791972412686955,
      "grad_norm": 0.15470515191555023,
      "learning_rate": 2.7025e-05,
      "loss": 0.5727,
      "step": 919
    },
    {
      "epoch": 1.1804803721079433,
      "grad_norm": 0.14943130314350128,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 0.6721,
      "step": 920
    },
    {
      "epoch": 1.1817635029471911,
      "grad_norm": 0.13314729928970337,
      "learning_rate": 2.6975000000000002e-05,
      "loss": 0.7726,
      "step": 921
    },
    {
      "epoch": 1.183046633786439,
      "grad_norm": 0.12177535146474838,
      "learning_rate": 2.6950000000000005e-05,
      "loss": 0.8218,
      "step": 922
    },
    {
      "epoch": 1.1843297646256867,
      "grad_norm": 0.1176556870341301,
      "learning_rate": 2.6925e-05,
      "loss": 0.621,
      "step": 923
    },
    {
      "epoch": 1.1856128954649345,
      "grad_norm": 0.1356288641691208,
      "learning_rate": 2.6900000000000003e-05,
      "loss": 0.6812,
      "step": 924
    },
    {
      "epoch": 1.1868960263041821,
      "grad_norm": 0.16944445669651031,
      "learning_rate": 2.6875e-05,
      "loss": 0.6837,
      "step": 925
    },
    {
      "epoch": 1.18817915714343,
      "grad_norm": 0.11072086542844772,
      "learning_rate": 2.6850000000000002e-05,
      "loss": 0.6015,
      "step": 926
    },
    {
      "epoch": 1.1894622879826777,
      "grad_norm": 0.12054786831140518,
      "learning_rate": 2.6825e-05,
      "loss": 0.7146,
      "step": 927
    },
    {
      "epoch": 1.1907454188219255,
      "grad_norm": 0.15583249926567078,
      "learning_rate": 2.6800000000000004e-05,
      "loss": 0.7056,
      "step": 928
    },
    {
      "epoch": 1.1920285496611733,
      "grad_norm": 0.11561548709869385,
      "learning_rate": 2.6775e-05,
      "loss": 0.6325,
      "step": 929
    },
    {
      "epoch": 1.193311680500421,
      "grad_norm": 0.09692132472991943,
      "learning_rate": 2.6750000000000003e-05,
      "loss": 0.7336,
      "step": 930
    },
    {
      "epoch": 1.1945948113396687,
      "grad_norm": 0.12546059489250183,
      "learning_rate": 2.6725e-05,
      "loss": 0.6581,
      "step": 931
    },
    {
      "epoch": 1.1958779421789165,
      "grad_norm": 0.15935397148132324,
      "learning_rate": 2.6700000000000002e-05,
      "loss": 0.7759,
      "step": 932
    },
    {
      "epoch": 1.1971610730181643,
      "grad_norm": 0.116784006357193,
      "learning_rate": 2.6675e-05,
      "loss": 0.7016,
      "step": 933
    },
    {
      "epoch": 1.1984442038574121,
      "grad_norm": 0.12242651730775833,
      "learning_rate": 2.6650000000000004e-05,
      "loss": 0.6406,
      "step": 934
    },
    {
      "epoch": 1.19972733469666,
      "grad_norm": 0.1190212070941925,
      "learning_rate": 2.6625e-05,
      "loss": 0.7528,
      "step": 935
    },
    {
      "epoch": 1.2010104655359077,
      "grad_norm": 0.10493738204240799,
      "learning_rate": 2.6600000000000003e-05,
      "loss": 0.6294,
      "step": 936
    },
    {
      "epoch": 1.2022935963751553,
      "grad_norm": 0.11646634340286255,
      "learning_rate": 2.6575e-05,
      "loss": 0.7344,
      "step": 937
    },
    {
      "epoch": 1.2035767272144031,
      "grad_norm": 0.16580480337142944,
      "learning_rate": 2.655e-05,
      "loss": 0.6346,
      "step": 938
    },
    {
      "epoch": 1.204859858053651,
      "grad_norm": 0.09907055646181107,
      "learning_rate": 2.6525e-05,
      "loss": 0.6907,
      "step": 939
    },
    {
      "epoch": 1.2061429888928987,
      "grad_norm": 0.11516622453927994,
      "learning_rate": 2.6500000000000004e-05,
      "loss": 0.7124,
      "step": 940
    },
    {
      "epoch": 1.2074261197321465,
      "grad_norm": 0.11797518283128738,
      "learning_rate": 2.6475e-05,
      "loss": 0.6035,
      "step": 941
    },
    {
      "epoch": 1.2087092505713941,
      "grad_norm": 0.13182587921619415,
      "learning_rate": 2.6450000000000003e-05,
      "loss": 0.5501,
      "step": 942
    },
    {
      "epoch": 1.209992381410642,
      "grad_norm": 0.12415479123592377,
      "learning_rate": 2.6425e-05,
      "loss": 0.7676,
      "step": 943
    },
    {
      "epoch": 1.2112755122498897,
      "grad_norm": 0.11650388687849045,
      "learning_rate": 2.64e-05,
      "loss": 0.6739,
      "step": 944
    },
    {
      "epoch": 1.2125586430891375,
      "grad_norm": 0.1635390967130661,
      "learning_rate": 2.6375e-05,
      "loss": 0.7001,
      "step": 945
    },
    {
      "epoch": 1.2138417739283853,
      "grad_norm": 0.11738403141498566,
      "learning_rate": 2.6350000000000004e-05,
      "loss": 0.6934,
      "step": 946
    },
    {
      "epoch": 1.2151249047676331,
      "grad_norm": 0.11615172028541565,
      "learning_rate": 2.6325e-05,
      "loss": 0.5707,
      "step": 947
    },
    {
      "epoch": 1.2164080356068807,
      "grad_norm": 0.17213615775108337,
      "learning_rate": 2.6300000000000002e-05,
      "loss": 0.7482,
      "step": 948
    },
    {
      "epoch": 1.2176911664461285,
      "grad_norm": 0.1065833643078804,
      "learning_rate": 2.6275e-05,
      "loss": 0.6951,
      "step": 949
    },
    {
      "epoch": 1.2189742972853763,
      "grad_norm": 0.12293951958417892,
      "learning_rate": 2.625e-05,
      "loss": 0.649,
      "step": 950
    },
    {
      "epoch": 1.2202574281246241,
      "grad_norm": 0.11064042896032333,
      "learning_rate": 2.6225e-05,
      "loss": 0.5912,
      "step": 951
    },
    {
      "epoch": 1.221540558963872,
      "grad_norm": 0.10693170875310898,
      "learning_rate": 2.6200000000000003e-05,
      "loss": 0.6951,
      "step": 952
    },
    {
      "epoch": 1.2228236898031195,
      "grad_norm": 0.12470885366201401,
      "learning_rate": 2.6175e-05,
      "loss": 0.577,
      "step": 953
    },
    {
      "epoch": 1.2241068206423673,
      "grad_norm": 0.11562110483646393,
      "learning_rate": 2.6150000000000002e-05,
      "loss": 0.6705,
      "step": 954
    },
    {
      "epoch": 1.2253899514816151,
      "grad_norm": 0.11370087414979935,
      "learning_rate": 2.6124999999999998e-05,
      "loss": 0.6191,
      "step": 955
    },
    {
      "epoch": 1.226673082320863,
      "grad_norm": 0.13788475096225739,
      "learning_rate": 2.61e-05,
      "loss": 0.6803,
      "step": 956
    },
    {
      "epoch": 1.2279562131601107,
      "grad_norm": 0.11830168217420578,
      "learning_rate": 2.6075e-05,
      "loss": 0.647,
      "step": 957
    },
    {
      "epoch": 1.2292393439993585,
      "grad_norm": 0.12710095942020416,
      "learning_rate": 2.6050000000000003e-05,
      "loss": 0.6619,
      "step": 958
    },
    {
      "epoch": 1.230522474838606,
      "grad_norm": 0.10318832099437714,
      "learning_rate": 2.6025e-05,
      "loss": 0.6116,
      "step": 959
    },
    {
      "epoch": 1.231805605677854,
      "grad_norm": 0.11693265289068222,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 0.6532,
      "step": 960
    },
    {
      "epoch": 1.2330887365171017,
      "grad_norm": 0.16333059966564178,
      "learning_rate": 2.5974999999999998e-05,
      "loss": 0.6669,
      "step": 961
    },
    {
      "epoch": 1.2343718673563495,
      "grad_norm": 0.12276259809732437,
      "learning_rate": 2.595e-05,
      "loss": 0.7274,
      "step": 962
    },
    {
      "epoch": 1.2356549981955973,
      "grad_norm": 0.10886857658624649,
      "learning_rate": 2.5925e-05,
      "loss": 0.6669,
      "step": 963
    },
    {
      "epoch": 1.236938129034845,
      "grad_norm": 0.1339716911315918,
      "learning_rate": 2.5900000000000003e-05,
      "loss": 0.6721,
      "step": 964
    },
    {
      "epoch": 1.2382212598740927,
      "grad_norm": 0.12599897384643555,
      "learning_rate": 2.5875e-05,
      "loss": 0.7109,
      "step": 965
    },
    {
      "epoch": 1.2395043907133405,
      "grad_norm": 0.10163362324237823,
      "learning_rate": 2.585e-05,
      "loss": 0.7305,
      "step": 966
    },
    {
      "epoch": 1.2407875215525883,
      "grad_norm": 0.1281915307044983,
      "learning_rate": 2.5824999999999998e-05,
      "loss": 0.6291,
      "step": 967
    },
    {
      "epoch": 1.2420706523918361,
      "grad_norm": 0.10489276796579361,
      "learning_rate": 2.58e-05,
      "loss": 0.5181,
      "step": 968
    },
    {
      "epoch": 1.243353783231084,
      "grad_norm": 0.13227961957454681,
      "learning_rate": 2.5775e-05,
      "loss": 0.7861,
      "step": 969
    },
    {
      "epoch": 1.2446369140703317,
      "grad_norm": 0.13125702738761902,
      "learning_rate": 2.5750000000000002e-05,
      "loss": 0.6256,
      "step": 970
    },
    {
      "epoch": 1.2459200449095793,
      "grad_norm": 0.16621635854244232,
      "learning_rate": 2.5725e-05,
      "loss": 0.7175,
      "step": 971
    },
    {
      "epoch": 1.247203175748827,
      "grad_norm": 0.1337946653366089,
      "learning_rate": 2.57e-05,
      "loss": 0.6826,
      "step": 972
    },
    {
      "epoch": 1.248486306588075,
      "grad_norm": 0.10191724449396133,
      "learning_rate": 2.5675e-05,
      "loss": 0.7202,
      "step": 973
    },
    {
      "epoch": 1.2497694374273227,
      "grad_norm": 0.10166647285223007,
      "learning_rate": 2.5650000000000003e-05,
      "loss": 0.7911,
      "step": 974
    },
    {
      "epoch": 1.2510525682665703,
      "grad_norm": 0.11532244086265564,
      "learning_rate": 2.5625e-05,
      "loss": 0.6769,
      "step": 975
    },
    {
      "epoch": 1.252335699105818,
      "grad_norm": 0.11956571787595749,
      "learning_rate": 2.5600000000000002e-05,
      "loss": 0.6361,
      "step": 976
    },
    {
      "epoch": 1.253618829945066,
      "grad_norm": 0.3616163730621338,
      "learning_rate": 2.5574999999999998e-05,
      "loss": 0.7219,
      "step": 977
    },
    {
      "epoch": 1.2549019607843137,
      "grad_norm": 0.1399337202310562,
      "learning_rate": 2.555e-05,
      "loss": 0.6205,
      "step": 978
    },
    {
      "epoch": 1.2561850916235615,
      "grad_norm": 0.1864340454339981,
      "learning_rate": 2.5525e-05,
      "loss": 0.6125,
      "step": 979
    },
    {
      "epoch": 1.2574682224628093,
      "grad_norm": 0.10967881232500076,
      "learning_rate": 2.5500000000000003e-05,
      "loss": 0.7184,
      "step": 980
    },
    {
      "epoch": 1.2587513533020571,
      "grad_norm": 0.12101941555738449,
      "learning_rate": 2.5475e-05,
      "loss": 0.646,
      "step": 981
    },
    {
      "epoch": 1.260034484141305,
      "grad_norm": 0.13449952006340027,
      "learning_rate": 2.5450000000000002e-05,
      "loss": 0.6648,
      "step": 982
    },
    {
      "epoch": 1.2613176149805525,
      "grad_norm": 0.1359936147928238,
      "learning_rate": 2.5424999999999998e-05,
      "loss": 0.7456,
      "step": 983
    },
    {
      "epoch": 1.2626007458198003,
      "grad_norm": 0.12955546379089355,
      "learning_rate": 2.54e-05,
      "loss": 0.7137,
      "step": 984
    },
    {
      "epoch": 1.2638838766590481,
      "grad_norm": 0.12510475516319275,
      "learning_rate": 2.5375e-05,
      "loss": 0.7361,
      "step": 985
    },
    {
      "epoch": 1.265167007498296,
      "grad_norm": 0.18031591176986694,
      "learning_rate": 2.5350000000000003e-05,
      "loss": 0.6644,
      "step": 986
    },
    {
      "epoch": 1.2664501383375435,
      "grad_norm": 0.13183267414569855,
      "learning_rate": 2.5325e-05,
      "loss": 0.6472,
      "step": 987
    },
    {
      "epoch": 1.2677332691767913,
      "grad_norm": 0.1619541347026825,
      "learning_rate": 2.5300000000000002e-05,
      "loss": 0.7627,
      "step": 988
    },
    {
      "epoch": 1.269016400016039,
      "grad_norm": 0.15462824702262878,
      "learning_rate": 2.5274999999999998e-05,
      "loss": 0.7073,
      "step": 989
    },
    {
      "epoch": 1.270299530855287,
      "grad_norm": 0.1697506308555603,
      "learning_rate": 2.525e-05,
      "loss": 0.5367,
      "step": 990
    },
    {
      "epoch": 1.2715826616945347,
      "grad_norm": 0.18223468959331512,
      "learning_rate": 2.5225e-05,
      "loss": 0.6379,
      "step": 991
    },
    {
      "epoch": 1.2728657925337825,
      "grad_norm": 0.1298186182975769,
      "learning_rate": 2.5200000000000003e-05,
      "loss": 0.6323,
      "step": 992
    },
    {
      "epoch": 1.2741489233730303,
      "grad_norm": 0.13207226991653442,
      "learning_rate": 2.5175e-05,
      "loss": 0.6537,
      "step": 993
    },
    {
      "epoch": 1.275432054212278,
      "grad_norm": 0.1383514702320099,
      "learning_rate": 2.515e-05,
      "loss": 0.75,
      "step": 994
    },
    {
      "epoch": 1.2767151850515257,
      "grad_norm": 0.14963077008724213,
      "learning_rate": 2.5124999999999997e-05,
      "loss": 0.7162,
      "step": 995
    },
    {
      "epoch": 1.2779983158907735,
      "grad_norm": 0.11118656396865845,
      "learning_rate": 2.51e-05,
      "loss": 0.5711,
      "step": 996
    },
    {
      "epoch": 1.2792814467300213,
      "grad_norm": 0.12384391576051712,
      "learning_rate": 2.5075e-05,
      "loss": 0.5434,
      "step": 997
    },
    {
      "epoch": 1.280564577569269,
      "grad_norm": 0.17304004728794098,
      "learning_rate": 2.5050000000000002e-05,
      "loss": 0.8326,
      "step": 998
    },
    {
      "epoch": 1.2818477084085167,
      "grad_norm": 0.1363946944475174,
      "learning_rate": 2.5025e-05,
      "loss": 0.5917,
      "step": 999
    },
    {
      "epoch": 1.2831308392477645,
      "grad_norm": 0.12078433483839035,
      "learning_rate": 2.5e-05,
      "loss": 0.7606,
      "step": 1000
    },
    {
      "epoch": 1.2844139700870123,
      "grad_norm": 0.1164424866437912,
      "learning_rate": 2.4975e-05,
      "loss": 0.6775,
      "step": 1001
    },
    {
      "epoch": 1.28569710092626,
      "grad_norm": 0.1490689218044281,
      "learning_rate": 2.495e-05,
      "loss": 0.5683,
      "step": 1002
    },
    {
      "epoch": 1.286980231765508,
      "grad_norm": 0.1189810112118721,
      "learning_rate": 2.4925000000000003e-05,
      "loss": 0.7295,
      "step": 1003
    },
    {
      "epoch": 1.2882633626047557,
      "grad_norm": 0.1118491068482399,
      "learning_rate": 2.4900000000000002e-05,
      "loss": 0.7053,
      "step": 1004
    },
    {
      "epoch": 1.2895464934440033,
      "grad_norm": 0.12503305077552795,
      "learning_rate": 2.4875e-05,
      "loss": 0.6253,
      "step": 1005
    },
    {
      "epoch": 1.290829624283251,
      "grad_norm": 0.10628114640712738,
      "learning_rate": 2.485e-05,
      "loss": 0.7128,
      "step": 1006
    },
    {
      "epoch": 1.292112755122499,
      "grad_norm": 0.11936908215284348,
      "learning_rate": 2.4825e-05,
      "loss": 0.6468,
      "step": 1007
    },
    {
      "epoch": 1.2933958859617467,
      "grad_norm": 0.11343146860599518,
      "learning_rate": 2.48e-05,
      "loss": 0.7056,
      "step": 1008
    },
    {
      "epoch": 1.2946790168009945,
      "grad_norm": 0.11086481809616089,
      "learning_rate": 2.4775000000000003e-05,
      "loss": 0.6016,
      "step": 1009
    },
    {
      "epoch": 1.295962147640242,
      "grad_norm": 0.09942571073770523,
      "learning_rate": 2.4750000000000002e-05,
      "loss": 0.6695,
      "step": 1010
    },
    {
      "epoch": 1.29724527847949,
      "grad_norm": 0.12788325548171997,
      "learning_rate": 2.4725e-05,
      "loss": 0.6088,
      "step": 1011
    },
    {
      "epoch": 1.2985284093187377,
      "grad_norm": 0.1217220202088356,
      "learning_rate": 2.47e-05,
      "loss": 0.7211,
      "step": 1012
    },
    {
      "epoch": 1.2998115401579855,
      "grad_norm": 0.11650361865758896,
      "learning_rate": 2.4675e-05,
      "loss": 0.6924,
      "step": 1013
    },
    {
      "epoch": 1.3010946709972333,
      "grad_norm": 0.10439816862344742,
      "learning_rate": 2.465e-05,
      "loss": 0.57,
      "step": 1014
    },
    {
      "epoch": 1.302377801836481,
      "grad_norm": 0.11899355798959732,
      "learning_rate": 2.4625000000000002e-05,
      "loss": 0.7415,
      "step": 1015
    },
    {
      "epoch": 1.303660932675729,
      "grad_norm": 0.1103866845369339,
      "learning_rate": 2.46e-05,
      "loss": 0.6781,
      "step": 1016
    },
    {
      "epoch": 1.3049440635149765,
      "grad_norm": 0.13989798724651337,
      "learning_rate": 2.4575e-05,
      "loss": 0.6155,
      "step": 1017
    },
    {
      "epoch": 1.3062271943542243,
      "grad_norm": 0.12829771637916565,
      "learning_rate": 2.455e-05,
      "loss": 0.6187,
      "step": 1018
    },
    {
      "epoch": 1.307510325193472,
      "grad_norm": 0.12145745754241943,
      "learning_rate": 2.4525e-05,
      "loss": 0.6938,
      "step": 1019
    },
    {
      "epoch": 1.30879345603272,
      "grad_norm": 0.15302674472332,
      "learning_rate": 2.45e-05,
      "loss": 0.7106,
      "step": 1020
    },
    {
      "epoch": 1.3100765868719675,
      "grad_norm": 0.12946799397468567,
      "learning_rate": 2.4475000000000002e-05,
      "loss": 0.7967,
      "step": 1021
    },
    {
      "epoch": 1.3113597177112153,
      "grad_norm": 0.13514703512191772,
      "learning_rate": 2.445e-05,
      "loss": 0.708,
      "step": 1022
    },
    {
      "epoch": 1.312642848550463,
      "grad_norm": 0.1839998960494995,
      "learning_rate": 2.4425e-05,
      "loss": 0.7555,
      "step": 1023
    },
    {
      "epoch": 1.313925979389711,
      "grad_norm": 0.10689768195152283,
      "learning_rate": 2.44e-05,
      "loss": 0.7399,
      "step": 1024
    },
    {
      "epoch": 1.3152091102289587,
      "grad_norm": 0.14267092943191528,
      "learning_rate": 2.4375e-05,
      "loss": 0.6812,
      "step": 1025
    },
    {
      "epoch": 1.3164922410682065,
      "grad_norm": 0.12077357620000839,
      "learning_rate": 2.435e-05,
      "loss": 0.7238,
      "step": 1026
    },
    {
      "epoch": 1.3177753719074543,
      "grad_norm": 0.14030152559280396,
      "learning_rate": 2.4325000000000002e-05,
      "loss": 0.6587,
      "step": 1027
    },
    {
      "epoch": 1.3190585027467019,
      "grad_norm": 0.13563835620880127,
      "learning_rate": 2.43e-05,
      "loss": 0.7005,
      "step": 1028
    },
    {
      "epoch": 1.3203416335859497,
      "grad_norm": 0.2371646761894226,
      "learning_rate": 2.4275e-05,
      "loss": 0.6872,
      "step": 1029
    },
    {
      "epoch": 1.3216247644251975,
      "grad_norm": 0.11393280327320099,
      "learning_rate": 2.425e-05,
      "loss": 0.6939,
      "step": 1030
    },
    {
      "epoch": 1.3229078952644453,
      "grad_norm": 0.1864841729402542,
      "learning_rate": 2.4225e-05,
      "loss": 0.7445,
      "step": 1031
    },
    {
      "epoch": 1.3241910261036929,
      "grad_norm": 0.11596901714801788,
      "learning_rate": 2.4200000000000002e-05,
      "loss": 0.7405,
      "step": 1032
    },
    {
      "epoch": 1.3254741569429407,
      "grad_norm": 0.12473506480455399,
      "learning_rate": 2.4175e-05,
      "loss": 0.7567,
      "step": 1033
    },
    {
      "epoch": 1.3267572877821885,
      "grad_norm": 0.14197614789009094,
      "learning_rate": 2.415e-05,
      "loss": 0.761,
      "step": 1034
    },
    {
      "epoch": 1.3280404186214363,
      "grad_norm": 0.12488014996051788,
      "learning_rate": 2.4125e-05,
      "loss": 0.6758,
      "step": 1035
    },
    {
      "epoch": 1.329323549460684,
      "grad_norm": 0.11762822419404984,
      "learning_rate": 2.41e-05,
      "loss": 0.622,
      "step": 1036
    },
    {
      "epoch": 1.330606680299932,
      "grad_norm": 0.13053925335407257,
      "learning_rate": 2.4075e-05,
      "loss": 0.7081,
      "step": 1037
    },
    {
      "epoch": 1.3318898111391797,
      "grad_norm": 0.10392630845308304,
      "learning_rate": 2.4050000000000002e-05,
      "loss": 0.7113,
      "step": 1038
    },
    {
      "epoch": 1.3331729419784273,
      "grad_norm": 0.21379707753658295,
      "learning_rate": 2.4025e-05,
      "loss": 0.6793,
      "step": 1039
    },
    {
      "epoch": 1.334456072817675,
      "grad_norm": 0.11627639085054398,
      "learning_rate": 2.4e-05,
      "loss": 0.5739,
      "step": 1040
    },
    {
      "epoch": 1.335739203656923,
      "grad_norm": 0.17493794858455658,
      "learning_rate": 2.3975e-05,
      "loss": 0.6695,
      "step": 1041
    },
    {
      "epoch": 1.3370223344961707,
      "grad_norm": 0.12904396653175354,
      "learning_rate": 2.395e-05,
      "loss": 0.6393,
      "step": 1042
    },
    {
      "epoch": 1.3383054653354185,
      "grad_norm": 0.10722175240516663,
      "learning_rate": 2.3925e-05,
      "loss": 0.6701,
      "step": 1043
    },
    {
      "epoch": 1.339588596174666,
      "grad_norm": 0.13406391441822052,
      "learning_rate": 2.39e-05,
      "loss": 0.7786,
      "step": 1044
    },
    {
      "epoch": 1.3408717270139139,
      "grad_norm": 0.18648472428321838,
      "learning_rate": 2.3875e-05,
      "loss": 0.7326,
      "step": 1045
    },
    {
      "epoch": 1.3421548578531617,
      "grad_norm": 0.11319169402122498,
      "learning_rate": 2.385e-05,
      "loss": 0.6914,
      "step": 1046
    },
    {
      "epoch": 1.3434379886924095,
      "grad_norm": 0.12263406068086624,
      "learning_rate": 2.3825e-05,
      "loss": 0.7468,
      "step": 1047
    },
    {
      "epoch": 1.3447211195316573,
      "grad_norm": 0.1497359275817871,
      "learning_rate": 2.38e-05,
      "loss": 0.6476,
      "step": 1048
    },
    {
      "epoch": 1.346004250370905,
      "grad_norm": 0.16893704235553741,
      "learning_rate": 2.3775e-05,
      "loss": 0.6905,
      "step": 1049
    },
    {
      "epoch": 1.347287381210153,
      "grad_norm": 0.11611008644104004,
      "learning_rate": 2.375e-05,
      "loss": 0.5949,
      "step": 1050
    },
    {
      "epoch": 1.3485705120494005,
      "grad_norm": 0.25994718074798584,
      "learning_rate": 2.3725e-05,
      "loss": 0.6881,
      "step": 1051
    },
    {
      "epoch": 1.3498536428886483,
      "grad_norm": 0.17757757008075714,
      "learning_rate": 2.37e-05,
      "loss": 0.7747,
      "step": 1052
    },
    {
      "epoch": 1.351136773727896,
      "grad_norm": 0.2415735274553299,
      "learning_rate": 2.3675e-05,
      "loss": 0.6032,
      "step": 1053
    },
    {
      "epoch": 1.352419904567144,
      "grad_norm": 0.14884178340435028,
      "learning_rate": 2.365e-05,
      "loss": 0.5853,
      "step": 1054
    },
    {
      "epoch": 1.3537030354063915,
      "grad_norm": 0.14867499470710754,
      "learning_rate": 2.3624999999999998e-05,
      "loss": 0.6199,
      "step": 1055
    },
    {
      "epoch": 1.3549861662456393,
      "grad_norm": 0.14620894193649292,
      "learning_rate": 2.36e-05,
      "loss": 0.6648,
      "step": 1056
    },
    {
      "epoch": 1.356269297084887,
      "grad_norm": 0.11804205179214478,
      "learning_rate": 2.3575e-05,
      "loss": 0.766,
      "step": 1057
    },
    {
      "epoch": 1.3575524279241349,
      "grad_norm": 0.18188659846782684,
      "learning_rate": 2.355e-05,
      "loss": 0.6309,
      "step": 1058
    },
    {
      "epoch": 1.3588355587633827,
      "grad_norm": 0.17240206897258759,
      "learning_rate": 2.3525e-05,
      "loss": 0.733,
      "step": 1059
    },
    {
      "epoch": 1.3601186896026305,
      "grad_norm": 0.11757597327232361,
      "learning_rate": 2.35e-05,
      "loss": 0.631,
      "step": 1060
    },
    {
      "epoch": 1.3614018204418783,
      "grad_norm": 0.16448509693145752,
      "learning_rate": 2.3475e-05,
      "loss": 0.6722,
      "step": 1061
    },
    {
      "epoch": 1.3626849512811259,
      "grad_norm": 0.12476751953363419,
      "learning_rate": 2.345e-05,
      "loss": 0.5502,
      "step": 1062
    },
    {
      "epoch": 1.3639680821203737,
      "grad_norm": 0.18635807931423187,
      "learning_rate": 2.3425000000000004e-05,
      "loss": 0.7195,
      "step": 1063
    },
    {
      "epoch": 1.3652512129596215,
      "grad_norm": 0.12079310417175293,
      "learning_rate": 2.3400000000000003e-05,
      "loss": 0.5864,
      "step": 1064
    },
    {
      "epoch": 1.3665343437988693,
      "grad_norm": 0.13351555168628693,
      "learning_rate": 2.3375000000000002e-05,
      "loss": 0.6352,
      "step": 1065
    },
    {
      "epoch": 1.3678174746381169,
      "grad_norm": 0.28210848569869995,
      "learning_rate": 2.3350000000000002e-05,
      "loss": 0.7817,
      "step": 1066
    },
    {
      "epoch": 1.3691006054773647,
      "grad_norm": 0.12239344418048859,
      "learning_rate": 2.3325e-05,
      "loss": 0.7295,
      "step": 1067
    },
    {
      "epoch": 1.3703837363166125,
      "grad_norm": 0.13832642138004303,
      "learning_rate": 2.3300000000000004e-05,
      "loss": 0.6513,
      "step": 1068
    },
    {
      "epoch": 1.3716668671558603,
      "grad_norm": 0.20540766417980194,
      "learning_rate": 2.3275000000000003e-05,
      "loss": 0.7003,
      "step": 1069
    },
    {
      "epoch": 1.372949997995108,
      "grad_norm": 0.14227239787578583,
      "learning_rate": 2.3250000000000003e-05,
      "loss": 0.6823,
      "step": 1070
    },
    {
      "epoch": 1.3742331288343559,
      "grad_norm": 0.11242067813873291,
      "learning_rate": 2.3225000000000002e-05,
      "loss": 0.695,
      "step": 1071
    },
    {
      "epoch": 1.3755162596736037,
      "grad_norm": 0.2395172119140625,
      "learning_rate": 2.32e-05,
      "loss": 0.7294,
      "step": 1072
    },
    {
      "epoch": 1.3767993905128515,
      "grad_norm": 0.14995355904102325,
      "learning_rate": 2.3175e-05,
      "loss": 0.6923,
      "step": 1073
    },
    {
      "epoch": 1.378082521352099,
      "grad_norm": 0.10733988881111145,
      "learning_rate": 2.3150000000000004e-05,
      "loss": 0.7498,
      "step": 1074
    },
    {
      "epoch": 1.3793656521913469,
      "grad_norm": 0.14108900725841522,
      "learning_rate": 2.3125000000000003e-05,
      "loss": 0.7162,
      "step": 1075
    },
    {
      "epoch": 1.3806487830305947,
      "grad_norm": 0.14789704978466034,
      "learning_rate": 2.3100000000000002e-05,
      "loss": 0.8312,
      "step": 1076
    },
    {
      "epoch": 1.3819319138698425,
      "grad_norm": 0.16460496187210083,
      "learning_rate": 2.3075000000000002e-05,
      "loss": 0.693,
      "step": 1077
    },
    {
      "epoch": 1.38321504470909,
      "grad_norm": 0.12011631578207016,
      "learning_rate": 2.305e-05,
      "loss": 0.6579,
      "step": 1078
    },
    {
      "epoch": 1.3844981755483379,
      "grad_norm": 0.23273155093193054,
      "learning_rate": 2.3025e-05,
      "loss": 0.6656,
      "step": 1079
    },
    {
      "epoch": 1.3857813063875857,
      "grad_norm": 0.14986133575439453,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 0.6642,
      "step": 1080
    },
    {
      "epoch": 1.3870644372268335,
      "grad_norm": 0.12668085098266602,
      "learning_rate": 2.2975000000000003e-05,
      "loss": 0.6013,
      "step": 1081
    },
    {
      "epoch": 1.3883475680660813,
      "grad_norm": 0.17946560680866241,
      "learning_rate": 2.2950000000000002e-05,
      "loss": 0.7723,
      "step": 1082
    },
    {
      "epoch": 1.389630698905329,
      "grad_norm": 0.1837294101715088,
      "learning_rate": 2.2925e-05,
      "loss": 0.6081,
      "step": 1083
    },
    {
      "epoch": 1.390913829744577,
      "grad_norm": 0.10806173831224442,
      "learning_rate": 2.29e-05,
      "loss": 0.6843,
      "step": 1084
    },
    {
      "epoch": 1.3921969605838245,
      "grad_norm": 0.12691102921962738,
      "learning_rate": 2.2875e-05,
      "loss": 0.6229,
      "step": 1085
    },
    {
      "epoch": 1.3934800914230723,
      "grad_norm": 0.2042393684387207,
      "learning_rate": 2.2850000000000003e-05,
      "loss": 0.6767,
      "step": 1086
    },
    {
      "epoch": 1.39476322226232,
      "grad_norm": 0.18985114991664886,
      "learning_rate": 2.2825000000000003e-05,
      "loss": 0.7548,
      "step": 1087
    },
    {
      "epoch": 1.3960463531015679,
      "grad_norm": 0.12734369933605194,
      "learning_rate": 2.2800000000000002e-05,
      "loss": 0.7067,
      "step": 1088
    },
    {
      "epoch": 1.3973294839408155,
      "grad_norm": 0.17573462426662445,
      "learning_rate": 2.2775e-05,
      "loss": 0.7703,
      "step": 1089
    },
    {
      "epoch": 1.3986126147800633,
      "grad_norm": 0.14725449681282043,
      "learning_rate": 2.275e-05,
      "loss": 0.6538,
      "step": 1090
    },
    {
      "epoch": 1.399895745619311,
      "grad_norm": 0.11268345266580582,
      "learning_rate": 2.2725000000000003e-05,
      "loss": 0.652,
      "step": 1091
    },
    {
      "epoch": 1.4011788764585589,
      "grad_norm": 0.1128515675663948,
      "learning_rate": 2.2700000000000003e-05,
      "loss": 0.7532,
      "step": 1092
    },
    {
      "epoch": 1.4024620072978067,
      "grad_norm": 0.19085878133773804,
      "learning_rate": 2.2675000000000002e-05,
      "loss": 0.6565,
      "step": 1093
    },
    {
      "epoch": 1.4037451381370545,
      "grad_norm": 0.1461411565542221,
      "learning_rate": 2.265e-05,
      "loss": 0.6589,
      "step": 1094
    },
    {
      "epoch": 1.4050282689763023,
      "grad_norm": 0.12743644416332245,
      "learning_rate": 2.2625e-05,
      "loss": 0.6777,
      "step": 1095
    },
    {
      "epoch": 1.4063113998155499,
      "grad_norm": 0.13085035979747772,
      "learning_rate": 2.26e-05,
      "loss": 0.7564,
      "step": 1096
    },
    {
      "epoch": 1.4075945306547977,
      "grad_norm": 0.16173502802848816,
      "learning_rate": 2.2575000000000003e-05,
      "loss": 0.7012,
      "step": 1097
    },
    {
      "epoch": 1.4088776614940455,
      "grad_norm": 0.13909080624580383,
      "learning_rate": 2.2550000000000003e-05,
      "loss": 0.7154,
      "step": 1098
    },
    {
      "epoch": 1.4101607923332933,
      "grad_norm": 0.1334976702928543,
      "learning_rate": 2.2525000000000002e-05,
      "loss": 0.6963,
      "step": 1099
    },
    {
      "epoch": 1.411443923172541,
      "grad_norm": 0.1790662705898285,
      "learning_rate": 2.25e-05,
      "loss": 0.5553,
      "step": 1100
    },
    {
      "epoch": 1.4127270540117887,
      "grad_norm": 0.1301458775997162,
      "learning_rate": 2.2475e-05,
      "loss": 0.7167,
      "step": 1101
    },
    {
      "epoch": 1.4140101848510365,
      "grad_norm": 0.14647060632705688,
      "learning_rate": 2.245e-05,
      "loss": 0.7003,
      "step": 1102
    },
    {
      "epoch": 1.4152933156902843,
      "grad_norm": 0.13639485836029053,
      "learning_rate": 2.2425000000000003e-05,
      "loss": 0.6566,
      "step": 1103
    },
    {
      "epoch": 1.416576446529532,
      "grad_norm": 0.18248987197875977,
      "learning_rate": 2.2400000000000002e-05,
      "loss": 0.7642,
      "step": 1104
    },
    {
      "epoch": 1.4178595773687799,
      "grad_norm": 0.10670022666454315,
      "learning_rate": 2.2375000000000002e-05,
      "loss": 0.551,
      "step": 1105
    },
    {
      "epoch": 1.4191427082080277,
      "grad_norm": 0.15404438972473145,
      "learning_rate": 2.235e-05,
      "loss": 0.6562,
      "step": 1106
    },
    {
      "epoch": 1.4204258390472755,
      "grad_norm": 0.12845264375209808,
      "learning_rate": 2.2325e-05,
      "loss": 0.6412,
      "step": 1107
    },
    {
      "epoch": 1.421708969886523,
      "grad_norm": 0.12882862985134125,
      "learning_rate": 2.23e-05,
      "loss": 0.6726,
      "step": 1108
    },
    {
      "epoch": 1.4229921007257709,
      "grad_norm": 0.11847760528326035,
      "learning_rate": 2.2275000000000003e-05,
      "loss": 0.7131,
      "step": 1109
    },
    {
      "epoch": 1.4242752315650187,
      "grad_norm": 0.11268723011016846,
      "learning_rate": 2.2250000000000002e-05,
      "loss": 0.693,
      "step": 1110
    },
    {
      "epoch": 1.4255583624042665,
      "grad_norm": 0.149946391582489,
      "learning_rate": 2.2225e-05,
      "loss": 0.7513,
      "step": 1111
    },
    {
      "epoch": 1.426841493243514,
      "grad_norm": 0.10811381787061691,
      "learning_rate": 2.22e-05,
      "loss": 0.6597,
      "step": 1112
    },
    {
      "epoch": 1.4281246240827619,
      "grad_norm": 0.11985135078430176,
      "learning_rate": 2.2175e-05,
      "loss": 0.663,
      "step": 1113
    },
    {
      "epoch": 1.4294077549220097,
      "grad_norm": 0.1555681824684143,
      "learning_rate": 2.215e-05,
      "loss": 0.5514,
      "step": 1114
    },
    {
      "epoch": 1.4306908857612575,
      "grad_norm": 0.13832812011241913,
      "learning_rate": 2.2125000000000002e-05,
      "loss": 0.7138,
      "step": 1115
    },
    {
      "epoch": 1.4319740166005053,
      "grad_norm": 0.09856922924518585,
      "learning_rate": 2.2100000000000002e-05,
      "loss": 0.6547,
      "step": 1116
    },
    {
      "epoch": 1.433257147439753,
      "grad_norm": 0.1875777244567871,
      "learning_rate": 2.2075e-05,
      "loss": 0.6169,
      "step": 1117
    },
    {
      "epoch": 1.4345402782790009,
      "grad_norm": 0.1743428260087967,
      "learning_rate": 2.205e-05,
      "loss": 0.613,
      "step": 1118
    },
    {
      "epoch": 1.4358234091182485,
      "grad_norm": 0.17224626243114471,
      "learning_rate": 2.2025e-05,
      "loss": 0.6856,
      "step": 1119
    },
    {
      "epoch": 1.4371065399574963,
      "grad_norm": 0.1443355828523636,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.6495,
      "step": 1120
    },
    {
      "epoch": 1.438389670796744,
      "grad_norm": 0.12037026137113571,
      "learning_rate": 2.1975000000000002e-05,
      "loss": 0.6485,
      "step": 1121
    },
    {
      "epoch": 1.4396728016359919,
      "grad_norm": 0.1384287178516388,
      "learning_rate": 2.195e-05,
      "loss": 0.6712,
      "step": 1122
    },
    {
      "epoch": 1.4409559324752395,
      "grad_norm": 0.15313078463077545,
      "learning_rate": 2.1925e-05,
      "loss": 0.5958,
      "step": 1123
    },
    {
      "epoch": 1.4422390633144873,
      "grad_norm": 0.13028162717819214,
      "learning_rate": 2.19e-05,
      "loss": 0.6476,
      "step": 1124
    },
    {
      "epoch": 1.443522194153735,
      "grad_norm": 0.20501457154750824,
      "learning_rate": 2.1875e-05,
      "loss": 0.7738,
      "step": 1125
    },
    {
      "epoch": 1.4448053249929829,
      "grad_norm": 0.1482904851436615,
      "learning_rate": 2.1850000000000003e-05,
      "loss": 0.6605,
      "step": 1126
    },
    {
      "epoch": 1.4460884558322307,
      "grad_norm": 0.15279611945152283,
      "learning_rate": 2.1825000000000002e-05,
      "loss": 0.6849,
      "step": 1127
    },
    {
      "epoch": 1.4473715866714785,
      "grad_norm": 0.14933930337429047,
      "learning_rate": 2.18e-05,
      "loss": 0.6533,
      "step": 1128
    },
    {
      "epoch": 1.4486547175107263,
      "grad_norm": 0.14123593270778656,
      "learning_rate": 2.1775e-05,
      "loss": 0.7738,
      "step": 1129
    },
    {
      "epoch": 1.4499378483499739,
      "grad_norm": 0.13760720193386078,
      "learning_rate": 2.175e-05,
      "loss": 0.7077,
      "step": 1130
    },
    {
      "epoch": 1.4512209791892217,
      "grad_norm": 0.11126912385225296,
      "learning_rate": 2.1725e-05,
      "loss": 0.6872,
      "step": 1131
    },
    {
      "epoch": 1.4525041100284695,
      "grad_norm": 0.12399459630250931,
      "learning_rate": 2.1700000000000002e-05,
      "loss": 0.705,
      "step": 1132
    },
    {
      "epoch": 1.4537872408677173,
      "grad_norm": 0.18581102788448334,
      "learning_rate": 2.1675e-05,
      "loss": 0.6387,
      "step": 1133
    },
    {
      "epoch": 1.455070371706965,
      "grad_norm": 0.11659476161003113,
      "learning_rate": 2.165e-05,
      "loss": 0.638,
      "step": 1134
    },
    {
      "epoch": 1.4563535025462127,
      "grad_norm": 0.11068329215049744,
      "learning_rate": 2.1625e-05,
      "loss": 0.5762,
      "step": 1135
    },
    {
      "epoch": 1.4576366333854605,
      "grad_norm": 0.12021217495203018,
      "learning_rate": 2.16e-05,
      "loss": 0.7739,
      "step": 1136
    },
    {
      "epoch": 1.4589197642247083,
      "grad_norm": 0.18542878329753876,
      "learning_rate": 2.1575e-05,
      "loss": 0.6716,
      "step": 1137
    },
    {
      "epoch": 1.460202895063956,
      "grad_norm": 0.1270538866519928,
      "learning_rate": 2.1550000000000002e-05,
      "loss": 0.6392,
      "step": 1138
    },
    {
      "epoch": 1.4614860259032039,
      "grad_norm": 0.14417818188667297,
      "learning_rate": 2.1525e-05,
      "loss": 0.7106,
      "step": 1139
    },
    {
      "epoch": 1.4627691567424517,
      "grad_norm": 0.19360293447971344,
      "learning_rate": 2.15e-05,
      "loss": 0.6856,
      "step": 1140
    },
    {
      "epoch": 1.4640522875816995,
      "grad_norm": 0.1428864747285843,
      "learning_rate": 2.1475e-05,
      "loss": 0.6426,
      "step": 1141
    },
    {
      "epoch": 1.465335418420947,
      "grad_norm": 0.1504928469657898,
      "learning_rate": 2.145e-05,
      "loss": 0.6252,
      "step": 1142
    },
    {
      "epoch": 1.4666185492601949,
      "grad_norm": 0.1692209541797638,
      "learning_rate": 2.1425e-05,
      "loss": 0.729,
      "step": 1143
    },
    {
      "epoch": 1.4679016800994427,
      "grad_norm": 0.13551710546016693,
      "learning_rate": 2.1400000000000002e-05,
      "loss": 0.7323,
      "step": 1144
    },
    {
      "epoch": 1.4691848109386905,
      "grad_norm": 0.11782413721084595,
      "learning_rate": 2.1375e-05,
      "loss": 0.6169,
      "step": 1145
    },
    {
      "epoch": 1.470467941777938,
      "grad_norm": 0.1493755728006363,
      "learning_rate": 2.135e-05,
      "loss": 0.6606,
      "step": 1146
    },
    {
      "epoch": 1.4717510726171859,
      "grad_norm": 0.16993238031864166,
      "learning_rate": 2.1325e-05,
      "loss": 0.7151,
      "step": 1147
    },
    {
      "epoch": 1.4730342034564337,
      "grad_norm": 0.14345870912075043,
      "learning_rate": 2.13e-05,
      "loss": 0.7187,
      "step": 1148
    },
    {
      "epoch": 1.4743173342956815,
      "grad_norm": 0.2202182114124298,
      "learning_rate": 2.1275000000000002e-05,
      "loss": 0.7627,
      "step": 1149
    },
    {
      "epoch": 1.4756004651349293,
      "grad_norm": 0.12015771865844727,
      "learning_rate": 2.125e-05,
      "loss": 0.5791,
      "step": 1150
    },
    {
      "epoch": 1.476883595974177,
      "grad_norm": 0.1657026708126068,
      "learning_rate": 2.1225e-05,
      "loss": 0.5606,
      "step": 1151
    },
    {
      "epoch": 1.4781667268134249,
      "grad_norm": 0.1524410843849182,
      "learning_rate": 2.12e-05,
      "loss": 0.7622,
      "step": 1152
    },
    {
      "epoch": 1.4794498576526725,
      "grad_norm": 0.12131082266569138,
      "learning_rate": 2.1175e-05,
      "loss": 0.6515,
      "step": 1153
    },
    {
      "epoch": 1.4807329884919203,
      "grad_norm": 0.12534214556217194,
      "learning_rate": 2.115e-05,
      "loss": 0.7124,
      "step": 1154
    },
    {
      "epoch": 1.482016119331168,
      "grad_norm": 0.12051219493150711,
      "learning_rate": 2.1125000000000002e-05,
      "loss": 0.5882,
      "step": 1155
    },
    {
      "epoch": 1.4832992501704159,
      "grad_norm": 0.16111978888511658,
      "learning_rate": 2.11e-05,
      "loss": 0.7523,
      "step": 1156
    },
    {
      "epoch": 1.4845823810096634,
      "grad_norm": 0.14418649673461914,
      "learning_rate": 2.1075e-05,
      "loss": 0.6409,
      "step": 1157
    },
    {
      "epoch": 1.4858655118489112,
      "grad_norm": 0.12204087525606155,
      "learning_rate": 2.105e-05,
      "loss": 0.8232,
      "step": 1158
    },
    {
      "epoch": 1.487148642688159,
      "grad_norm": 0.13546761870384216,
      "learning_rate": 2.1025e-05,
      "loss": 0.7376,
      "step": 1159
    },
    {
      "epoch": 1.4884317735274069,
      "grad_norm": 0.13471700251102448,
      "learning_rate": 2.1e-05,
      "loss": 0.5984,
      "step": 1160
    },
    {
      "epoch": 1.4897149043666547,
      "grad_norm": 0.1155252531170845,
      "learning_rate": 2.0975e-05,
      "loss": 0.6956,
      "step": 1161
    },
    {
      "epoch": 1.4909980352059025,
      "grad_norm": 0.11469168961048126,
      "learning_rate": 2.095e-05,
      "loss": 0.5656,
      "step": 1162
    },
    {
      "epoch": 1.4922811660451503,
      "grad_norm": 0.13567231595516205,
      "learning_rate": 2.0925e-05,
      "loss": 0.634,
      "step": 1163
    },
    {
      "epoch": 1.493564296884398,
      "grad_norm": 0.11076784878969193,
      "learning_rate": 2.09e-05,
      "loss": 0.6202,
      "step": 1164
    },
    {
      "epoch": 1.4948474277236456,
      "grad_norm": 0.10771938413381577,
      "learning_rate": 2.0875e-05,
      "loss": 0.7106,
      "step": 1165
    },
    {
      "epoch": 1.4961305585628935,
      "grad_norm": 0.12163422256708145,
      "learning_rate": 2.085e-05,
      "loss": 0.7524,
      "step": 1166
    },
    {
      "epoch": 1.4974136894021413,
      "grad_norm": 0.13888458907604218,
      "learning_rate": 2.0825e-05,
      "loss": 0.6652,
      "step": 1167
    },
    {
      "epoch": 1.498696820241389,
      "grad_norm": 0.12717197835445404,
      "learning_rate": 2.08e-05,
      "loss": 0.6545,
      "step": 1168
    },
    {
      "epoch": 1.4999799510806366,
      "grad_norm": 0.1300230622291565,
      "learning_rate": 2.0775e-05,
      "loss": 0.7003,
      "step": 1169
    },
    {
      "epoch": 1.5012630819198844,
      "grad_norm": 0.13557158410549164,
      "learning_rate": 2.075e-05,
      "loss": 0.5592,
      "step": 1170
    },
    {
      "epoch": 1.5025462127591322,
      "grad_norm": 0.10597646981477737,
      "learning_rate": 2.0725e-05,
      "loss": 0.6916,
      "step": 1171
    },
    {
      "epoch": 1.50382934359838,
      "grad_norm": 0.11596749722957611,
      "learning_rate": 2.07e-05,
      "loss": 0.7404,
      "step": 1172
    },
    {
      "epoch": 1.5051124744376279,
      "grad_norm": 0.12383753806352615,
      "learning_rate": 2.0675e-05,
      "loss": 0.7041,
      "step": 1173
    },
    {
      "epoch": 1.5063956052768757,
      "grad_norm": 0.9166615605354309,
      "learning_rate": 2.065e-05,
      "loss": 0.6614,
      "step": 1174
    },
    {
      "epoch": 1.5076787361161235,
      "grad_norm": 0.12161948531866074,
      "learning_rate": 2.0625e-05,
      "loss": 0.7178,
      "step": 1175
    },
    {
      "epoch": 1.5089618669553713,
      "grad_norm": 0.1278965324163437,
      "learning_rate": 2.06e-05,
      "loss": 0.7493,
      "step": 1176
    },
    {
      "epoch": 1.5102449977946188,
      "grad_norm": 0.19567012786865234,
      "learning_rate": 2.0575e-05,
      "loss": 0.6521,
      "step": 1177
    },
    {
      "epoch": 1.5115281286338667,
      "grad_norm": 0.13046441972255707,
      "learning_rate": 2.055e-05,
      "loss": 0.6669,
      "step": 1178
    },
    {
      "epoch": 1.5128112594731142,
      "grad_norm": 0.1338386982679367,
      "learning_rate": 2.0525e-05,
      "loss": 0.6718,
      "step": 1179
    },
    {
      "epoch": 1.514094390312362,
      "grad_norm": 0.11612948030233383,
      "learning_rate": 2.05e-05,
      "loss": 0.6271,
      "step": 1180
    },
    {
      "epoch": 1.5153775211516098,
      "grad_norm": 0.1374698430299759,
      "learning_rate": 2.0475e-05,
      "loss": 0.7032,
      "step": 1181
    },
    {
      "epoch": 1.5166606519908576,
      "grad_norm": 0.14481447637081146,
      "learning_rate": 2.045e-05,
      "loss": 0.7325,
      "step": 1182
    },
    {
      "epoch": 1.5179437828301054,
      "grad_norm": 0.12722648680210114,
      "learning_rate": 2.0425e-05,
      "loss": 0.6544,
      "step": 1183
    },
    {
      "epoch": 1.5192269136693533,
      "grad_norm": 0.13937945663928986,
      "learning_rate": 2.04e-05,
      "loss": 0.6037,
      "step": 1184
    },
    {
      "epoch": 1.520510044508601,
      "grad_norm": 0.12462259083986282,
      "learning_rate": 2.0375e-05,
      "loss": 0.6836,
      "step": 1185
    },
    {
      "epoch": 1.5217931753478489,
      "grad_norm": 0.10917101055383682,
      "learning_rate": 2.035e-05,
      "loss": 0.5627,
      "step": 1186
    },
    {
      "epoch": 1.5230763061870967,
      "grad_norm": 0.11286066472530365,
      "learning_rate": 2.0325e-05,
      "loss": 0.6355,
      "step": 1187
    },
    {
      "epoch": 1.5243594370263442,
      "grad_norm": 0.13300608098506927,
      "learning_rate": 2.0300000000000002e-05,
      "loss": 0.6745,
      "step": 1188
    },
    {
      "epoch": 1.525642567865592,
      "grad_norm": 0.14726601541042328,
      "learning_rate": 2.0275e-05,
      "loss": 0.6807,
      "step": 1189
    },
    {
      "epoch": 1.5269256987048399,
      "grad_norm": 0.13778601586818695,
      "learning_rate": 2.025e-05,
      "loss": 0.7363,
      "step": 1190
    },
    {
      "epoch": 1.5282088295440874,
      "grad_norm": 0.12509533762931824,
      "learning_rate": 2.0225000000000004e-05,
      "loss": 0.6689,
      "step": 1191
    },
    {
      "epoch": 1.5294919603833352,
      "grad_norm": 0.11750698834657669,
      "learning_rate": 2.0200000000000003e-05,
      "loss": 0.6145,
      "step": 1192
    },
    {
      "epoch": 1.530775091222583,
      "grad_norm": 0.1375279426574707,
      "learning_rate": 2.0175000000000003e-05,
      "loss": 0.6678,
      "step": 1193
    },
    {
      "epoch": 1.5320582220618308,
      "grad_norm": 0.13817015290260315,
      "learning_rate": 2.0150000000000002e-05,
      "loss": 0.6887,
      "step": 1194
    },
    {
      "epoch": 1.5333413529010786,
      "grad_norm": 0.12550437450408936,
      "learning_rate": 2.0125e-05,
      "loss": 0.612,
      "step": 1195
    },
    {
      "epoch": 1.5346244837403265,
      "grad_norm": 0.12547250092029572,
      "learning_rate": 2.01e-05,
      "loss": 0.7514,
      "step": 1196
    },
    {
      "epoch": 1.5359076145795743,
      "grad_norm": 0.11022087186574936,
      "learning_rate": 2.0075000000000003e-05,
      "loss": 0.6816,
      "step": 1197
    },
    {
      "epoch": 1.537190745418822,
      "grad_norm": 0.13550901412963867,
      "learning_rate": 2.0050000000000003e-05,
      "loss": 0.6019,
      "step": 1198
    },
    {
      "epoch": 1.5384738762580696,
      "grad_norm": 0.15220233798027039,
      "learning_rate": 2.0025000000000002e-05,
      "loss": 0.7279,
      "step": 1199
    },
    {
      "epoch": 1.5397570070973174,
      "grad_norm": 0.12883928418159485,
      "learning_rate": 2e-05,
      "loss": 0.7093,
      "step": 1200
    },
    {
      "epoch": 1.5410401379365652,
      "grad_norm": 0.11407290399074554,
      "learning_rate": 1.9975e-05,
      "loss": 0.6253,
      "step": 1201
    },
    {
      "epoch": 1.5423232687758128,
      "grad_norm": 0.1161826103925705,
      "learning_rate": 1.995e-05,
      "loss": 0.7263,
      "step": 1202
    },
    {
      "epoch": 1.5436063996150606,
      "grad_norm": 0.10845784097909927,
      "learning_rate": 1.9925000000000003e-05,
      "loss": 0.6291,
      "step": 1203
    },
    {
      "epoch": 1.5448895304543084,
      "grad_norm": 0.11213939636945724,
      "learning_rate": 1.9900000000000003e-05,
      "loss": 0.6451,
      "step": 1204
    },
    {
      "epoch": 1.5461726612935562,
      "grad_norm": 0.11751475930213928,
      "learning_rate": 1.9875000000000002e-05,
      "loss": 0.7288,
      "step": 1205
    },
    {
      "epoch": 1.547455792132804,
      "grad_norm": 0.17337869107723236,
      "learning_rate": 1.985e-05,
      "loss": 0.7215,
      "step": 1206
    },
    {
      "epoch": 1.5487389229720518,
      "grad_norm": 0.12952551245689392,
      "learning_rate": 1.9825e-05,
      "loss": 0.7143,
      "step": 1207
    },
    {
      "epoch": 1.5500220538112996,
      "grad_norm": 0.21707835793495178,
      "learning_rate": 1.9800000000000004e-05,
      "loss": 0.7133,
      "step": 1208
    },
    {
      "epoch": 1.5513051846505475,
      "grad_norm": 0.11903823167085648,
      "learning_rate": 1.9775000000000003e-05,
      "loss": 0.6251,
      "step": 1209
    },
    {
      "epoch": 1.5525883154897953,
      "grad_norm": 0.15934951603412628,
      "learning_rate": 1.9750000000000002e-05,
      "loss": 0.6351,
      "step": 1210
    },
    {
      "epoch": 1.5538714463290428,
      "grad_norm": 0.23085235059261322,
      "learning_rate": 1.9725000000000002e-05,
      "loss": 0.6571,
      "step": 1211
    },
    {
      "epoch": 1.5551545771682906,
      "grad_norm": 0.11457837373018265,
      "learning_rate": 1.97e-05,
      "loss": 0.6744,
      "step": 1212
    },
    {
      "epoch": 1.5564377080075384,
      "grad_norm": 0.24926653504371643,
      "learning_rate": 1.9675e-05,
      "loss": 0.6322,
      "step": 1213
    },
    {
      "epoch": 1.557720838846786,
      "grad_norm": 0.14555948972702026,
      "learning_rate": 1.9650000000000003e-05,
      "loss": 0.6086,
      "step": 1214
    },
    {
      "epoch": 1.5590039696860338,
      "grad_norm": 0.1458958387374878,
      "learning_rate": 1.9625000000000003e-05,
      "loss": 0.6752,
      "step": 1215
    },
    {
      "epoch": 1.5602871005252816,
      "grad_norm": 0.2551647424697876,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 0.5892,
      "step": 1216
    },
    {
      "epoch": 1.5615702313645294,
      "grad_norm": 0.12587259709835052,
      "learning_rate": 1.9575e-05,
      "loss": 0.7124,
      "step": 1217
    },
    {
      "epoch": 1.5628533622037772,
      "grad_norm": 0.12688258290290833,
      "learning_rate": 1.955e-05,
      "loss": 0.7588,
      "step": 1218
    },
    {
      "epoch": 1.564136493043025,
      "grad_norm": 0.1534375250339508,
      "learning_rate": 1.9525e-05,
      "loss": 0.7045,
      "step": 1219
    },
    {
      "epoch": 1.5654196238822728,
      "grad_norm": 0.1590505689382553,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 0.7329,
      "step": 1220
    },
    {
      "epoch": 1.5667027547215207,
      "grad_norm": 0.12459621578454971,
      "learning_rate": 1.9475000000000002e-05,
      "loss": 0.7586,
      "step": 1221
    },
    {
      "epoch": 1.5679858855607682,
      "grad_norm": 0.14487729966640472,
      "learning_rate": 1.9450000000000002e-05,
      "loss": 0.7036,
      "step": 1222
    },
    {
      "epoch": 1.569269016400016,
      "grad_norm": 0.12553448975086212,
      "learning_rate": 1.9425e-05,
      "loss": 0.6523,
      "step": 1223
    },
    {
      "epoch": 1.5705521472392638,
      "grad_norm": 0.12217416614294052,
      "learning_rate": 1.94e-05,
      "loss": 0.6548,
      "step": 1224
    },
    {
      "epoch": 1.5718352780785114,
      "grad_norm": 0.1370861679315567,
      "learning_rate": 1.9375e-05,
      "loss": 0.6693,
      "step": 1225
    },
    {
      "epoch": 1.5731184089177592,
      "grad_norm": 0.12897707521915436,
      "learning_rate": 1.9350000000000003e-05,
      "loss": 0.7183,
      "step": 1226
    },
    {
      "epoch": 1.574401539757007,
      "grad_norm": 0.12946870923042297,
      "learning_rate": 1.9325000000000002e-05,
      "loss": 0.6775,
      "step": 1227
    },
    {
      "epoch": 1.5756846705962548,
      "grad_norm": 0.11160869896411896,
      "learning_rate": 1.93e-05,
      "loss": 0.6462,
      "step": 1228
    },
    {
      "epoch": 1.5769678014355026,
      "grad_norm": 0.19780348241329193,
      "learning_rate": 1.9275e-05,
      "loss": 0.6921,
      "step": 1229
    },
    {
      "epoch": 1.5782509322747504,
      "grad_norm": 0.11126334965229034,
      "learning_rate": 1.925e-05,
      "loss": 0.641,
      "step": 1230
    },
    {
      "epoch": 1.5795340631139982,
      "grad_norm": 0.17710551619529724,
      "learning_rate": 1.9225e-05,
      "loss": 0.7526,
      "step": 1231
    },
    {
      "epoch": 1.580817193953246,
      "grad_norm": 0.1221880167722702,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.6032,
      "step": 1232
    },
    {
      "epoch": 1.5821003247924939,
      "grad_norm": 0.130191370844841,
      "learning_rate": 1.9175000000000002e-05,
      "loss": 0.6932,
      "step": 1233
    },
    {
      "epoch": 1.5833834556317414,
      "grad_norm": 0.1171320453286171,
      "learning_rate": 1.915e-05,
      "loss": 0.7188,
      "step": 1234
    },
    {
      "epoch": 1.5846665864709892,
      "grad_norm": 0.12760233879089355,
      "learning_rate": 1.9125e-05,
      "loss": 0.7081,
      "step": 1235
    },
    {
      "epoch": 1.5859497173102368,
      "grad_norm": 0.1895088106393814,
      "learning_rate": 1.91e-05,
      "loss": 0.6376,
      "step": 1236
    },
    {
      "epoch": 1.5872328481494846,
      "grad_norm": 0.15296760201454163,
      "learning_rate": 1.9075000000000003e-05,
      "loss": 0.8098,
      "step": 1237
    },
    {
      "epoch": 1.5885159789887324,
      "grad_norm": 0.16338662803173065,
      "learning_rate": 1.9050000000000002e-05,
      "loss": 0.6793,
      "step": 1238
    },
    {
      "epoch": 1.5897991098279802,
      "grad_norm": 0.13704895973205566,
      "learning_rate": 1.9025e-05,
      "loss": 0.5983,
      "step": 1239
    },
    {
      "epoch": 1.591082240667228,
      "grad_norm": 0.14559198915958405,
      "learning_rate": 1.9e-05,
      "loss": 0.6121,
      "step": 1240
    },
    {
      "epoch": 1.5923653715064758,
      "grad_norm": 0.6883216500282288,
      "learning_rate": 1.8975e-05,
      "loss": 0.6797,
      "step": 1241
    },
    {
      "epoch": 1.5936485023457236,
      "grad_norm": 0.13931411504745483,
      "learning_rate": 1.895e-05,
      "loss": 0.5479,
      "step": 1242
    },
    {
      "epoch": 1.5949316331849714,
      "grad_norm": 0.14977627992630005,
      "learning_rate": 1.8925000000000003e-05,
      "loss": 0.6748,
      "step": 1243
    },
    {
      "epoch": 1.5962147640242192,
      "grad_norm": 0.1692972481250763,
      "learning_rate": 1.8900000000000002e-05,
      "loss": 0.6613,
      "step": 1244
    },
    {
      "epoch": 1.5974978948634668,
      "grad_norm": 0.11552184075117111,
      "learning_rate": 1.8875e-05,
      "loss": 0.5997,
      "step": 1245
    },
    {
      "epoch": 1.5987810257027146,
      "grad_norm": 0.12523971498012543,
      "learning_rate": 1.885e-05,
      "loss": 0.7111,
      "step": 1246
    },
    {
      "epoch": 1.6000641565419624,
      "grad_norm": 0.2136063575744629,
      "learning_rate": 1.8825e-05,
      "loss": 0.5912,
      "step": 1247
    },
    {
      "epoch": 1.60134728738121,
      "grad_norm": 0.16620782017707825,
      "learning_rate": 1.88e-05,
      "loss": 0.657,
      "step": 1248
    },
    {
      "epoch": 1.6026304182204578,
      "grad_norm": 0.15379056334495544,
      "learning_rate": 1.8775000000000002e-05,
      "loss": 0.6811,
      "step": 1249
    },
    {
      "epoch": 1.6039135490597056,
      "grad_norm": 0.14723680913448334,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.7559,
      "step": 1250
    },
    {
      "epoch": 1.6051966798989534,
      "grad_norm": 0.17427508533000946,
      "learning_rate": 1.8725e-05,
      "loss": 0.667,
      "step": 1251
    },
    {
      "epoch": 1.6064798107382012,
      "grad_norm": 0.13866527378559113,
      "learning_rate": 1.87e-05,
      "loss": 0.6543,
      "step": 1252
    },
    {
      "epoch": 1.607762941577449,
      "grad_norm": 0.18355853855609894,
      "learning_rate": 1.8675e-05,
      "loss": 0.7313,
      "step": 1253
    },
    {
      "epoch": 1.6090460724166968,
      "grad_norm": 0.12581904232501984,
      "learning_rate": 1.865e-05,
      "loss": 0.6867,
      "step": 1254
    },
    {
      "epoch": 1.6103292032559446,
      "grad_norm": 0.11011559516191483,
      "learning_rate": 1.8625000000000002e-05,
      "loss": 0.6411,
      "step": 1255
    },
    {
      "epoch": 1.6116123340951922,
      "grad_norm": 0.13260339200496674,
      "learning_rate": 1.86e-05,
      "loss": 0.7494,
      "step": 1256
    },
    {
      "epoch": 1.61289546493444,
      "grad_norm": 0.1335648149251938,
      "learning_rate": 1.8575e-05,
      "loss": 0.6364,
      "step": 1257
    },
    {
      "epoch": 1.6141785957736878,
      "grad_norm": 0.2350221872329712,
      "learning_rate": 1.855e-05,
      "loss": 0.7443,
      "step": 1258
    },
    {
      "epoch": 1.6154617266129354,
      "grad_norm": 0.14860591292381287,
      "learning_rate": 1.8525e-05,
      "loss": 0.6493,
      "step": 1259
    },
    {
      "epoch": 1.6167448574521832,
      "grad_norm": 0.1255294680595398,
      "learning_rate": 1.85e-05,
      "loss": 0.7485,
      "step": 1260
    },
    {
      "epoch": 1.618027988291431,
      "grad_norm": 0.15059664845466614,
      "learning_rate": 1.8475000000000002e-05,
      "loss": 0.621,
      "step": 1261
    },
    {
      "epoch": 1.6193111191306788,
      "grad_norm": 0.161246657371521,
      "learning_rate": 1.845e-05,
      "loss": 0.5558,
      "step": 1262
    },
    {
      "epoch": 1.6205942499699266,
      "grad_norm": 0.2036411464214325,
      "learning_rate": 1.8425e-05,
      "loss": 0.6047,
      "step": 1263
    },
    {
      "epoch": 1.6218773808091744,
      "grad_norm": 0.18932987749576569,
      "learning_rate": 1.84e-05,
      "loss": 0.7533,
      "step": 1264
    },
    {
      "epoch": 1.6231605116484222,
      "grad_norm": 0.14403486251831055,
      "learning_rate": 1.8375e-05,
      "loss": 0.7981,
      "step": 1265
    },
    {
      "epoch": 1.62444364248767,
      "grad_norm": 0.22595533728599548,
      "learning_rate": 1.8350000000000002e-05,
      "loss": 0.6184,
      "step": 1266
    },
    {
      "epoch": 1.6257267733269178,
      "grad_norm": 0.14521941542625427,
      "learning_rate": 1.8325e-05,
      "loss": 0.6674,
      "step": 1267
    },
    {
      "epoch": 1.6270099041661654,
      "grad_norm": 0.20160236954689026,
      "learning_rate": 1.83e-05,
      "loss": 0.7071,
      "step": 1268
    },
    {
      "epoch": 1.6282930350054132,
      "grad_norm": 0.1447755992412567,
      "learning_rate": 1.8275e-05,
      "loss": 0.7153,
      "step": 1269
    },
    {
      "epoch": 1.6295761658446608,
      "grad_norm": 0.223125159740448,
      "learning_rate": 1.825e-05,
      "loss": 0.6336,
      "step": 1270
    },
    {
      "epoch": 1.6308592966839086,
      "grad_norm": 0.16308586299419403,
      "learning_rate": 1.8225e-05,
      "loss": 0.7544,
      "step": 1271
    },
    {
      "epoch": 1.6321424275231564,
      "grad_norm": 0.10494381189346313,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 0.728,
      "step": 1272
    },
    {
      "epoch": 1.6334255583624042,
      "grad_norm": 0.15711893141269684,
      "learning_rate": 1.8175e-05,
      "loss": 0.7198,
      "step": 1273
    },
    {
      "epoch": 1.634708689201652,
      "grad_norm": 0.2158462256193161,
      "learning_rate": 1.815e-05,
      "loss": 0.7071,
      "step": 1274
    },
    {
      "epoch": 1.6359918200408998,
      "grad_norm": 0.13445796072483063,
      "learning_rate": 1.8125e-05,
      "loss": 0.6775,
      "step": 1275
    },
    {
      "epoch": 1.6372749508801476,
      "grad_norm": 0.18404191732406616,
      "learning_rate": 1.81e-05,
      "loss": 0.6938,
      "step": 1276
    },
    {
      "epoch": 1.6385580817193954,
      "grad_norm": 0.15086588263511658,
      "learning_rate": 1.8075e-05,
      "loss": 0.657,
      "step": 1277
    },
    {
      "epoch": 1.6398412125586432,
      "grad_norm": 0.1281137317419052,
      "learning_rate": 1.805e-05,
      "loss": 0.6566,
      "step": 1278
    },
    {
      "epoch": 1.6411243433978908,
      "grad_norm": 0.16500942409038544,
      "learning_rate": 1.8025e-05,
      "loss": 0.6417,
      "step": 1279
    },
    {
      "epoch": 1.6424074742371386,
      "grad_norm": 0.18856273591518402,
      "learning_rate": 1.8e-05,
      "loss": 0.5978,
      "step": 1280
    },
    {
      "epoch": 1.6436906050763864,
      "grad_norm": 0.10975392162799835,
      "learning_rate": 1.7975e-05,
      "loss": 0.6529,
      "step": 1281
    },
    {
      "epoch": 1.644973735915634,
      "grad_norm": 0.17665152251720428,
      "learning_rate": 1.795e-05,
      "loss": 0.6754,
      "step": 1282
    },
    {
      "epoch": 1.6462568667548818,
      "grad_norm": 0.12219380587339401,
      "learning_rate": 1.7925e-05,
      "loss": 0.8276,
      "step": 1283
    },
    {
      "epoch": 1.6475399975941296,
      "grad_norm": 0.12733003497123718,
      "learning_rate": 1.79e-05,
      "loss": 0.7495,
      "step": 1284
    },
    {
      "epoch": 1.6488231284333774,
      "grad_norm": 0.11480991542339325,
      "learning_rate": 1.7875e-05,
      "loss": 0.6771,
      "step": 1285
    },
    {
      "epoch": 1.6501062592726252,
      "grad_norm": 0.15071417391300201,
      "learning_rate": 1.785e-05,
      "loss": 0.6531,
      "step": 1286
    },
    {
      "epoch": 1.651389390111873,
      "grad_norm": 0.146849125623703,
      "learning_rate": 1.7825e-05,
      "loss": 0.5849,
      "step": 1287
    },
    {
      "epoch": 1.6526725209511208,
      "grad_norm": 0.15811583399772644,
      "learning_rate": 1.78e-05,
      "loss": 0.8328,
      "step": 1288
    },
    {
      "epoch": 1.6539556517903686,
      "grad_norm": 0.21095775067806244,
      "learning_rate": 1.7775e-05,
      "loss": 0.6173,
      "step": 1289
    },
    {
      "epoch": 1.6552387826296162,
      "grad_norm": 0.11063028872013092,
      "learning_rate": 1.775e-05,
      "loss": 0.5906,
      "step": 1290
    },
    {
      "epoch": 1.656521913468864,
      "grad_norm": 0.16982293128967285,
      "learning_rate": 1.7725e-05,
      "loss": 0.6609,
      "step": 1291
    },
    {
      "epoch": 1.6578050443081118,
      "grad_norm": 0.11614705622196198,
      "learning_rate": 1.77e-05,
      "loss": 0.5513,
      "step": 1292
    },
    {
      "epoch": 1.6590881751473594,
      "grad_norm": 0.13795951008796692,
      "learning_rate": 1.7675e-05,
      "loss": 0.6455,
      "step": 1293
    },
    {
      "epoch": 1.6603713059866072,
      "grad_norm": 0.14651727676391602,
      "learning_rate": 1.765e-05,
      "loss": 0.656,
      "step": 1294
    },
    {
      "epoch": 1.661654436825855,
      "grad_norm": 0.1461266279220581,
      "learning_rate": 1.7625e-05,
      "loss": 0.5759,
      "step": 1295
    },
    {
      "epoch": 1.6629375676651028,
      "grad_norm": 0.1302463859319687,
      "learning_rate": 1.76e-05,
      "loss": 0.7157,
      "step": 1296
    },
    {
      "epoch": 1.6642206985043506,
      "grad_norm": 0.2130376100540161,
      "learning_rate": 1.7575e-05,
      "loss": 0.6436,
      "step": 1297
    },
    {
      "epoch": 1.6655038293435984,
      "grad_norm": 0.1362278312444687,
      "learning_rate": 1.755e-05,
      "loss": 0.7305,
      "step": 1298
    },
    {
      "epoch": 1.6667869601828462,
      "grad_norm": 0.12812939286231995,
      "learning_rate": 1.7525e-05,
      "loss": 0.8903,
      "step": 1299
    },
    {
      "epoch": 1.668070091022094,
      "grad_norm": 0.13095980882644653,
      "learning_rate": 1.75e-05,
      "loss": 0.7565,
      "step": 1300
    },
    {
      "epoch": 1.6693532218613418,
      "grad_norm": 0.1730969399213791,
      "learning_rate": 1.7475e-05,
      "loss": 0.6385,
      "step": 1301
    },
    {
      "epoch": 1.6706363527005894,
      "grad_norm": 0.14023330807685852,
      "learning_rate": 1.745e-05,
      "loss": 0.751,
      "step": 1302
    },
    {
      "epoch": 1.6719194835398372,
      "grad_norm": 0.21199192106723785,
      "learning_rate": 1.7425e-05,
      "loss": 0.71,
      "step": 1303
    },
    {
      "epoch": 1.673202614379085,
      "grad_norm": 0.151366725564003,
      "learning_rate": 1.74e-05,
      "loss": 0.5774,
      "step": 1304
    },
    {
      "epoch": 1.6744857452183326,
      "grad_norm": 0.13684917986392975,
      "learning_rate": 1.7375e-05,
      "loss": 0.7425,
      "step": 1305
    },
    {
      "epoch": 1.6757688760575804,
      "grad_norm": 0.2789534032344818,
      "learning_rate": 1.7349999999999998e-05,
      "loss": 0.6169,
      "step": 1306
    },
    {
      "epoch": 1.6770520068968282,
      "grad_norm": 0.19137178361415863,
      "learning_rate": 1.7325e-05,
      "loss": 0.7564,
      "step": 1307
    },
    {
      "epoch": 1.678335137736076,
      "grad_norm": 0.12667708098888397,
      "learning_rate": 1.73e-05,
      "loss": 0.7472,
      "step": 1308
    },
    {
      "epoch": 1.6796182685753238,
      "grad_norm": 0.1945316642522812,
      "learning_rate": 1.7275e-05,
      "loss": 0.81,
      "step": 1309
    },
    {
      "epoch": 1.6809013994145716,
      "grad_norm": 0.19703683257102966,
      "learning_rate": 1.725e-05,
      "loss": 0.7421,
      "step": 1310
    },
    {
      "epoch": 1.6821845302538194,
      "grad_norm": 0.132840096950531,
      "learning_rate": 1.7225e-05,
      "loss": 0.7335,
      "step": 1311
    },
    {
      "epoch": 1.6834676610930672,
      "grad_norm": 0.2428734302520752,
      "learning_rate": 1.7199999999999998e-05,
      "loss": 0.6064,
      "step": 1312
    },
    {
      "epoch": 1.6847507919323148,
      "grad_norm": 0.2817400097846985,
      "learning_rate": 1.7175e-05,
      "loss": 0.6294,
      "step": 1313
    },
    {
      "epoch": 1.6860339227715626,
      "grad_norm": 0.15893571078777313,
      "learning_rate": 1.7150000000000004e-05,
      "loss": 0.7349,
      "step": 1314
    },
    {
      "epoch": 1.6873170536108104,
      "grad_norm": 0.2220536172389984,
      "learning_rate": 1.7125000000000003e-05,
      "loss": 0.6543,
      "step": 1315
    },
    {
      "epoch": 1.688600184450058,
      "grad_norm": 0.20911221206188202,
      "learning_rate": 1.7100000000000002e-05,
      "loss": 0.638,
      "step": 1316
    },
    {
      "epoch": 1.6898833152893058,
      "grad_norm": 0.1306636929512024,
      "learning_rate": 1.7075e-05,
      "loss": 0.7431,
      "step": 1317
    },
    {
      "epoch": 1.6911664461285536,
      "grad_norm": 0.3337976932525635,
      "learning_rate": 1.705e-05,
      "loss": 0.6701,
      "step": 1318
    },
    {
      "epoch": 1.6924495769678014,
      "grad_norm": 0.18874673545360565,
      "learning_rate": 1.7025e-05,
      "loss": 0.6936,
      "step": 1319
    },
    {
      "epoch": 1.6937327078070492,
      "grad_norm": 0.13233716785907745,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 0.7227,
      "step": 1320
    },
    {
      "epoch": 1.695015838646297,
      "grad_norm": 0.17669826745986938,
      "learning_rate": 1.6975000000000003e-05,
      "loss": 0.656,
      "step": 1321
    },
    {
      "epoch": 1.6962989694855448,
      "grad_norm": 0.20482803881168365,
      "learning_rate": 1.6950000000000002e-05,
      "loss": 0.6898,
      "step": 1322
    },
    {
      "epoch": 1.6975821003247926,
      "grad_norm": 0.13892102241516113,
      "learning_rate": 1.6925e-05,
      "loss": 0.6507,
      "step": 1323
    },
    {
      "epoch": 1.6988652311640404,
      "grad_norm": 0.14034418761730194,
      "learning_rate": 1.69e-05,
      "loss": 0.7016,
      "step": 1324
    },
    {
      "epoch": 1.700148362003288,
      "grad_norm": 0.1607329398393631,
      "learning_rate": 1.6875000000000004e-05,
      "loss": 0.6529,
      "step": 1325
    },
    {
      "epoch": 1.7014314928425358,
      "grad_norm": 0.1523558646440506,
      "learning_rate": 1.6850000000000003e-05,
      "loss": 0.6791,
      "step": 1326
    },
    {
      "epoch": 1.7027146236817834,
      "grad_norm": 0.11956484615802765,
      "learning_rate": 1.6825000000000002e-05,
      "loss": 0.8127,
      "step": 1327
    },
    {
      "epoch": 1.7039977545210312,
      "grad_norm": 0.13559770584106445,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.7115,
      "step": 1328
    },
    {
      "epoch": 1.705280885360279,
      "grad_norm": 0.16262871026992798,
      "learning_rate": 1.6775e-05,
      "loss": 0.5616,
      "step": 1329
    },
    {
      "epoch": 1.7065640161995268,
      "grad_norm": 0.12426955997943878,
      "learning_rate": 1.675e-05,
      "loss": 0.6668,
      "step": 1330
    },
    {
      "epoch": 1.7078471470387746,
      "grad_norm": 0.1751808375120163,
      "learning_rate": 1.6725000000000003e-05,
      "loss": 0.6553,
      "step": 1331
    },
    {
      "epoch": 1.7091302778780224,
      "grad_norm": 0.6798585057258606,
      "learning_rate": 1.6700000000000003e-05,
      "loss": 0.6068,
      "step": 1332
    },
    {
      "epoch": 1.7104134087172702,
      "grad_norm": 0.12011666595935822,
      "learning_rate": 1.6675000000000002e-05,
      "loss": 0.6351,
      "step": 1333
    },
    {
      "epoch": 1.711696539556518,
      "grad_norm": 0.11038540303707123,
      "learning_rate": 1.665e-05,
      "loss": 0.6627,
      "step": 1334
    },
    {
      "epoch": 1.7129796703957658,
      "grad_norm": 0.1322665512561798,
      "learning_rate": 1.6625e-05,
      "loss": 0.7068,
      "step": 1335
    },
    {
      "epoch": 1.7142628012350134,
      "grad_norm": 0.1126861572265625,
      "learning_rate": 1.66e-05,
      "loss": 0.6081,
      "step": 1336
    },
    {
      "epoch": 1.7155459320742612,
      "grad_norm": 0.11329224705696106,
      "learning_rate": 1.6575000000000003e-05,
      "loss": 0.7142,
      "step": 1337
    },
    {
      "epoch": 1.716829062913509,
      "grad_norm": 0.13173171877861023,
      "learning_rate": 1.6550000000000002e-05,
      "loss": 0.6287,
      "step": 1338
    },
    {
      "epoch": 1.7181121937527566,
      "grad_norm": 0.13154132664203644,
      "learning_rate": 1.6525000000000002e-05,
      "loss": 0.7233,
      "step": 1339
    },
    {
      "epoch": 1.7193953245920044,
      "grad_norm": 0.14961585402488708,
      "learning_rate": 1.65e-05,
      "loss": 0.7504,
      "step": 1340
    },
    {
      "epoch": 1.7206784554312522,
      "grad_norm": 0.13134168088436127,
      "learning_rate": 1.6475e-05,
      "loss": 0.6917,
      "step": 1341
    },
    {
      "epoch": 1.7219615862705,
      "grad_norm": 0.1265396773815155,
      "learning_rate": 1.645e-05,
      "loss": 0.5979,
      "step": 1342
    },
    {
      "epoch": 1.7232447171097478,
      "grad_norm": 0.14768539369106293,
      "learning_rate": 1.6425000000000003e-05,
      "loss": 0.6626,
      "step": 1343
    },
    {
      "epoch": 1.7245278479489956,
      "grad_norm": 0.13000629842281342,
      "learning_rate": 1.6400000000000002e-05,
      "loss": 0.8261,
      "step": 1344
    },
    {
      "epoch": 1.7258109787882434,
      "grad_norm": 0.15658777952194214,
      "learning_rate": 1.6375e-05,
      "loss": 0.6346,
      "step": 1345
    },
    {
      "epoch": 1.7270941096274912,
      "grad_norm": 0.15850672125816345,
      "learning_rate": 1.635e-05,
      "loss": 0.6221,
      "step": 1346
    },
    {
      "epoch": 1.7283772404667388,
      "grad_norm": 0.13517631590366364,
      "learning_rate": 1.6325e-05,
      "loss": 0.7203,
      "step": 1347
    },
    {
      "epoch": 1.7296603713059866,
      "grad_norm": 0.10801047086715698,
      "learning_rate": 1.63e-05,
      "loss": 0.7091,
      "step": 1348
    },
    {
      "epoch": 1.7309435021452344,
      "grad_norm": 0.12752363085746765,
      "learning_rate": 1.6275000000000003e-05,
      "loss": 0.7071,
      "step": 1349
    },
    {
      "epoch": 1.732226632984482,
      "grad_norm": 0.13949206471443176,
      "learning_rate": 1.6250000000000002e-05,
      "loss": 0.677,
      "step": 1350
    },
    {
      "epoch": 1.7335097638237298,
      "grad_norm": 0.177191361784935,
      "learning_rate": 1.6225e-05,
      "loss": 0.6613,
      "step": 1351
    },
    {
      "epoch": 1.7347928946629776,
      "grad_norm": 0.1404072642326355,
      "learning_rate": 1.62e-05,
      "loss": 0.6074,
      "step": 1352
    },
    {
      "epoch": 1.7360760255022254,
      "grad_norm": 0.14838001132011414,
      "learning_rate": 1.6175e-05,
      "loss": 0.7742,
      "step": 1353
    },
    {
      "epoch": 1.7373591563414732,
      "grad_norm": 0.12018391489982605,
      "learning_rate": 1.6150000000000003e-05,
      "loss": 0.7156,
      "step": 1354
    },
    {
      "epoch": 1.738642287180721,
      "grad_norm": 0.16992993652820587,
      "learning_rate": 1.6125000000000002e-05,
      "loss": 0.6307,
      "step": 1355
    },
    {
      "epoch": 1.7399254180199688,
      "grad_norm": 0.1509513258934021,
      "learning_rate": 1.6100000000000002e-05,
      "loss": 0.624,
      "step": 1356
    },
    {
      "epoch": 1.7412085488592166,
      "grad_norm": 0.14725223183631897,
      "learning_rate": 1.6075e-05,
      "loss": 0.6656,
      "step": 1357
    },
    {
      "epoch": 1.7424916796984644,
      "grad_norm": 0.1576245129108429,
      "learning_rate": 1.605e-05,
      "loss": 0.7102,
      "step": 1358
    },
    {
      "epoch": 1.743774810537712,
      "grad_norm": 0.1702420562505722,
      "learning_rate": 1.6025e-05,
      "loss": 0.6527,
      "step": 1359
    },
    {
      "epoch": 1.7450579413769598,
      "grad_norm": 0.1570025086402893,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.6267,
      "step": 1360
    },
    {
      "epoch": 1.7463410722162076,
      "grad_norm": 0.1654820442199707,
      "learning_rate": 1.5975000000000002e-05,
      "loss": 0.69,
      "step": 1361
    },
    {
      "epoch": 1.7476242030554552,
      "grad_norm": 0.13784408569335938,
      "learning_rate": 1.595e-05,
      "loss": 0.6937,
      "step": 1362
    },
    {
      "epoch": 1.748907333894703,
      "grad_norm": 0.1685132384300232,
      "learning_rate": 1.5925e-05,
      "loss": 0.586,
      "step": 1363
    },
    {
      "epoch": 1.7501904647339508,
      "grad_norm": 0.13721473515033722,
      "learning_rate": 1.59e-05,
      "loss": 0.6815,
      "step": 1364
    },
    {
      "epoch": 1.7514735955731986,
      "grad_norm": 0.11430935561656952,
      "learning_rate": 1.5875e-05,
      "loss": 0.6336,
      "step": 1365
    },
    {
      "epoch": 1.7527567264124464,
      "grad_norm": 0.12601076066493988,
      "learning_rate": 1.5850000000000002e-05,
      "loss": 0.7915,
      "step": 1366
    },
    {
      "epoch": 1.7540398572516942,
      "grad_norm": 0.15483878552913666,
      "learning_rate": 1.5825000000000002e-05,
      "loss": 0.6418,
      "step": 1367
    },
    {
      "epoch": 1.755322988090942,
      "grad_norm": 0.24551011621952057,
      "learning_rate": 1.58e-05,
      "loss": 0.7264,
      "step": 1368
    },
    {
      "epoch": 1.7566061189301898,
      "grad_norm": 0.12263841181993484,
      "learning_rate": 1.5775e-05,
      "loss": 0.6412,
      "step": 1369
    },
    {
      "epoch": 1.7578892497694374,
      "grad_norm": 0.11007744073867798,
      "learning_rate": 1.575e-05,
      "loss": 0.6667,
      "step": 1370
    },
    {
      "epoch": 1.7591723806086852,
      "grad_norm": 0.1387297511100769,
      "learning_rate": 1.5725e-05,
      "loss": 0.6307,
      "step": 1371
    },
    {
      "epoch": 1.760455511447933,
      "grad_norm": 0.18302080035209656,
      "learning_rate": 1.5700000000000002e-05,
      "loss": 0.6157,
      "step": 1372
    },
    {
      "epoch": 1.7617386422871806,
      "grad_norm": 0.12509283423423767,
      "learning_rate": 1.5675e-05,
      "loss": 0.7221,
      "step": 1373
    },
    {
      "epoch": 1.7630217731264284,
      "grad_norm": 0.23319111764431,
      "learning_rate": 1.565e-05,
      "loss": 0.6915,
      "step": 1374
    },
    {
      "epoch": 1.7643049039656762,
      "grad_norm": 0.11508772522211075,
      "learning_rate": 1.5625e-05,
      "loss": 0.7471,
      "step": 1375
    },
    {
      "epoch": 1.765588034804924,
      "grad_norm": 0.1235569417476654,
      "learning_rate": 1.56e-05,
      "loss": 0.6777,
      "step": 1376
    },
    {
      "epoch": 1.7668711656441718,
      "grad_norm": 0.1645171195268631,
      "learning_rate": 1.5575e-05,
      "loss": 0.654,
      "step": 1377
    },
    {
      "epoch": 1.7681542964834196,
      "grad_norm": 0.12215454131364822,
      "learning_rate": 1.5550000000000002e-05,
      "loss": 0.7353,
      "step": 1378
    },
    {
      "epoch": 1.7694374273226674,
      "grad_norm": 0.12350727617740631,
      "learning_rate": 1.5525e-05,
      "loss": 0.6734,
      "step": 1379
    },
    {
      "epoch": 1.7707205581619152,
      "grad_norm": 0.16377438604831696,
      "learning_rate": 1.55e-05,
      "loss": 0.6452,
      "step": 1380
    },
    {
      "epoch": 1.7720036890011628,
      "grad_norm": 0.17555522918701172,
      "learning_rate": 1.5475e-05,
      "loss": 0.6326,
      "step": 1381
    },
    {
      "epoch": 1.7732868198404106,
      "grad_norm": 0.20337384939193726,
      "learning_rate": 1.545e-05,
      "loss": 0.6757,
      "step": 1382
    },
    {
      "epoch": 1.7745699506796584,
      "grad_norm": 0.13655917346477509,
      "learning_rate": 1.5425000000000002e-05,
      "loss": 0.7743,
      "step": 1383
    },
    {
      "epoch": 1.775853081518906,
      "grad_norm": 0.12117540836334229,
      "learning_rate": 1.54e-05,
      "loss": 0.7137,
      "step": 1384
    },
    {
      "epoch": 1.7771362123581538,
      "grad_norm": 0.13435031473636627,
      "learning_rate": 1.5375e-05,
      "loss": 0.6145,
      "step": 1385
    },
    {
      "epoch": 1.7784193431974016,
      "grad_norm": 0.14893195033073425,
      "learning_rate": 1.535e-05,
      "loss": 0.6051,
      "step": 1386
    },
    {
      "epoch": 1.7797024740366494,
      "grad_norm": 0.22929029166698456,
      "learning_rate": 1.5325e-05,
      "loss": 0.7562,
      "step": 1387
    },
    {
      "epoch": 1.7809856048758972,
      "grad_norm": 0.13588419556617737,
      "learning_rate": 1.53e-05,
      "loss": 0.69,
      "step": 1388
    },
    {
      "epoch": 1.782268735715145,
      "grad_norm": 0.18602107465267181,
      "learning_rate": 1.5275000000000002e-05,
      "loss": 0.7215,
      "step": 1389
    },
    {
      "epoch": 1.7835518665543928,
      "grad_norm": 0.18229863047599792,
      "learning_rate": 1.525e-05,
      "loss": 0.6812,
      "step": 1390
    },
    {
      "epoch": 1.7848349973936406,
      "grad_norm": 0.11314523965120316,
      "learning_rate": 1.5225e-05,
      "loss": 0.6085,
      "step": 1391
    },
    {
      "epoch": 1.7861181282328884,
      "grad_norm": 0.19920094311237335,
      "learning_rate": 1.52e-05,
      "loss": 0.7272,
      "step": 1392
    },
    {
      "epoch": 1.787401259072136,
      "grad_norm": 0.1503182202577591,
      "learning_rate": 1.5175e-05,
      "loss": 0.623,
      "step": 1393
    },
    {
      "epoch": 1.7886843899113838,
      "grad_norm": 0.15200340747833252,
      "learning_rate": 1.515e-05,
      "loss": 0.6282,
      "step": 1394
    },
    {
      "epoch": 1.7899675207506316,
      "grad_norm": 0.11768672615289688,
      "learning_rate": 1.5125e-05,
      "loss": 0.6924,
      "step": 1395
    },
    {
      "epoch": 1.7912506515898792,
      "grad_norm": 0.13647012412548065,
      "learning_rate": 1.51e-05,
      "loss": 0.6154,
      "step": 1396
    },
    {
      "epoch": 1.792533782429127,
      "grad_norm": 0.1407051384449005,
      "learning_rate": 1.5075e-05,
      "loss": 0.7332,
      "step": 1397
    },
    {
      "epoch": 1.7938169132683748,
      "grad_norm": 0.13454091548919678,
      "learning_rate": 1.505e-05,
      "loss": 0.6508,
      "step": 1398
    },
    {
      "epoch": 1.7951000441076226,
      "grad_norm": 0.14547018706798553,
      "learning_rate": 1.5025000000000001e-05,
      "loss": 0.6835,
      "step": 1399
    },
    {
      "epoch": 1.7963831749468704,
      "grad_norm": 0.1292678862810135,
      "learning_rate": 1.5e-05,
      "loss": 0.7144,
      "step": 1400
    },
    {
      "epoch": 1.7976663057861182,
      "grad_norm": 0.12477909028530121,
      "learning_rate": 1.4975e-05,
      "loss": 0.5957,
      "step": 1401
    },
    {
      "epoch": 1.798949436625366,
      "grad_norm": 0.15178154408931732,
      "learning_rate": 1.4950000000000001e-05,
      "loss": 0.5926,
      "step": 1402
    },
    {
      "epoch": 1.8002325674646138,
      "grad_norm": 0.1250164806842804,
      "learning_rate": 1.4925e-05,
      "loss": 0.6591,
      "step": 1403
    },
    {
      "epoch": 1.8015156983038614,
      "grad_norm": 0.16010189056396484,
      "learning_rate": 1.49e-05,
      "loss": 0.695,
      "step": 1404
    },
    {
      "epoch": 1.8027988291431092,
      "grad_norm": 0.12774349749088287,
      "learning_rate": 1.4875e-05,
      "loss": 0.6418,
      "step": 1405
    },
    {
      "epoch": 1.804081959982357,
      "grad_norm": 0.1260654479265213,
      "learning_rate": 1.485e-05,
      "loss": 0.6865,
      "step": 1406
    },
    {
      "epoch": 1.8053650908216046,
      "grad_norm": 0.12946397066116333,
      "learning_rate": 1.4825e-05,
      "loss": 0.7041,
      "step": 1407
    },
    {
      "epoch": 1.8066482216608524,
      "grad_norm": 0.1322648674249649,
      "learning_rate": 1.48e-05,
      "loss": 0.6313,
      "step": 1408
    },
    {
      "epoch": 1.8079313525001002,
      "grad_norm": 0.17252841591835022,
      "learning_rate": 1.4775e-05,
      "loss": 0.6641,
      "step": 1409
    },
    {
      "epoch": 1.809214483339348,
      "grad_norm": 0.13678297400474548,
      "learning_rate": 1.475e-05,
      "loss": 0.6743,
      "step": 1410
    },
    {
      "epoch": 1.8104976141785958,
      "grad_norm": 0.13106144964694977,
      "learning_rate": 1.4725e-05,
      "loss": 0.6402,
      "step": 1411
    },
    {
      "epoch": 1.8117807450178436,
      "grad_norm": 0.14982549846172333,
      "learning_rate": 1.47e-05,
      "loss": 0.6928,
      "step": 1412
    },
    {
      "epoch": 1.8130638758570914,
      "grad_norm": 0.13002970814704895,
      "learning_rate": 1.4675e-05,
      "loss": 0.5503,
      "step": 1413
    },
    {
      "epoch": 1.8143470066963392,
      "grad_norm": 0.17042480409145355,
      "learning_rate": 1.465e-05,
      "loss": 0.7887,
      "step": 1414
    },
    {
      "epoch": 1.815630137535587,
      "grad_norm": 0.1825023889541626,
      "learning_rate": 1.4625e-05,
      "loss": 0.6204,
      "step": 1415
    },
    {
      "epoch": 1.8169132683748346,
      "grad_norm": 0.15606853365898132,
      "learning_rate": 1.4599999999999999e-05,
      "loss": 0.63,
      "step": 1416
    },
    {
      "epoch": 1.8181963992140824,
      "grad_norm": 0.21935994923114777,
      "learning_rate": 1.4575e-05,
      "loss": 0.6708,
      "step": 1417
    },
    {
      "epoch": 1.81947953005333,
      "grad_norm": 0.143601194024086,
      "learning_rate": 1.455e-05,
      "loss": 0.7063,
      "step": 1418
    },
    {
      "epoch": 1.8207626608925778,
      "grad_norm": 0.17767834663391113,
      "learning_rate": 1.4524999999999999e-05,
      "loss": 0.7282,
      "step": 1419
    },
    {
      "epoch": 1.8220457917318256,
      "grad_norm": 0.14025205373764038,
      "learning_rate": 1.45e-05,
      "loss": 0.6645,
      "step": 1420
    },
    {
      "epoch": 1.8233289225710734,
      "grad_norm": 0.13456682860851288,
      "learning_rate": 1.4475e-05,
      "loss": 0.6083,
      "step": 1421
    },
    {
      "epoch": 1.8246120534103212,
      "grad_norm": 0.1716422289609909,
      "learning_rate": 1.4449999999999999e-05,
      "loss": 0.6597,
      "step": 1422
    },
    {
      "epoch": 1.825895184249569,
      "grad_norm": 0.15895067155361176,
      "learning_rate": 1.4425e-05,
      "loss": 0.6863,
      "step": 1423
    },
    {
      "epoch": 1.8271783150888168,
      "grad_norm": 0.13537825644016266,
      "learning_rate": 1.44e-05,
      "loss": 0.7607,
      "step": 1424
    },
    {
      "epoch": 1.8284614459280646,
      "grad_norm": 0.17973536252975464,
      "learning_rate": 1.4374999999999999e-05,
      "loss": 0.5877,
      "step": 1425
    },
    {
      "epoch": 1.8297445767673124,
      "grad_norm": 0.15202227234840393,
      "learning_rate": 1.435e-05,
      "loss": 0.639,
      "step": 1426
    },
    {
      "epoch": 1.83102770760656,
      "grad_norm": 0.1577380895614624,
      "learning_rate": 1.4325e-05,
      "loss": 0.699,
      "step": 1427
    },
    {
      "epoch": 1.8323108384458078,
      "grad_norm": 0.17819897830486298,
      "learning_rate": 1.43e-05,
      "loss": 0.7261,
      "step": 1428
    },
    {
      "epoch": 1.8335939692850556,
      "grad_norm": 0.13524161279201508,
      "learning_rate": 1.4275e-05,
      "loss": 0.5644,
      "step": 1429
    },
    {
      "epoch": 1.8348771001243032,
      "grad_norm": 0.19910091161727905,
      "learning_rate": 1.4249999999999999e-05,
      "loss": 0.5978,
      "step": 1430
    },
    {
      "epoch": 1.836160230963551,
      "grad_norm": 0.1504639834165573,
      "learning_rate": 1.4225e-05,
      "loss": 0.7159,
      "step": 1431
    },
    {
      "epoch": 1.8374433618027988,
      "grad_norm": 0.1357908397912979,
      "learning_rate": 1.42e-05,
      "loss": 0.7522,
      "step": 1432
    },
    {
      "epoch": 1.8387264926420466,
      "grad_norm": 0.15317556262016296,
      "learning_rate": 1.4174999999999999e-05,
      "loss": 0.6425,
      "step": 1433
    },
    {
      "epoch": 1.8400096234812944,
      "grad_norm": 0.1392197459936142,
      "learning_rate": 1.415e-05,
      "loss": 0.669,
      "step": 1434
    },
    {
      "epoch": 1.8412927543205422,
      "grad_norm": 0.1639438271522522,
      "learning_rate": 1.4125e-05,
      "loss": 0.593,
      "step": 1435
    },
    {
      "epoch": 1.84257588515979,
      "grad_norm": 0.12353084981441498,
      "learning_rate": 1.4099999999999999e-05,
      "loss": 0.6178,
      "step": 1436
    },
    {
      "epoch": 1.8438590159990378,
      "grad_norm": 0.13696624338626862,
      "learning_rate": 1.4075e-05,
      "loss": 0.5264,
      "step": 1437
    },
    {
      "epoch": 1.8451421468382854,
      "grad_norm": 0.2005489468574524,
      "learning_rate": 1.4050000000000003e-05,
      "loss": 0.7025,
      "step": 1438
    },
    {
      "epoch": 1.8464252776775332,
      "grad_norm": 0.16725142300128937,
      "learning_rate": 1.4025000000000002e-05,
      "loss": 0.7252,
      "step": 1439
    },
    {
      "epoch": 1.847708408516781,
      "grad_norm": 0.1941082775592804,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 0.7128,
      "step": 1440
    },
    {
      "epoch": 1.8489915393560286,
      "grad_norm": 0.12054339796304703,
      "learning_rate": 1.3975000000000003e-05,
      "loss": 0.6982,
      "step": 1441
    },
    {
      "epoch": 1.8502746701952764,
      "grad_norm": 0.17013908922672272,
      "learning_rate": 1.3950000000000002e-05,
      "loss": 0.5812,
      "step": 1442
    },
    {
      "epoch": 1.8515578010345242,
      "grad_norm": 0.1486055552959442,
      "learning_rate": 1.3925000000000001e-05,
      "loss": 0.6591,
      "step": 1443
    },
    {
      "epoch": 1.852840931873772,
      "grad_norm": 0.13397161662578583,
      "learning_rate": 1.3900000000000002e-05,
      "loss": 0.6583,
      "step": 1444
    },
    {
      "epoch": 1.8541240627130198,
      "grad_norm": 0.14717556536197662,
      "learning_rate": 1.3875000000000002e-05,
      "loss": 0.6026,
      "step": 1445
    },
    {
      "epoch": 1.8554071935522676,
      "grad_norm": 0.14378762245178223,
      "learning_rate": 1.3850000000000001e-05,
      "loss": 0.687,
      "step": 1446
    },
    {
      "epoch": 1.8566903243915154,
      "grad_norm": 0.12297289073467255,
      "learning_rate": 1.3825000000000002e-05,
      "loss": 0.6137,
      "step": 1447
    },
    {
      "epoch": 1.8579734552307632,
      "grad_norm": 0.12644359469413757,
      "learning_rate": 1.3800000000000002e-05,
      "loss": 0.6581,
      "step": 1448
    },
    {
      "epoch": 1.859256586070011,
      "grad_norm": 0.1382635235786438,
      "learning_rate": 1.3775000000000001e-05,
      "loss": 0.6351,
      "step": 1449
    },
    {
      "epoch": 1.8605397169092586,
      "grad_norm": 0.1874188929796219,
      "learning_rate": 1.3750000000000002e-05,
      "loss": 0.6932,
      "step": 1450
    },
    {
      "epoch": 1.8618228477485064,
      "grad_norm": 0.13920597732067108,
      "learning_rate": 1.3725000000000002e-05,
      "loss": 0.7122,
      "step": 1451
    },
    {
      "epoch": 1.8631059785877542,
      "grad_norm": 0.1680184006690979,
      "learning_rate": 1.3700000000000001e-05,
      "loss": 0.6553,
      "step": 1452
    },
    {
      "epoch": 1.8643891094270018,
      "grad_norm": 0.1353582888841629,
      "learning_rate": 1.3675000000000002e-05,
      "loss": 0.7125,
      "step": 1453
    },
    {
      "epoch": 1.8656722402662496,
      "grad_norm": 0.1404341161251068,
      "learning_rate": 1.3650000000000001e-05,
      "loss": 0.6671,
      "step": 1454
    },
    {
      "epoch": 1.8669553711054974,
      "grad_norm": 0.20652219653129578,
      "learning_rate": 1.3625e-05,
      "loss": 0.7543,
      "step": 1455
    },
    {
      "epoch": 1.8682385019447452,
      "grad_norm": 0.13235893845558167,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.7011,
      "step": 1456
    },
    {
      "epoch": 1.869521632783993,
      "grad_norm": 0.2352202981710434,
      "learning_rate": 1.3575000000000001e-05,
      "loss": 0.7055,
      "step": 1457
    },
    {
      "epoch": 1.8708047636232408,
      "grad_norm": 0.12791766226291656,
      "learning_rate": 1.3550000000000002e-05,
      "loss": 0.7386,
      "step": 1458
    },
    {
      "epoch": 1.8720878944624886,
      "grad_norm": 0.19631804525852203,
      "learning_rate": 1.3525000000000002e-05,
      "loss": 0.7162,
      "step": 1459
    },
    {
      "epoch": 1.8733710253017364,
      "grad_norm": 0.18851901590824127,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.6348,
      "step": 1460
    },
    {
      "epoch": 1.874654156140984,
      "grad_norm": 0.22382518649101257,
      "learning_rate": 1.3475000000000002e-05,
      "loss": 0.7586,
      "step": 1461
    },
    {
      "epoch": 1.8759372869802318,
      "grad_norm": 0.15082311630249023,
      "learning_rate": 1.3450000000000002e-05,
      "loss": 0.6858,
      "step": 1462
    },
    {
      "epoch": 1.8772204178194796,
      "grad_norm": 0.19259871542453766,
      "learning_rate": 1.3425000000000001e-05,
      "loss": 0.6228,
      "step": 1463
    },
    {
      "epoch": 1.8785035486587272,
      "grad_norm": 0.1378374993801117,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 0.5711,
      "step": 1464
    },
    {
      "epoch": 1.879786679497975,
      "grad_norm": 0.1362425535917282,
      "learning_rate": 1.3375000000000002e-05,
      "loss": 0.66,
      "step": 1465
    },
    {
      "epoch": 1.8810698103372228,
      "grad_norm": 0.15910939872264862,
      "learning_rate": 1.3350000000000001e-05,
      "loss": 0.663,
      "step": 1466
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 0.2237401008605957,
      "learning_rate": 1.3325000000000002e-05,
      "loss": 0.6308,
      "step": 1467
    },
    {
      "epoch": 1.8836360720157184,
      "grad_norm": 0.1142146959900856,
      "learning_rate": 1.3300000000000001e-05,
      "loss": 0.7748,
      "step": 1468
    },
    {
      "epoch": 1.8849192028549662,
      "grad_norm": 0.16637159883975983,
      "learning_rate": 1.3275e-05,
      "loss": 0.6516,
      "step": 1469
    },
    {
      "epoch": 1.886202333694214,
      "grad_norm": 0.3004128634929657,
      "learning_rate": 1.3250000000000002e-05,
      "loss": 0.65,
      "step": 1470
    },
    {
      "epoch": 1.8874854645334618,
      "grad_norm": 0.18116283416748047,
      "learning_rate": 1.3225000000000001e-05,
      "loss": 0.5883,
      "step": 1471
    },
    {
      "epoch": 1.8887685953727094,
      "grad_norm": 0.1987968534231186,
      "learning_rate": 1.32e-05,
      "loss": 0.7048,
      "step": 1472
    },
    {
      "epoch": 1.8900517262119572,
      "grad_norm": 0.23374143242835999,
      "learning_rate": 1.3175000000000002e-05,
      "loss": 0.6694,
      "step": 1473
    },
    {
      "epoch": 1.891334857051205,
      "grad_norm": 0.17063762247562408,
      "learning_rate": 1.3150000000000001e-05,
      "loss": 0.6317,
      "step": 1474
    },
    {
      "epoch": 1.8926179878904525,
      "grad_norm": 0.14853981137275696,
      "learning_rate": 1.3125e-05,
      "loss": 0.6941,
      "step": 1475
    },
    {
      "epoch": 1.8939011187297004,
      "grad_norm": 0.13045956194400787,
      "learning_rate": 1.3100000000000002e-05,
      "loss": 0.6678,
      "step": 1476
    },
    {
      "epoch": 1.8951842495689482,
      "grad_norm": 0.16737574338912964,
      "learning_rate": 1.3075000000000001e-05,
      "loss": 0.6743,
      "step": 1477
    },
    {
      "epoch": 1.896467380408196,
      "grad_norm": 0.12220614403486252,
      "learning_rate": 1.305e-05,
      "loss": 0.64,
      "step": 1478
    },
    {
      "epoch": 1.8977505112474438,
      "grad_norm": 0.15594695508480072,
      "learning_rate": 1.3025000000000002e-05,
      "loss": 0.753,
      "step": 1479
    },
    {
      "epoch": 1.8990336420866916,
      "grad_norm": 0.23242434859275818,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.6713,
      "step": 1480
    },
    {
      "epoch": 1.9003167729259394,
      "grad_norm": 0.23542509973049164,
      "learning_rate": 1.2975e-05,
      "loss": 0.5955,
      "step": 1481
    },
    {
      "epoch": 1.9015999037651872,
      "grad_norm": 0.1638963371515274,
      "learning_rate": 1.2950000000000001e-05,
      "loss": 0.6821,
      "step": 1482
    },
    {
      "epoch": 1.902883034604435,
      "grad_norm": 0.2904958724975586,
      "learning_rate": 1.2925e-05,
      "loss": 0.5711,
      "step": 1483
    },
    {
      "epoch": 1.9041661654436826,
      "grad_norm": 0.18574759364128113,
      "learning_rate": 1.29e-05,
      "loss": 0.7268,
      "step": 1484
    },
    {
      "epoch": 1.9054492962829304,
      "grad_norm": 0.14318229258060455,
      "learning_rate": 1.2875000000000001e-05,
      "loss": 0.6866,
      "step": 1485
    },
    {
      "epoch": 1.9067324271221782,
      "grad_norm": 0.16622266173362732,
      "learning_rate": 1.285e-05,
      "loss": 0.7743,
      "step": 1486
    },
    {
      "epoch": 1.9080155579614257,
      "grad_norm": 0.14762765169143677,
      "learning_rate": 1.2825000000000002e-05,
      "loss": 0.7071,
      "step": 1487
    },
    {
      "epoch": 1.9092986888006735,
      "grad_norm": 0.1915442943572998,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.5762,
      "step": 1488
    },
    {
      "epoch": 1.9105818196399214,
      "grad_norm": 0.1702679544687271,
      "learning_rate": 1.2775e-05,
      "loss": 0.7077,
      "step": 1489
    },
    {
      "epoch": 1.9118649504791692,
      "grad_norm": 0.24630101025104523,
      "learning_rate": 1.2750000000000002e-05,
      "loss": 0.5773,
      "step": 1490
    },
    {
      "epoch": 1.913148081318417,
      "grad_norm": 0.14439129829406738,
      "learning_rate": 1.2725000000000001e-05,
      "loss": 0.7751,
      "step": 1491
    },
    {
      "epoch": 1.9144312121576648,
      "grad_norm": 0.15124794840812683,
      "learning_rate": 1.27e-05,
      "loss": 0.7616,
      "step": 1492
    },
    {
      "epoch": 1.9157143429969126,
      "grad_norm": 0.16320733726024628,
      "learning_rate": 1.2675000000000001e-05,
      "loss": 0.7419,
      "step": 1493
    },
    {
      "epoch": 1.9169974738361604,
      "grad_norm": 0.2036639153957367,
      "learning_rate": 1.2650000000000001e-05,
      "loss": 0.6663,
      "step": 1494
    },
    {
      "epoch": 1.918280604675408,
      "grad_norm": 0.14129449427127838,
      "learning_rate": 1.2625e-05,
      "loss": 0.6325,
      "step": 1495
    },
    {
      "epoch": 1.9195637355146558,
      "grad_norm": 0.1338912397623062,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 0.7947,
      "step": 1496
    },
    {
      "epoch": 1.9208468663539036,
      "grad_norm": 0.1264069378376007,
      "learning_rate": 1.2575e-05,
      "loss": 0.7159,
      "step": 1497
    },
    {
      "epoch": 1.9221299971931511,
      "grad_norm": 0.12937413156032562,
      "learning_rate": 1.255e-05,
      "loss": 0.7477,
      "step": 1498
    },
    {
      "epoch": 1.923413128032399,
      "grad_norm": 0.18050144612789154,
      "learning_rate": 1.2525000000000001e-05,
      "loss": 0.7045,
      "step": 1499
    },
    {
      "epoch": 1.9246962588716467,
      "grad_norm": 0.18447187542915344,
      "learning_rate": 1.25e-05,
      "loss": 0.6533,
      "step": 1500
    },
    {
      "epoch": 1.9259793897108946,
      "grad_norm": 0.13267916440963745,
      "learning_rate": 1.2475e-05,
      "loss": 0.6447,
      "step": 1501
    },
    {
      "epoch": 1.9272625205501424,
      "grad_norm": 0.14592881500720978,
      "learning_rate": 1.2450000000000001e-05,
      "loss": 0.6442,
      "step": 1502
    },
    {
      "epoch": 1.9285456513893902,
      "grad_norm": 0.149007186293602,
      "learning_rate": 1.2425e-05,
      "loss": 0.6936,
      "step": 1503
    },
    {
      "epoch": 1.929828782228638,
      "grad_norm": 0.1606464982032776,
      "learning_rate": 1.24e-05,
      "loss": 0.7392,
      "step": 1504
    },
    {
      "epoch": 1.9311119130678858,
      "grad_norm": 0.14403480291366577,
      "learning_rate": 1.2375000000000001e-05,
      "loss": 0.6153,
      "step": 1505
    },
    {
      "epoch": 1.9323950439071336,
      "grad_norm": 0.1547730565071106,
      "learning_rate": 1.235e-05,
      "loss": 0.6977,
      "step": 1506
    },
    {
      "epoch": 1.9336781747463812,
      "grad_norm": 0.12914840877056122,
      "learning_rate": 1.2325e-05,
      "loss": 0.7372,
      "step": 1507
    },
    {
      "epoch": 1.934961305585629,
      "grad_norm": 0.1220213919878006,
      "learning_rate": 1.23e-05,
      "loss": 0.6434,
      "step": 1508
    },
    {
      "epoch": 1.9362444364248765,
      "grad_norm": 0.1548054963350296,
      "learning_rate": 1.2275e-05,
      "loss": 0.627,
      "step": 1509
    },
    {
      "epoch": 1.9375275672641243,
      "grad_norm": 0.18845023214817047,
      "learning_rate": 1.225e-05,
      "loss": 0.6917,
      "step": 1510
    },
    {
      "epoch": 1.9388106981033721,
      "grad_norm": 0.1394408941268921,
      "learning_rate": 1.2225e-05,
      "loss": 0.7319,
      "step": 1511
    },
    {
      "epoch": 1.94009382894262,
      "grad_norm": 0.1565275937318802,
      "learning_rate": 1.22e-05,
      "loss": 0.6014,
      "step": 1512
    },
    {
      "epoch": 1.9413769597818678,
      "grad_norm": 0.16563591361045837,
      "learning_rate": 1.2175e-05,
      "loss": 0.623,
      "step": 1513
    },
    {
      "epoch": 1.9426600906211156,
      "grad_norm": 0.235155388712883,
      "learning_rate": 1.215e-05,
      "loss": 0.6872,
      "step": 1514
    },
    {
      "epoch": 1.9439432214603634,
      "grad_norm": 0.14621759951114655,
      "learning_rate": 1.2125e-05,
      "loss": 0.6969,
      "step": 1515
    },
    {
      "epoch": 1.9452263522996112,
      "grad_norm": 0.14707410335540771,
      "learning_rate": 1.2100000000000001e-05,
      "loss": 0.6226,
      "step": 1516
    },
    {
      "epoch": 1.946509483138859,
      "grad_norm": 0.2238488644361496,
      "learning_rate": 1.2075e-05,
      "loss": 0.6851,
      "step": 1517
    },
    {
      "epoch": 1.9477926139781065,
      "grad_norm": 0.13058927655220032,
      "learning_rate": 1.205e-05,
      "loss": 0.6231,
      "step": 1518
    },
    {
      "epoch": 1.9490757448173543,
      "grad_norm": 0.18839363753795624,
      "learning_rate": 1.2025000000000001e-05,
      "loss": 0.6785,
      "step": 1519
    },
    {
      "epoch": 1.9503588756566022,
      "grad_norm": 0.13708648085594177,
      "learning_rate": 1.2e-05,
      "loss": 0.7154,
      "step": 1520
    },
    {
      "epoch": 1.9516420064958497,
      "grad_norm": 0.1808280646800995,
      "learning_rate": 1.1975e-05,
      "loss": 0.5893,
      "step": 1521
    },
    {
      "epoch": 1.9529251373350975,
      "grad_norm": 0.13736313581466675,
      "learning_rate": 1.195e-05,
      "loss": 0.7093,
      "step": 1522
    },
    {
      "epoch": 1.9542082681743453,
      "grad_norm": 0.14489923417568207,
      "learning_rate": 1.1925e-05,
      "loss": 0.7508,
      "step": 1523
    },
    {
      "epoch": 1.9554913990135931,
      "grad_norm": 0.2685537040233612,
      "learning_rate": 1.19e-05,
      "loss": 0.7657,
      "step": 1524
    },
    {
      "epoch": 1.956774529852841,
      "grad_norm": 0.25685834884643555,
      "learning_rate": 1.1875e-05,
      "loss": 0.7049,
      "step": 1525
    },
    {
      "epoch": 1.9580576606920888,
      "grad_norm": 0.1670665144920349,
      "learning_rate": 1.185e-05,
      "loss": 0.5325,
      "step": 1526
    },
    {
      "epoch": 1.9593407915313366,
      "grad_norm": 0.3145669102668762,
      "learning_rate": 1.1825e-05,
      "loss": 0.6984,
      "step": 1527
    },
    {
      "epoch": 1.9606239223705844,
      "grad_norm": 0.2640247642993927,
      "learning_rate": 1.18e-05,
      "loss": 0.7855,
      "step": 1528
    },
    {
      "epoch": 1.961907053209832,
      "grad_norm": 0.14513202011585236,
      "learning_rate": 1.1775e-05,
      "loss": 0.7081,
      "step": 1529
    },
    {
      "epoch": 1.9631901840490797,
      "grad_norm": 0.317045658826828,
      "learning_rate": 1.175e-05,
      "loss": 0.6648,
      "step": 1530
    },
    {
      "epoch": 1.9644733148883275,
      "grad_norm": 0.27277788519859314,
      "learning_rate": 1.1725e-05,
      "loss": 0.677,
      "step": 1531
    },
    {
      "epoch": 1.9657564457275751,
      "grad_norm": 0.2551770508289337,
      "learning_rate": 1.1700000000000001e-05,
      "loss": 0.7118,
      "step": 1532
    },
    {
      "epoch": 1.967039576566823,
      "grad_norm": 0.1343768686056137,
      "learning_rate": 1.1675000000000001e-05,
      "loss": 0.5847,
      "step": 1533
    },
    {
      "epoch": 1.9683227074060707,
      "grad_norm": 0.2789246439933777,
      "learning_rate": 1.1650000000000002e-05,
      "loss": 0.7215,
      "step": 1534
    },
    {
      "epoch": 1.9696058382453185,
      "grad_norm": 0.2754831910133362,
      "learning_rate": 1.1625000000000001e-05,
      "loss": 0.6959,
      "step": 1535
    },
    {
      "epoch": 1.9708889690845663,
      "grad_norm": 0.16960623860359192,
      "learning_rate": 1.16e-05,
      "loss": 0.7342,
      "step": 1536
    },
    {
      "epoch": 1.9721720999238141,
      "grad_norm": 0.1266731470823288,
      "learning_rate": 1.1575000000000002e-05,
      "loss": 0.6376,
      "step": 1537
    },
    {
      "epoch": 1.973455230763062,
      "grad_norm": 0.21306449174880981,
      "learning_rate": 1.1550000000000001e-05,
      "loss": 0.7186,
      "step": 1538
    },
    {
      "epoch": 1.9747383616023098,
      "grad_norm": 0.2596346437931061,
      "learning_rate": 1.1525e-05,
      "loss": 0.5741,
      "step": 1539
    },
    {
      "epoch": 1.9760214924415576,
      "grad_norm": 0.2460293471813202,
      "learning_rate": 1.1500000000000002e-05,
      "loss": 0.6486,
      "step": 1540
    },
    {
      "epoch": 1.9773046232808051,
      "grad_norm": 0.14821656048297882,
      "learning_rate": 1.1475000000000001e-05,
      "loss": 0.6394,
      "step": 1541
    },
    {
      "epoch": 1.978587754120053,
      "grad_norm": 0.18355102837085724,
      "learning_rate": 1.145e-05,
      "loss": 0.7329,
      "step": 1542
    },
    {
      "epoch": 1.9798708849593007,
      "grad_norm": 0.2686123549938202,
      "learning_rate": 1.1425000000000002e-05,
      "loss": 0.6268,
      "step": 1543
    },
    {
      "epoch": 1.9811540157985483,
      "grad_norm": 0.19398203492164612,
      "learning_rate": 1.1400000000000001e-05,
      "loss": 0.553,
      "step": 1544
    },
    {
      "epoch": 1.9824371466377961,
      "grad_norm": 0.1478540599346161,
      "learning_rate": 1.1375e-05,
      "loss": 0.6467,
      "step": 1545
    },
    {
      "epoch": 1.983720277477044,
      "grad_norm": 0.25116533041000366,
      "learning_rate": 1.1350000000000001e-05,
      "loss": 0.7256,
      "step": 1546
    },
    {
      "epoch": 1.9850034083162917,
      "grad_norm": 0.34043675661087036,
      "learning_rate": 1.1325e-05,
      "loss": 0.5385,
      "step": 1547
    },
    {
      "epoch": 1.9862865391555395,
      "grad_norm": 0.11930978298187256,
      "learning_rate": 1.13e-05,
      "loss": 0.6745,
      "step": 1548
    },
    {
      "epoch": 1.9875696699947873,
      "grad_norm": 0.16984637081623077,
      "learning_rate": 1.1275000000000001e-05,
      "loss": 0.6842,
      "step": 1549
    },
    {
      "epoch": 1.9888528008340352,
      "grad_norm": 0.2523157298564911,
      "learning_rate": 1.125e-05,
      "loss": 0.7033,
      "step": 1550
    },
    {
      "epoch": 1.990135931673283,
      "grad_norm": 0.2185674011707306,
      "learning_rate": 1.1225e-05,
      "loss": 0.7077,
      "step": 1551
    },
    {
      "epoch": 1.9914190625125305,
      "grad_norm": 0.1426929235458374,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.7058,
      "step": 1552
    },
    {
      "epoch": 1.9927021933517783,
      "grad_norm": 0.17399674654006958,
      "learning_rate": 1.1175e-05,
      "loss": 0.6886,
      "step": 1553
    },
    {
      "epoch": 1.9939853241910261,
      "grad_norm": 0.21891486644744873,
      "learning_rate": 1.115e-05,
      "loss": 0.6641,
      "step": 1554
    },
    {
      "epoch": 1.9952684550302737,
      "grad_norm": 2.256788492202759,
      "learning_rate": 1.1125000000000001e-05,
      "loss": 0.614,
      "step": 1555
    },
    {
      "epoch": 1.9965515858695215,
      "grad_norm": 0.13035763800144196,
      "learning_rate": 1.11e-05,
      "loss": 0.7122,
      "step": 1556
    },
    {
      "epoch": 1.9978347167087693,
      "grad_norm": 0.1669803112745285,
      "learning_rate": 1.1075e-05,
      "loss": 0.6363,
      "step": 1557
    },
    {
      "epoch": 1.9991178475480171,
      "grad_norm": 0.16696332395076752,
      "learning_rate": 1.1050000000000001e-05,
      "loss": 0.7633,
      "step": 1558
    },
    {
      "epoch": 2.000400978387265,
      "grad_norm": 0.17861796915531158,
      "learning_rate": 1.1025e-05,
      "loss": 0.7056,
      "step": 1559
    },
    {
      "epoch": 2.0016841092265127,
      "grad_norm": 0.12357054650783539,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.715,
      "step": 1560
    },
    {
      "epoch": 2.0029672400657605,
      "grad_norm": 0.1746291071176529,
      "learning_rate": 1.0975e-05,
      "loss": 0.6716,
      "step": 1561
    },
    {
      "epoch": 2.0042503709050083,
      "grad_norm": 0.19787736237049103,
      "learning_rate": 1.095e-05,
      "loss": 0.6501,
      "step": 1562
    },
    {
      "epoch": 2.005533501744256,
      "grad_norm": 0.14717698097229004,
      "learning_rate": 1.0925000000000001e-05,
      "loss": 0.5817,
      "step": 1563
    },
    {
      "epoch": 2.006816632583504,
      "grad_norm": 0.16182328760623932,
      "learning_rate": 1.09e-05,
      "loss": 0.639,
      "step": 1564
    },
    {
      "epoch": 2.0080997634227513,
      "grad_norm": 0.15550200641155243,
      "learning_rate": 1.0875e-05,
      "loss": 0.6215,
      "step": 1565
    },
    {
      "epoch": 2.009382894261999,
      "grad_norm": 0.21498633921146393,
      "learning_rate": 1.0850000000000001e-05,
      "loss": 0.5887,
      "step": 1566
    },
    {
      "epoch": 2.010666025101247,
      "grad_norm": 0.12479162961244583,
      "learning_rate": 1.0825e-05,
      "loss": 0.5479,
      "step": 1567
    },
    {
      "epoch": 2.0119491559404947,
      "grad_norm": 0.16472657024860382,
      "learning_rate": 1.08e-05,
      "loss": 0.6773,
      "step": 1568
    },
    {
      "epoch": 2.0132322867797425,
      "grad_norm": 0.18999527394771576,
      "learning_rate": 1.0775000000000001e-05,
      "loss": 0.7436,
      "step": 1569
    },
    {
      "epoch": 2.0145154176189903,
      "grad_norm": 0.17993372678756714,
      "learning_rate": 1.075e-05,
      "loss": 0.5537,
      "step": 1570
    },
    {
      "epoch": 2.015798548458238,
      "grad_norm": 0.11974009871482849,
      "learning_rate": 1.0725e-05,
      "loss": 0.714,
      "step": 1571
    },
    {
      "epoch": 2.017081679297486,
      "grad_norm": 0.13963720202445984,
      "learning_rate": 1.0700000000000001e-05,
      "loss": 0.7372,
      "step": 1572
    },
    {
      "epoch": 2.0183648101367337,
      "grad_norm": 0.1924862414598465,
      "learning_rate": 1.0675e-05,
      "loss": 0.6206,
      "step": 1573
    },
    {
      "epoch": 2.0196479409759815,
      "grad_norm": 0.14117860794067383,
      "learning_rate": 1.065e-05,
      "loss": 0.6234,
      "step": 1574
    },
    {
      "epoch": 2.0209310718152294,
      "grad_norm": 0.14933103322982788,
      "learning_rate": 1.0625e-05,
      "loss": 0.6657,
      "step": 1575
    },
    {
      "epoch": 2.0222142026544767,
      "grad_norm": 0.20177270472049713,
      "learning_rate": 1.06e-05,
      "loss": 0.6627,
      "step": 1576
    },
    {
      "epoch": 2.0234973334937245,
      "grad_norm": 0.15579940378665924,
      "learning_rate": 1.0575e-05,
      "loss": 0.6631,
      "step": 1577
    },
    {
      "epoch": 2.0247804643329723,
      "grad_norm": 0.11961943656206131,
      "learning_rate": 1.055e-05,
      "loss": 0.7346,
      "step": 1578
    },
    {
      "epoch": 2.02606359517222,
      "grad_norm": 0.173823744058609,
      "learning_rate": 1.0525e-05,
      "loss": 0.6143,
      "step": 1579
    },
    {
      "epoch": 2.027346726011468,
      "grad_norm": 0.173222154378891,
      "learning_rate": 1.05e-05,
      "loss": 0.7029,
      "step": 1580
    },
    {
      "epoch": 2.0286298568507157,
      "grad_norm": 0.12250307947397232,
      "learning_rate": 1.0475e-05,
      "loss": 0.6483,
      "step": 1581
    },
    {
      "epoch": 2.0299129876899635,
      "grad_norm": 0.14830245077610016,
      "learning_rate": 1.045e-05,
      "loss": 0.7198,
      "step": 1582
    },
    {
      "epoch": 2.0311961185292113,
      "grad_norm": 0.14896734058856964,
      "learning_rate": 1.0425e-05,
      "loss": 0.6305,
      "step": 1583
    },
    {
      "epoch": 2.032479249368459,
      "grad_norm": 0.14422857761383057,
      "learning_rate": 1.04e-05,
      "loss": 0.706,
      "step": 1584
    },
    {
      "epoch": 2.033762380207707,
      "grad_norm": 0.1509624570608139,
      "learning_rate": 1.0375e-05,
      "loss": 0.7425,
      "step": 1585
    },
    {
      "epoch": 2.0350455110469547,
      "grad_norm": 0.1578909456729889,
      "learning_rate": 1.035e-05,
      "loss": 0.7511,
      "step": 1586
    },
    {
      "epoch": 2.0363286418862026,
      "grad_norm": 0.15453536808490753,
      "learning_rate": 1.0325e-05,
      "loss": 0.7138,
      "step": 1587
    },
    {
      "epoch": 2.03761177272545,
      "grad_norm": 0.12898020446300507,
      "learning_rate": 1.03e-05,
      "loss": 0.654,
      "step": 1588
    },
    {
      "epoch": 2.0388949035646977,
      "grad_norm": 0.15156765282154083,
      "learning_rate": 1.0275e-05,
      "loss": 0.647,
      "step": 1589
    },
    {
      "epoch": 2.0401780344039455,
      "grad_norm": 0.1488633155822754,
      "learning_rate": 1.025e-05,
      "loss": 0.6616,
      "step": 1590
    },
    {
      "epoch": 2.0414611652431933,
      "grad_norm": 0.139505997300148,
      "learning_rate": 1.0225e-05,
      "loss": 0.699,
      "step": 1591
    },
    {
      "epoch": 2.042744296082441,
      "grad_norm": 0.17622753977775574,
      "learning_rate": 1.02e-05,
      "loss": 0.7628,
      "step": 1592
    },
    {
      "epoch": 2.044027426921689,
      "grad_norm": 0.14387580752372742,
      "learning_rate": 1.0175e-05,
      "loss": 0.6737,
      "step": 1593
    },
    {
      "epoch": 2.0453105577609367,
      "grad_norm": 0.15649189054965973,
      "learning_rate": 1.0150000000000001e-05,
      "loss": 0.7542,
      "step": 1594
    },
    {
      "epoch": 2.0465936886001845,
      "grad_norm": 0.13956481218338013,
      "learning_rate": 1.0125e-05,
      "loss": 0.6197,
      "step": 1595
    },
    {
      "epoch": 2.0478768194394323,
      "grad_norm": 0.12911321222782135,
      "learning_rate": 1.0100000000000002e-05,
      "loss": 0.6348,
      "step": 1596
    },
    {
      "epoch": 2.04915995027868,
      "grad_norm": 0.1348663866519928,
      "learning_rate": 1.0075000000000001e-05,
      "loss": 0.6512,
      "step": 1597
    },
    {
      "epoch": 2.050443081117928,
      "grad_norm": 0.16021768748760223,
      "learning_rate": 1.005e-05,
      "loss": 0.8014,
      "step": 1598
    },
    {
      "epoch": 2.0517262119571753,
      "grad_norm": 0.12531141936779022,
      "learning_rate": 1.0025000000000001e-05,
      "loss": 0.7182,
      "step": 1599
    },
    {
      "epoch": 2.053009342796423,
      "grad_norm": 0.18550220131874084,
      "learning_rate": 1e-05,
      "loss": 0.6216,
      "step": 1600
    },
    {
      "epoch": 2.054292473635671,
      "grad_norm": 0.1856919825077057,
      "learning_rate": 9.975e-06,
      "loss": 0.6259,
      "step": 1601
    },
    {
      "epoch": 2.0555756044749187,
      "grad_norm": 0.134526327252388,
      "learning_rate": 9.950000000000001e-06,
      "loss": 0.7913,
      "step": 1602
    },
    {
      "epoch": 2.0568587353141665,
      "grad_norm": 0.1557689607143402,
      "learning_rate": 9.925e-06,
      "loss": 0.6501,
      "step": 1603
    },
    {
      "epoch": 2.0581418661534143,
      "grad_norm": 0.15253138542175293,
      "learning_rate": 9.900000000000002e-06,
      "loss": 0.6427,
      "step": 1604
    },
    {
      "epoch": 2.059424996992662,
      "grad_norm": 0.22030320763587952,
      "learning_rate": 9.875000000000001e-06,
      "loss": 0.6088,
      "step": 1605
    },
    {
      "epoch": 2.06070812783191,
      "grad_norm": 0.12460494786500931,
      "learning_rate": 9.85e-06,
      "loss": 0.6781,
      "step": 1606
    },
    {
      "epoch": 2.0619912586711577,
      "grad_norm": 0.24061134457588196,
      "learning_rate": 9.825000000000002e-06,
      "loss": 0.6973,
      "step": 1607
    },
    {
      "epoch": 2.0632743895104055,
      "grad_norm": 0.19142265617847443,
      "learning_rate": 9.800000000000001e-06,
      "loss": 0.6261,
      "step": 1608
    },
    {
      "epoch": 2.0645575203496533,
      "grad_norm": 0.1283390074968338,
      "learning_rate": 9.775e-06,
      "loss": 0.6572,
      "step": 1609
    },
    {
      "epoch": 2.0658406511889007,
      "grad_norm": 0.18083253502845764,
      "learning_rate": 9.750000000000002e-06,
      "loss": 0.6439,
      "step": 1610
    },
    {
      "epoch": 2.0671237820281485,
      "grad_norm": 0.23683129251003265,
      "learning_rate": 9.725000000000001e-06,
      "loss": 0.6436,
      "step": 1611
    },
    {
      "epoch": 2.0684069128673963,
      "grad_norm": 0.17096452414989471,
      "learning_rate": 9.7e-06,
      "loss": 0.6492,
      "step": 1612
    },
    {
      "epoch": 2.069690043706644,
      "grad_norm": 0.16207605600357056,
      "learning_rate": 9.675000000000001e-06,
      "loss": 0.6973,
      "step": 1613
    },
    {
      "epoch": 2.070973174545892,
      "grad_norm": 0.18866023421287537,
      "learning_rate": 9.65e-06,
      "loss": 0.7399,
      "step": 1614
    },
    {
      "epoch": 2.0722563053851397,
      "grad_norm": 0.18542003631591797,
      "learning_rate": 9.625e-06,
      "loss": 0.7095,
      "step": 1615
    },
    {
      "epoch": 2.0735394362243875,
      "grad_norm": 0.144646555185318,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.6762,
      "step": 1616
    },
    {
      "epoch": 2.0748225670636353,
      "grad_norm": 0.20865269005298615,
      "learning_rate": 9.575e-06,
      "loss": 0.6665,
      "step": 1617
    },
    {
      "epoch": 2.076105697902883,
      "grad_norm": 0.12314004451036453,
      "learning_rate": 9.55e-06,
      "loss": 0.6963,
      "step": 1618
    },
    {
      "epoch": 2.077388828742131,
      "grad_norm": 0.1609227955341339,
      "learning_rate": 9.525000000000001e-06,
      "loss": 0.6885,
      "step": 1619
    },
    {
      "epoch": 2.0786719595813787,
      "grad_norm": 0.16372966766357422,
      "learning_rate": 9.5e-06,
      "loss": 0.5866,
      "step": 1620
    },
    {
      "epoch": 2.0799550904206265,
      "grad_norm": 0.17851406335830688,
      "learning_rate": 9.475e-06,
      "loss": 0.6619,
      "step": 1621
    },
    {
      "epoch": 2.081238221259874,
      "grad_norm": 0.1362927109003067,
      "learning_rate": 9.450000000000001e-06,
      "loss": 0.6328,
      "step": 1622
    },
    {
      "epoch": 2.0825213520991217,
      "grad_norm": 0.1211976483464241,
      "learning_rate": 9.425e-06,
      "loss": 0.6998,
      "step": 1623
    },
    {
      "epoch": 2.0838044829383695,
      "grad_norm": 0.14508037269115448,
      "learning_rate": 9.4e-06,
      "loss": 0.5824,
      "step": 1624
    },
    {
      "epoch": 2.0850876137776173,
      "grad_norm": 0.1406944841146469,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.6887,
      "step": 1625
    },
    {
      "epoch": 2.086370744616865,
      "grad_norm": 0.14786794781684875,
      "learning_rate": 9.35e-06,
      "loss": 0.6754,
      "step": 1626
    },
    {
      "epoch": 2.087653875456113,
      "grad_norm": 0.13699856400489807,
      "learning_rate": 9.325e-06,
      "loss": 0.6555,
      "step": 1627
    },
    {
      "epoch": 2.0889370062953607,
      "grad_norm": 0.1486446112394333,
      "learning_rate": 9.3e-06,
      "loss": 0.5904,
      "step": 1628
    },
    {
      "epoch": 2.0902201371346085,
      "grad_norm": 0.14794659614562988,
      "learning_rate": 9.275e-06,
      "loss": 0.6728,
      "step": 1629
    },
    {
      "epoch": 2.0915032679738563,
      "grad_norm": 0.13809779286384583,
      "learning_rate": 9.25e-06,
      "loss": 0.6819,
      "step": 1630
    },
    {
      "epoch": 2.092786398813104,
      "grad_norm": 0.11856380105018616,
      "learning_rate": 9.225e-06,
      "loss": 0.7111,
      "step": 1631
    },
    {
      "epoch": 2.094069529652352,
      "grad_norm": 0.1256600022315979,
      "learning_rate": 9.2e-06,
      "loss": 0.6014,
      "step": 1632
    },
    {
      "epoch": 2.0953526604915993,
      "grad_norm": 0.20802545547485352,
      "learning_rate": 9.175000000000001e-06,
      "loss": 0.6922,
      "step": 1633
    },
    {
      "epoch": 2.096635791330847,
      "grad_norm": 0.12396729737520218,
      "learning_rate": 9.15e-06,
      "loss": 0.7246,
      "step": 1634
    },
    {
      "epoch": 2.097918922170095,
      "grad_norm": 0.15272389352321625,
      "learning_rate": 9.125e-06,
      "loss": 0.7472,
      "step": 1635
    },
    {
      "epoch": 2.0992020530093427,
      "grad_norm": 0.12153248488903046,
      "learning_rate": 9.100000000000001e-06,
      "loss": 0.5397,
      "step": 1636
    },
    {
      "epoch": 2.1004851838485905,
      "grad_norm": 0.13438157737255096,
      "learning_rate": 9.075e-06,
      "loss": 0.7145,
      "step": 1637
    },
    {
      "epoch": 2.1017683146878383,
      "grad_norm": 0.14056336879730225,
      "learning_rate": 9.05e-06,
      "loss": 0.6393,
      "step": 1638
    },
    {
      "epoch": 2.103051445527086,
      "grad_norm": 0.14400939643383026,
      "learning_rate": 9.025e-06,
      "loss": 0.6125,
      "step": 1639
    },
    {
      "epoch": 2.104334576366334,
      "grad_norm": 0.153264120221138,
      "learning_rate": 9e-06,
      "loss": 0.6221,
      "step": 1640
    },
    {
      "epoch": 2.1056177072055817,
      "grad_norm": 0.2843489646911621,
      "learning_rate": 8.975e-06,
      "loss": 0.6368,
      "step": 1641
    },
    {
      "epoch": 2.1069008380448295,
      "grad_norm": 0.1402052640914917,
      "learning_rate": 8.95e-06,
      "loss": 0.6906,
      "step": 1642
    },
    {
      "epoch": 2.1081839688840773,
      "grad_norm": 0.1446068286895752,
      "learning_rate": 8.925e-06,
      "loss": 0.7427,
      "step": 1643
    },
    {
      "epoch": 2.109467099723325,
      "grad_norm": 0.2499261349439621,
      "learning_rate": 8.9e-06,
      "loss": 0.5638,
      "step": 1644
    },
    {
      "epoch": 2.1107502305625725,
      "grad_norm": 0.16890954971313477,
      "learning_rate": 8.875e-06,
      "loss": 0.7408,
      "step": 1645
    },
    {
      "epoch": 2.1120333614018203,
      "grad_norm": 0.15861776471138,
      "learning_rate": 8.85e-06,
      "loss": 0.6565,
      "step": 1646
    },
    {
      "epoch": 2.113316492241068,
      "grad_norm": 0.18237391114234924,
      "learning_rate": 8.825e-06,
      "loss": 0.6227,
      "step": 1647
    },
    {
      "epoch": 2.114599623080316,
      "grad_norm": 0.2225266396999359,
      "learning_rate": 8.8e-06,
      "loss": 0.7491,
      "step": 1648
    },
    {
      "epoch": 2.1158827539195637,
      "grad_norm": 0.25529658794403076,
      "learning_rate": 8.775e-06,
      "loss": 0.8027,
      "step": 1649
    },
    {
      "epoch": 2.1171658847588115,
      "grad_norm": 0.1572997272014618,
      "learning_rate": 8.75e-06,
      "loss": 0.7173,
      "step": 1650
    },
    {
      "epoch": 2.1184490155980593,
      "grad_norm": 0.17005662620067596,
      "learning_rate": 8.725e-06,
      "loss": 0.7917,
      "step": 1651
    },
    {
      "epoch": 2.119732146437307,
      "grad_norm": 0.21567296981811523,
      "learning_rate": 8.7e-06,
      "loss": 0.7259,
      "step": 1652
    },
    {
      "epoch": 2.121015277276555,
      "grad_norm": 0.17847298085689545,
      "learning_rate": 8.674999999999999e-06,
      "loss": 0.6748,
      "step": 1653
    },
    {
      "epoch": 2.1222984081158027,
      "grad_norm": 0.16393904387950897,
      "learning_rate": 8.65e-06,
      "loss": 0.6929,
      "step": 1654
    },
    {
      "epoch": 2.1235815389550505,
      "grad_norm": 0.16520340740680695,
      "learning_rate": 8.625e-06,
      "loss": 0.6855,
      "step": 1655
    },
    {
      "epoch": 2.124864669794298,
      "grad_norm": 0.19630448520183563,
      "learning_rate": 8.599999999999999e-06,
      "loss": 0.8463,
      "step": 1656
    },
    {
      "epoch": 2.1261478006335457,
      "grad_norm": 0.1445370614528656,
      "learning_rate": 8.575000000000002e-06,
      "loss": 0.7168,
      "step": 1657
    },
    {
      "epoch": 2.1274309314727935,
      "grad_norm": 0.12170284986495972,
      "learning_rate": 8.550000000000001e-06,
      "loss": 0.659,
      "step": 1658
    },
    {
      "epoch": 2.1287140623120413,
      "grad_norm": 0.2065662145614624,
      "learning_rate": 8.525e-06,
      "loss": 0.7868,
      "step": 1659
    },
    {
      "epoch": 2.129997193151289,
      "grad_norm": 0.2783021926879883,
      "learning_rate": 8.500000000000002e-06,
      "loss": 0.6819,
      "step": 1660
    },
    {
      "epoch": 2.131280323990537,
      "grad_norm": 0.15881720185279846,
      "learning_rate": 8.475000000000001e-06,
      "loss": 0.5914,
      "step": 1661
    },
    {
      "epoch": 2.1325634548297847,
      "grad_norm": 0.2467505782842636,
      "learning_rate": 8.45e-06,
      "loss": 0.558,
      "step": 1662
    },
    {
      "epoch": 2.1338465856690325,
      "grad_norm": 0.2538348436355591,
      "learning_rate": 8.425000000000001e-06,
      "loss": 0.7247,
      "step": 1663
    },
    {
      "epoch": 2.1351297165082803,
      "grad_norm": 0.12376474589109421,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.6322,
      "step": 1664
    },
    {
      "epoch": 2.136412847347528,
      "grad_norm": 0.1395341008901596,
      "learning_rate": 8.375e-06,
      "loss": 0.6321,
      "step": 1665
    },
    {
      "epoch": 2.137695978186776,
      "grad_norm": 0.1714840978384018,
      "learning_rate": 8.350000000000001e-06,
      "loss": 0.6555,
      "step": 1666
    },
    {
      "epoch": 2.1389791090260237,
      "grad_norm": 0.14701274037361145,
      "learning_rate": 8.325e-06,
      "loss": 0.6915,
      "step": 1667
    },
    {
      "epoch": 2.140262239865271,
      "grad_norm": 0.13203680515289307,
      "learning_rate": 8.3e-06,
      "loss": 0.6802,
      "step": 1668
    },
    {
      "epoch": 2.141545370704519,
      "grad_norm": 0.13032852113246918,
      "learning_rate": 8.275000000000001e-06,
      "loss": 0.7321,
      "step": 1669
    },
    {
      "epoch": 2.1428285015437667,
      "grad_norm": 0.21127323806285858,
      "learning_rate": 8.25e-06,
      "loss": 0.6534,
      "step": 1670
    },
    {
      "epoch": 2.1441116323830145,
      "grad_norm": 0.12883509695529938,
      "learning_rate": 8.225e-06,
      "loss": 0.6282,
      "step": 1671
    },
    {
      "epoch": 2.1453947632222623,
      "grad_norm": 0.179097518324852,
      "learning_rate": 8.200000000000001e-06,
      "loss": 0.5532,
      "step": 1672
    },
    {
      "epoch": 2.14667789406151,
      "grad_norm": 0.13370627164840698,
      "learning_rate": 8.175e-06,
      "loss": 0.6996,
      "step": 1673
    },
    {
      "epoch": 2.147961024900758,
      "grad_norm": 0.1555623561143875,
      "learning_rate": 8.15e-06,
      "loss": 0.7189,
      "step": 1674
    },
    {
      "epoch": 2.1492441557400057,
      "grad_norm": 0.18491564691066742,
      "learning_rate": 8.125000000000001e-06,
      "loss": 0.7548,
      "step": 1675
    },
    {
      "epoch": 2.1505272865792535,
      "grad_norm": 0.1790684461593628,
      "learning_rate": 8.1e-06,
      "loss": 0.713,
      "step": 1676
    },
    {
      "epoch": 2.1518104174185013,
      "grad_norm": 0.13741157948970795,
      "learning_rate": 8.075000000000001e-06,
      "loss": 0.6818,
      "step": 1677
    },
    {
      "epoch": 2.1530935482577487,
      "grad_norm": 0.3462422490119934,
      "learning_rate": 8.050000000000001e-06,
      "loss": 0.594,
      "step": 1678
    },
    {
      "epoch": 2.1543766790969965,
      "grad_norm": 0.2955045998096466,
      "learning_rate": 8.025e-06,
      "loss": 0.6227,
      "step": 1679
    },
    {
      "epoch": 2.1556598099362443,
      "grad_norm": 0.21362438797950745,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.5954,
      "step": 1680
    },
    {
      "epoch": 2.156942940775492,
      "grad_norm": 0.2304479479789734,
      "learning_rate": 7.975e-06,
      "loss": 0.663,
      "step": 1681
    },
    {
      "epoch": 2.15822607161474,
      "grad_norm": 0.19798746705055237,
      "learning_rate": 7.95e-06,
      "loss": 0.5824,
      "step": 1682
    },
    {
      "epoch": 2.1595092024539877,
      "grad_norm": 0.15682680904865265,
      "learning_rate": 7.925000000000001e-06,
      "loss": 0.6083,
      "step": 1683
    },
    {
      "epoch": 2.1607923332932355,
      "grad_norm": 0.1363999992609024,
      "learning_rate": 7.9e-06,
      "loss": 0.6279,
      "step": 1684
    },
    {
      "epoch": 2.1620754641324833,
      "grad_norm": 0.3346864879131317,
      "learning_rate": 7.875e-06,
      "loss": 0.6093,
      "step": 1685
    },
    {
      "epoch": 2.163358594971731,
      "grad_norm": 0.17582584917545319,
      "learning_rate": 7.850000000000001e-06,
      "loss": 0.6746,
      "step": 1686
    },
    {
      "epoch": 2.164641725810979,
      "grad_norm": 0.1282002329826355,
      "learning_rate": 7.825e-06,
      "loss": 0.6371,
      "step": 1687
    },
    {
      "epoch": 2.1659248566502267,
      "grad_norm": 0.19875232875347137,
      "learning_rate": 7.8e-06,
      "loss": 0.7288,
      "step": 1688
    },
    {
      "epoch": 2.1672079874894745,
      "grad_norm": 0.22021135687828064,
      "learning_rate": 7.775000000000001e-06,
      "loss": 0.6434,
      "step": 1689
    },
    {
      "epoch": 2.168491118328722,
      "grad_norm": 0.16815219819545746,
      "learning_rate": 7.75e-06,
      "loss": 0.7122,
      "step": 1690
    },
    {
      "epoch": 2.1697742491679697,
      "grad_norm": 0.15925660729408264,
      "learning_rate": 7.725e-06,
      "loss": 0.7007,
      "step": 1691
    },
    {
      "epoch": 2.1710573800072175,
      "grad_norm": 0.18915124237537384,
      "learning_rate": 7.7e-06,
      "loss": 0.6508,
      "step": 1692
    },
    {
      "epoch": 2.1723405108464653,
      "grad_norm": 0.2605852484703064,
      "learning_rate": 7.675e-06,
      "loss": 0.6016,
      "step": 1693
    },
    {
      "epoch": 2.173623641685713,
      "grad_norm": 0.17195607721805573,
      "learning_rate": 7.65e-06,
      "loss": 0.6396,
      "step": 1694
    },
    {
      "epoch": 2.174906772524961,
      "grad_norm": 0.1878781020641327,
      "learning_rate": 7.625e-06,
      "loss": 0.6543,
      "step": 1695
    },
    {
      "epoch": 2.1761899033642087,
      "grad_norm": 0.1787392646074295,
      "learning_rate": 7.6e-06,
      "loss": 0.687,
      "step": 1696
    },
    {
      "epoch": 2.1774730342034565,
      "grad_norm": 0.16841579973697662,
      "learning_rate": 7.575e-06,
      "loss": 0.6708,
      "step": 1697
    },
    {
      "epoch": 2.1787561650427043,
      "grad_norm": 0.1305128037929535,
      "learning_rate": 7.55e-06,
      "loss": 0.6537,
      "step": 1698
    },
    {
      "epoch": 2.180039295881952,
      "grad_norm": 0.12137843668460846,
      "learning_rate": 7.525e-06,
      "loss": 0.6424,
      "step": 1699
    },
    {
      "epoch": 2.1813224267212,
      "grad_norm": 0.2070666253566742,
      "learning_rate": 7.5e-06,
      "loss": 0.7845,
      "step": 1700
    },
    {
      "epoch": 2.1826055575604473,
      "grad_norm": 0.13479135930538177,
      "learning_rate": 7.4750000000000004e-06,
      "loss": 0.5868,
      "step": 1701
    },
    {
      "epoch": 2.183888688399695,
      "grad_norm": 0.15035103261470795,
      "learning_rate": 7.45e-06,
      "loss": 0.6711,
      "step": 1702
    },
    {
      "epoch": 2.185171819238943,
      "grad_norm": 0.1421874761581421,
      "learning_rate": 7.425e-06,
      "loss": 0.6757,
      "step": 1703
    },
    {
      "epoch": 2.1864549500781907,
      "grad_norm": 0.1204628124833107,
      "learning_rate": 7.4e-06,
      "loss": 0.5981,
      "step": 1704
    },
    {
      "epoch": 2.1877380809174385,
      "grad_norm": 0.17500166594982147,
      "learning_rate": 7.375e-06,
      "loss": 0.6793,
      "step": 1705
    },
    {
      "epoch": 2.1890212117566863,
      "grad_norm": 0.15598618984222412,
      "learning_rate": 7.35e-06,
      "loss": 0.7309,
      "step": 1706
    },
    {
      "epoch": 2.190304342595934,
      "grad_norm": 0.14523835480213165,
      "learning_rate": 7.325e-06,
      "loss": 0.733,
      "step": 1707
    },
    {
      "epoch": 2.191587473435182,
      "grad_norm": 0.1546453833580017,
      "learning_rate": 7.2999999999999996e-06,
      "loss": 0.7586,
      "step": 1708
    },
    {
      "epoch": 2.1928706042744297,
      "grad_norm": 0.23904559016227722,
      "learning_rate": 7.275e-06,
      "loss": 0.7853,
      "step": 1709
    },
    {
      "epoch": 2.1941537351136775,
      "grad_norm": 0.12216968089342117,
      "learning_rate": 7.25e-06,
      "loss": 0.7499,
      "step": 1710
    },
    {
      "epoch": 2.1954368659529253,
      "grad_norm": 0.135368213057518,
      "learning_rate": 7.2249999999999994e-06,
      "loss": 0.6644,
      "step": 1711
    },
    {
      "epoch": 2.196719996792173,
      "grad_norm": 0.1824193149805069,
      "learning_rate": 7.2e-06,
      "loss": 0.7654,
      "step": 1712
    },
    {
      "epoch": 2.1980031276314205,
      "grad_norm": 0.1380457580089569,
      "learning_rate": 7.175e-06,
      "loss": 0.6764,
      "step": 1713
    },
    {
      "epoch": 2.1992862584706683,
      "grad_norm": 0.14091970026493073,
      "learning_rate": 7.15e-06,
      "loss": 0.6874,
      "step": 1714
    },
    {
      "epoch": 2.200569389309916,
      "grad_norm": 0.15919935703277588,
      "learning_rate": 7.1249999999999995e-06,
      "loss": 0.6372,
      "step": 1715
    },
    {
      "epoch": 2.201852520149164,
      "grad_norm": 0.14193716645240784,
      "learning_rate": 7.1e-06,
      "loss": 0.7364,
      "step": 1716
    },
    {
      "epoch": 2.2031356509884117,
      "grad_norm": 0.22038009762763977,
      "learning_rate": 7.075e-06,
      "loss": 0.652,
      "step": 1717
    },
    {
      "epoch": 2.2044187818276595,
      "grad_norm": 0.18721088767051697,
      "learning_rate": 7.049999999999999e-06,
      "loss": 0.6731,
      "step": 1718
    },
    {
      "epoch": 2.2057019126669073,
      "grad_norm": 0.193436861038208,
      "learning_rate": 7.025000000000001e-06,
      "loss": 0.6658,
      "step": 1719
    },
    {
      "epoch": 2.206985043506155,
      "grad_norm": 0.17044712603092194,
      "learning_rate": 7.000000000000001e-06,
      "loss": 0.689,
      "step": 1720
    },
    {
      "epoch": 2.208268174345403,
      "grad_norm": 0.2885238230228424,
      "learning_rate": 6.975000000000001e-06,
      "loss": 0.5995,
      "step": 1721
    },
    {
      "epoch": 2.2095513051846507,
      "grad_norm": 0.12533560395240784,
      "learning_rate": 6.950000000000001e-06,
      "loss": 0.6896,
      "step": 1722
    },
    {
      "epoch": 2.2108344360238985,
      "grad_norm": 0.15469686686992645,
      "learning_rate": 6.925000000000001e-06,
      "loss": 0.6482,
      "step": 1723
    },
    {
      "epoch": 2.212117566863146,
      "grad_norm": 0.12172360718250275,
      "learning_rate": 6.900000000000001e-06,
      "loss": 0.5874,
      "step": 1724
    },
    {
      "epoch": 2.2134006977023937,
      "grad_norm": 0.15161921083927155,
      "learning_rate": 6.875000000000001e-06,
      "loss": 0.6249,
      "step": 1725
    },
    {
      "epoch": 2.2146838285416415,
      "grad_norm": 0.17440353333950043,
      "learning_rate": 6.8500000000000005e-06,
      "loss": 0.5002,
      "step": 1726
    },
    {
      "epoch": 2.2159669593808893,
      "grad_norm": 0.13320526480674744,
      "learning_rate": 6.825000000000001e-06,
      "loss": 0.7504,
      "step": 1727
    },
    {
      "epoch": 2.217250090220137,
      "grad_norm": 0.13276419043540955,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.6493,
      "step": 1728
    },
    {
      "epoch": 2.218533221059385,
      "grad_norm": 0.13539136946201324,
      "learning_rate": 6.775000000000001e-06,
      "loss": 0.6764,
      "step": 1729
    },
    {
      "epoch": 2.2198163518986327,
      "grad_norm": 0.12818317115306854,
      "learning_rate": 6.750000000000001e-06,
      "loss": 0.6932,
      "step": 1730
    },
    {
      "epoch": 2.2210994827378805,
      "grad_norm": 0.11954977363348007,
      "learning_rate": 6.725000000000001e-06,
      "loss": 0.6451,
      "step": 1731
    },
    {
      "epoch": 2.2223826135771283,
      "grad_norm": 0.144534170627594,
      "learning_rate": 6.700000000000001e-06,
      "loss": 0.7819,
      "step": 1732
    },
    {
      "epoch": 2.223665744416376,
      "grad_norm": 0.14043539762496948,
      "learning_rate": 6.6750000000000005e-06,
      "loss": 0.7258,
      "step": 1733
    },
    {
      "epoch": 2.224948875255624,
      "grad_norm": 0.18055632710456848,
      "learning_rate": 6.650000000000001e-06,
      "loss": 0.7105,
      "step": 1734
    },
    {
      "epoch": 2.2262320060948717,
      "grad_norm": 0.14630240201950073,
      "learning_rate": 6.625000000000001e-06,
      "loss": 0.6285,
      "step": 1735
    },
    {
      "epoch": 2.227515136934119,
      "grad_norm": 0.19263723492622375,
      "learning_rate": 6.6e-06,
      "loss": 0.7665,
      "step": 1736
    },
    {
      "epoch": 2.228798267773367,
      "grad_norm": 0.20304565131664276,
      "learning_rate": 6.5750000000000006e-06,
      "loss": 0.7909,
      "step": 1737
    },
    {
      "epoch": 2.2300813986126147,
      "grad_norm": 0.1459382176399231,
      "learning_rate": 6.550000000000001e-06,
      "loss": 0.6623,
      "step": 1738
    },
    {
      "epoch": 2.2313645294518625,
      "grad_norm": 0.26151278614997864,
      "learning_rate": 6.525e-06,
      "loss": 0.6127,
      "step": 1739
    },
    {
      "epoch": 2.2326476602911103,
      "grad_norm": 0.15971115231513977,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.6699,
      "step": 1740
    },
    {
      "epoch": 2.233930791130358,
      "grad_norm": 0.1514868587255478,
      "learning_rate": 6.475000000000001e-06,
      "loss": 0.6176,
      "step": 1741
    },
    {
      "epoch": 2.235213921969606,
      "grad_norm": 0.13898637890815735,
      "learning_rate": 6.45e-06,
      "loss": 0.5719,
      "step": 1742
    },
    {
      "epoch": 2.2364970528088537,
      "grad_norm": 0.17938609421253204,
      "learning_rate": 6.425e-06,
      "loss": 0.7357,
      "step": 1743
    },
    {
      "epoch": 2.2377801836481015,
      "grad_norm": 0.16142819821834564,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.6644,
      "step": 1744
    },
    {
      "epoch": 2.2390633144873493,
      "grad_norm": 0.1518702358007431,
      "learning_rate": 6.375000000000001e-06,
      "loss": 0.7004,
      "step": 1745
    },
    {
      "epoch": 2.240346445326597,
      "grad_norm": 0.16036461293697357,
      "learning_rate": 6.35e-06,
      "loss": 0.6415,
      "step": 1746
    },
    {
      "epoch": 2.2416295761658445,
      "grad_norm": 0.15337897837162018,
      "learning_rate": 6.3250000000000004e-06,
      "loss": 0.6991,
      "step": 1747
    },
    {
      "epoch": 2.2429127070050923,
      "grad_norm": 0.13635998964309692,
      "learning_rate": 6.300000000000001e-06,
      "loss": 0.6871,
      "step": 1748
    },
    {
      "epoch": 2.24419583784434,
      "grad_norm": 0.1608695089817047,
      "learning_rate": 6.275e-06,
      "loss": 0.6344,
      "step": 1749
    },
    {
      "epoch": 2.245478968683588,
      "grad_norm": 0.18258242309093475,
      "learning_rate": 6.25e-06,
      "loss": 0.6682,
      "step": 1750
    },
    {
      "epoch": 2.2467620995228357,
      "grad_norm": 0.1698656529188156,
      "learning_rate": 6.2250000000000005e-06,
      "loss": 0.6107,
      "step": 1751
    },
    {
      "epoch": 2.2480452303620835,
      "grad_norm": 0.1514923870563507,
      "learning_rate": 6.2e-06,
      "loss": 0.6708,
      "step": 1752
    },
    {
      "epoch": 2.2493283612013313,
      "grad_norm": 0.20906352996826172,
      "learning_rate": 6.175e-06,
      "loss": 0.7143,
      "step": 1753
    },
    {
      "epoch": 2.250611492040579,
      "grad_norm": 0.15213285386562347,
      "learning_rate": 6.15e-06,
      "loss": 0.6928,
      "step": 1754
    },
    {
      "epoch": 2.251894622879827,
      "grad_norm": 0.14127208292484283,
      "learning_rate": 6.125e-06,
      "loss": 0.6588,
      "step": 1755
    },
    {
      "epoch": 2.2531777537190747,
      "grad_norm": 0.14866508543491364,
      "learning_rate": 6.1e-06,
      "loss": 0.7408,
      "step": 1756
    },
    {
      "epoch": 2.2544608845583225,
      "grad_norm": 0.1756955087184906,
      "learning_rate": 6.075e-06,
      "loss": 0.6266,
      "step": 1757
    },
    {
      "epoch": 2.2557440153975703,
      "grad_norm": 0.1728198379278183,
      "learning_rate": 6.0500000000000005e-06,
      "loss": 0.6327,
      "step": 1758
    },
    {
      "epoch": 2.2570271462368177,
      "grad_norm": 0.14398327469825745,
      "learning_rate": 6.025e-06,
      "loss": 0.6474,
      "step": 1759
    },
    {
      "epoch": 2.2583102770760655,
      "grad_norm": 0.13076895475387573,
      "learning_rate": 6e-06,
      "loss": 0.7688,
      "step": 1760
    },
    {
      "epoch": 2.2595934079153133,
      "grad_norm": 0.19304291903972626,
      "learning_rate": 5.975e-06,
      "loss": 0.6385,
      "step": 1761
    },
    {
      "epoch": 2.260876538754561,
      "grad_norm": 0.11369101703166962,
      "learning_rate": 5.95e-06,
      "loss": 0.6758,
      "step": 1762
    },
    {
      "epoch": 2.262159669593809,
      "grad_norm": 0.1310070902109146,
      "learning_rate": 5.925e-06,
      "loss": 0.675,
      "step": 1763
    },
    {
      "epoch": 2.2634428004330567,
      "grad_norm": 0.15209239721298218,
      "learning_rate": 5.9e-06,
      "loss": 0.6554,
      "step": 1764
    },
    {
      "epoch": 2.2647259312723045,
      "grad_norm": 0.18606530129909515,
      "learning_rate": 5.875e-06,
      "loss": 0.5265,
      "step": 1765
    },
    {
      "epoch": 2.2660090621115523,
      "grad_norm": 0.1213434711098671,
      "learning_rate": 5.850000000000001e-06,
      "loss": 0.5974,
      "step": 1766
    },
    {
      "epoch": 2.2672921929508,
      "grad_norm": 0.1257883757352829,
      "learning_rate": 5.825000000000001e-06,
      "loss": 0.617,
      "step": 1767
    },
    {
      "epoch": 2.268575323790048,
      "grad_norm": 0.1754256933927536,
      "learning_rate": 5.8e-06,
      "loss": 0.6024,
      "step": 1768
    },
    {
      "epoch": 2.2698584546292953,
      "grad_norm": 0.22791916131973267,
      "learning_rate": 5.775000000000001e-06,
      "loss": 0.6333,
      "step": 1769
    },
    {
      "epoch": 2.271141585468543,
      "grad_norm": 0.15326513350009918,
      "learning_rate": 5.750000000000001e-06,
      "loss": 0.6391,
      "step": 1770
    },
    {
      "epoch": 2.272424716307791,
      "grad_norm": 0.16976433992385864,
      "learning_rate": 5.725e-06,
      "loss": 0.6363,
      "step": 1771
    },
    {
      "epoch": 2.2737078471470387,
      "grad_norm": 0.14141680300235748,
      "learning_rate": 5.7000000000000005e-06,
      "loss": 0.6854,
      "step": 1772
    },
    {
      "epoch": 2.2749909779862865,
      "grad_norm": 0.20347145199775696,
      "learning_rate": 5.675000000000001e-06,
      "loss": 0.6987,
      "step": 1773
    },
    {
      "epoch": 2.2762741088255343,
      "grad_norm": 0.16503895819187164,
      "learning_rate": 5.65e-06,
      "loss": 0.7151,
      "step": 1774
    },
    {
      "epoch": 2.277557239664782,
      "grad_norm": 0.15288709104061127,
      "learning_rate": 5.625e-06,
      "loss": 0.7636,
      "step": 1775
    },
    {
      "epoch": 2.27884037050403,
      "grad_norm": 0.14346033334732056,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.6746,
      "step": 1776
    },
    {
      "epoch": 2.2801235013432777,
      "grad_norm": 0.2069840431213379,
      "learning_rate": 5.575e-06,
      "loss": 0.7442,
      "step": 1777
    },
    {
      "epoch": 2.2814066321825255,
      "grad_norm": 0.18324162065982819,
      "learning_rate": 5.55e-06,
      "loss": 0.6384,
      "step": 1778
    },
    {
      "epoch": 2.2826897630217733,
      "grad_norm": 0.12865564227104187,
      "learning_rate": 5.5250000000000005e-06,
      "loss": 0.7159,
      "step": 1779
    },
    {
      "epoch": 2.283972893861021,
      "grad_norm": 0.17469723522663116,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.6385,
      "step": 1780
    },
    {
      "epoch": 2.285256024700269,
      "grad_norm": 0.17101655900478363,
      "learning_rate": 5.475e-06,
      "loss": 0.7047,
      "step": 1781
    },
    {
      "epoch": 2.2865391555395163,
      "grad_norm": 0.16254979372024536,
      "learning_rate": 5.45e-06,
      "loss": 0.6696,
      "step": 1782
    },
    {
      "epoch": 2.287822286378764,
      "grad_norm": 0.2329155057668686,
      "learning_rate": 5.4250000000000006e-06,
      "loss": 0.7225,
      "step": 1783
    },
    {
      "epoch": 2.289105417218012,
      "grad_norm": 0.17540696263313293,
      "learning_rate": 5.4e-06,
      "loss": 0.6482,
      "step": 1784
    },
    {
      "epoch": 2.2903885480572597,
      "grad_norm": 0.15673121809959412,
      "learning_rate": 5.375e-06,
      "loss": 0.6544,
      "step": 1785
    },
    {
      "epoch": 2.2916716788965075,
      "grad_norm": 0.15571188926696777,
      "learning_rate": 5.3500000000000004e-06,
      "loss": 0.767,
      "step": 1786
    },
    {
      "epoch": 2.2929548097357553,
      "grad_norm": 0.1779564768075943,
      "learning_rate": 5.325e-06,
      "loss": 0.8099,
      "step": 1787
    },
    {
      "epoch": 2.294237940575003,
      "grad_norm": 0.22156600654125214,
      "learning_rate": 5.3e-06,
      "loss": 0.7425,
      "step": 1788
    },
    {
      "epoch": 2.295521071414251,
      "grad_norm": 0.1420728713274002,
      "learning_rate": 5.275e-06,
      "loss": 0.6642,
      "step": 1789
    },
    {
      "epoch": 2.2968042022534987,
      "grad_norm": 0.17445795238018036,
      "learning_rate": 5.25e-06,
      "loss": 0.6666,
      "step": 1790
    },
    {
      "epoch": 2.2980873330927465,
      "grad_norm": 0.12195762246847153,
      "learning_rate": 5.225e-06,
      "loss": 0.6757,
      "step": 1791
    },
    {
      "epoch": 2.299370463931994,
      "grad_norm": 0.1290019452571869,
      "learning_rate": 5.2e-06,
      "loss": 0.6832,
      "step": 1792
    },
    {
      "epoch": 2.3006535947712417,
      "grad_norm": 0.1515694111585617,
      "learning_rate": 5.175e-06,
      "loss": 0.7097,
      "step": 1793
    },
    {
      "epoch": 2.3019367256104895,
      "grad_norm": 0.14995551109313965,
      "learning_rate": 5.15e-06,
      "loss": 0.731,
      "step": 1794
    },
    {
      "epoch": 2.3032198564497373,
      "grad_norm": 0.2452741116285324,
      "learning_rate": 5.125e-06,
      "loss": 0.7205,
      "step": 1795
    },
    {
      "epoch": 2.304502987288985,
      "grad_norm": 0.20354309678077698,
      "learning_rate": 5.1e-06,
      "loss": 0.6232,
      "step": 1796
    },
    {
      "epoch": 2.305786118128233,
      "grad_norm": 0.1639031022787094,
      "learning_rate": 5.0750000000000005e-06,
      "loss": 0.701,
      "step": 1797
    },
    {
      "epoch": 2.3070692489674807,
      "grad_norm": 0.2124990075826645,
      "learning_rate": 5.050000000000001e-06,
      "loss": 0.654,
      "step": 1798
    },
    {
      "epoch": 2.3083523798067285,
      "grad_norm": 0.13865113258361816,
      "learning_rate": 5.025e-06,
      "loss": 0.6917,
      "step": 1799
    },
    {
      "epoch": 2.3096355106459763,
      "grad_norm": 0.13579049706459045,
      "learning_rate": 5e-06,
      "loss": 0.7484,
      "step": 1800
    },
    {
      "epoch": 2.310918641485224,
      "grad_norm": 0.24273373186588287,
      "learning_rate": 4.975000000000001e-06,
      "loss": 0.7272,
      "step": 1801
    },
    {
      "epoch": 2.312201772324472,
      "grad_norm": 0.2409195601940155,
      "learning_rate": 4.950000000000001e-06,
      "loss": 0.6591,
      "step": 1802
    },
    {
      "epoch": 2.3134849031637197,
      "grad_norm": 0.1992218792438507,
      "learning_rate": 4.925e-06,
      "loss": 0.7022,
      "step": 1803
    },
    {
      "epoch": 2.3147680340029675,
      "grad_norm": 0.13271629810333252,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 0.6376,
      "step": 1804
    },
    {
      "epoch": 2.316051164842215,
      "grad_norm": 0.1718410700559616,
      "learning_rate": 4.875000000000001e-06,
      "loss": 0.6844,
      "step": 1805
    },
    {
      "epoch": 2.3173342956814627,
      "grad_norm": 0.19865791499614716,
      "learning_rate": 4.85e-06,
      "loss": 0.7659,
      "step": 1806
    },
    {
      "epoch": 2.3186174265207105,
      "grad_norm": 0.36948734521865845,
      "learning_rate": 4.825e-06,
      "loss": 0.6294,
      "step": 1807
    },
    {
      "epoch": 2.3199005573599583,
      "grad_norm": 0.16237518191337585,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.6844,
      "step": 1808
    },
    {
      "epoch": 2.321183688199206,
      "grad_norm": 0.15354694426059723,
      "learning_rate": 4.775e-06,
      "loss": 0.6658,
      "step": 1809
    },
    {
      "epoch": 2.322466819038454,
      "grad_norm": 0.320901095867157,
      "learning_rate": 4.75e-06,
      "loss": 0.6105,
      "step": 1810
    },
    {
      "epoch": 2.3237499498777017,
      "grad_norm": 0.2136460840702057,
      "learning_rate": 4.7250000000000005e-06,
      "loss": 0.5697,
      "step": 1811
    },
    {
      "epoch": 2.3250330807169495,
      "grad_norm": 0.20062057673931122,
      "learning_rate": 4.7e-06,
      "loss": 0.691,
      "step": 1812
    },
    {
      "epoch": 2.3263162115561973,
      "grad_norm": 0.20206096768379211,
      "learning_rate": 4.675e-06,
      "loss": 0.6408,
      "step": 1813
    },
    {
      "epoch": 2.3275993423954446,
      "grad_norm": 0.15030571818351746,
      "learning_rate": 4.65e-06,
      "loss": 0.6817,
      "step": 1814
    },
    {
      "epoch": 2.3288824732346924,
      "grad_norm": 0.21458406746387482,
      "learning_rate": 4.625e-06,
      "loss": 0.6386,
      "step": 1815
    },
    {
      "epoch": 2.3301656040739402,
      "grad_norm": 0.36531540751457214,
      "learning_rate": 4.6e-06,
      "loss": 0.7589,
      "step": 1816
    },
    {
      "epoch": 2.331448734913188,
      "grad_norm": 0.20997947454452515,
      "learning_rate": 4.575e-06,
      "loss": 0.6121,
      "step": 1817
    },
    {
      "epoch": 2.332731865752436,
      "grad_norm": 0.15924055874347687,
      "learning_rate": 4.5500000000000005e-06,
      "loss": 0.6445,
      "step": 1818
    },
    {
      "epoch": 2.3340149965916837,
      "grad_norm": 0.18460945785045624,
      "learning_rate": 4.525e-06,
      "loss": 0.6006,
      "step": 1819
    },
    {
      "epoch": 2.3352981274309315,
      "grad_norm": 0.3289165496826172,
      "learning_rate": 4.5e-06,
      "loss": 0.7165,
      "step": 1820
    },
    {
      "epoch": 2.3365812582701793,
      "grad_norm": 0.14990635216236115,
      "learning_rate": 4.475e-06,
      "loss": 0.7568,
      "step": 1821
    },
    {
      "epoch": 2.337864389109427,
      "grad_norm": 0.15205344557762146,
      "learning_rate": 4.45e-06,
      "loss": 0.7485,
      "step": 1822
    },
    {
      "epoch": 2.339147519948675,
      "grad_norm": 0.14835873246192932,
      "learning_rate": 4.425e-06,
      "loss": 0.6629,
      "step": 1823
    },
    {
      "epoch": 2.3404306507879227,
      "grad_norm": 0.14888514578342438,
      "learning_rate": 4.4e-06,
      "loss": 0.6255,
      "step": 1824
    },
    {
      "epoch": 2.3417137816271705,
      "grad_norm": 0.13398034870624542,
      "learning_rate": 4.375e-06,
      "loss": 0.7379,
      "step": 1825
    },
    {
      "epoch": 2.3429969124664183,
      "grad_norm": 0.11306151002645493,
      "learning_rate": 4.35e-06,
      "loss": 0.6246,
      "step": 1826
    },
    {
      "epoch": 2.3442800433056656,
      "grad_norm": 0.14197863638401031,
      "learning_rate": 4.325e-06,
      "loss": 0.6836,
      "step": 1827
    },
    {
      "epoch": 2.3455631741449134,
      "grad_norm": 0.2663162648677826,
      "learning_rate": 4.2999999999999995e-06,
      "loss": 0.565,
      "step": 1828
    },
    {
      "epoch": 2.3468463049841612,
      "grad_norm": 0.15789471566677094,
      "learning_rate": 4.2750000000000006e-06,
      "loss": 0.6511,
      "step": 1829
    },
    {
      "epoch": 2.348129435823409,
      "grad_norm": 0.1293058544397354,
      "learning_rate": 4.250000000000001e-06,
      "loss": 0.651,
      "step": 1830
    },
    {
      "epoch": 2.349412566662657,
      "grad_norm": 0.15461447834968567,
      "learning_rate": 4.225e-06,
      "loss": 0.7026,
      "step": 1831
    },
    {
      "epoch": 2.3506956975019047,
      "grad_norm": 0.1547372192144394,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.592,
      "step": 1832
    },
    {
      "epoch": 2.3519788283411525,
      "grad_norm": 0.13594333827495575,
      "learning_rate": 4.175000000000001e-06,
      "loss": 0.6298,
      "step": 1833
    },
    {
      "epoch": 2.3532619591804003,
      "grad_norm": 0.19110356271266937,
      "learning_rate": 4.15e-06,
      "loss": 0.7446,
      "step": 1834
    },
    {
      "epoch": 2.354545090019648,
      "grad_norm": 0.12567587196826935,
      "learning_rate": 4.125e-06,
      "loss": 0.7211,
      "step": 1835
    },
    {
      "epoch": 2.355828220858896,
      "grad_norm": 0.1582501381635666,
      "learning_rate": 4.1000000000000006e-06,
      "loss": 0.7273,
      "step": 1836
    },
    {
      "epoch": 2.3571113516981432,
      "grad_norm": 0.24335543811321259,
      "learning_rate": 4.075e-06,
      "loss": 0.6487,
      "step": 1837
    },
    {
      "epoch": 2.358394482537391,
      "grad_norm": 0.12796013057231903,
      "learning_rate": 4.05e-06,
      "loss": 0.6995,
      "step": 1838
    },
    {
      "epoch": 2.359677613376639,
      "grad_norm": 0.16038630902767181,
      "learning_rate": 4.0250000000000004e-06,
      "loss": 0.6995,
      "step": 1839
    },
    {
      "epoch": 2.3609607442158866,
      "grad_norm": 0.17377327382564545,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.7509,
      "step": 1840
    },
    {
      "epoch": 2.3622438750551344,
      "grad_norm": 0.14089252054691315,
      "learning_rate": 3.975e-06,
      "loss": 0.7472,
      "step": 1841
    },
    {
      "epoch": 2.3635270058943822,
      "grad_norm": 0.1695878952741623,
      "learning_rate": 3.95e-06,
      "loss": 0.6896,
      "step": 1842
    },
    {
      "epoch": 2.36481013673363,
      "grad_norm": 0.19289609789848328,
      "learning_rate": 3.9250000000000005e-06,
      "loss": 0.5972,
      "step": 1843
    },
    {
      "epoch": 2.366093267572878,
      "grad_norm": 0.17196987569332123,
      "learning_rate": 3.9e-06,
      "loss": 0.6352,
      "step": 1844
    },
    {
      "epoch": 2.3673763984121257,
      "grad_norm": 0.16529642045497894,
      "learning_rate": 3.875e-06,
      "loss": 0.5981,
      "step": 1845
    },
    {
      "epoch": 2.3686595292513735,
      "grad_norm": 0.2297869622707367,
      "learning_rate": 3.85e-06,
      "loss": 0.7363,
      "step": 1846
    },
    {
      "epoch": 2.3699426600906213,
      "grad_norm": 0.12826687097549438,
      "learning_rate": 3.825e-06,
      "loss": 0.552,
      "step": 1847
    },
    {
      "epoch": 2.371225790929869,
      "grad_norm": 0.2224760353565216,
      "learning_rate": 3.8e-06,
      "loss": 0.669,
      "step": 1848
    },
    {
      "epoch": 2.372508921769117,
      "grad_norm": 0.16175910830497742,
      "learning_rate": 3.775e-06,
      "loss": 0.7258,
      "step": 1849
    },
    {
      "epoch": 2.3737920526083642,
      "grad_norm": 0.12144355475902557,
      "learning_rate": 3.75e-06,
      "loss": 0.5795,
      "step": 1850
    },
    {
      "epoch": 2.375075183447612,
      "grad_norm": 0.16625931859016418,
      "learning_rate": 3.725e-06,
      "loss": 0.6291,
      "step": 1851
    },
    {
      "epoch": 2.37635831428686,
      "grad_norm": 0.14619119465351105,
      "learning_rate": 3.7e-06,
      "loss": 0.6319,
      "step": 1852
    },
    {
      "epoch": 2.3776414451261076,
      "grad_norm": 0.18616823852062225,
      "learning_rate": 3.675e-06,
      "loss": 0.6574,
      "step": 1853
    },
    {
      "epoch": 2.3789245759653554,
      "grad_norm": 0.19217754900455475,
      "learning_rate": 3.6499999999999998e-06,
      "loss": 0.6376,
      "step": 1854
    },
    {
      "epoch": 2.3802077068046033,
      "grad_norm": 0.14618663489818573,
      "learning_rate": 3.625e-06,
      "loss": 0.6157,
      "step": 1855
    },
    {
      "epoch": 2.381490837643851,
      "grad_norm": 0.15706594288349152,
      "learning_rate": 3.6e-06,
      "loss": 0.655,
      "step": 1856
    },
    {
      "epoch": 2.382773968483099,
      "grad_norm": 0.16629637777805328,
      "learning_rate": 3.575e-06,
      "loss": 0.6465,
      "step": 1857
    },
    {
      "epoch": 2.3840570993223467,
      "grad_norm": 0.12229061871767044,
      "learning_rate": 3.55e-06,
      "loss": 0.6697,
      "step": 1858
    },
    {
      "epoch": 2.3853402301615945,
      "grad_norm": 0.13999316096305847,
      "learning_rate": 3.5249999999999997e-06,
      "loss": 0.6304,
      "step": 1859
    },
    {
      "epoch": 2.386623361000842,
      "grad_norm": 0.1271730363368988,
      "learning_rate": 3.5000000000000004e-06,
      "loss": 0.7514,
      "step": 1860
    },
    {
      "epoch": 2.3879064918400896,
      "grad_norm": 0.18263642489910126,
      "learning_rate": 3.4750000000000006e-06,
      "loss": 0.6808,
      "step": 1861
    },
    {
      "epoch": 2.3891896226793374,
      "grad_norm": 0.1465403139591217,
      "learning_rate": 3.4500000000000004e-06,
      "loss": 0.6759,
      "step": 1862
    },
    {
      "epoch": 2.3904727535185852,
      "grad_norm": 0.18245361745357513,
      "learning_rate": 3.4250000000000002e-06,
      "loss": 0.6769,
      "step": 1863
    },
    {
      "epoch": 2.391755884357833,
      "grad_norm": 0.16645410656929016,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.5779,
      "step": 1864
    },
    {
      "epoch": 2.393039015197081,
      "grad_norm": 0.21561074256896973,
      "learning_rate": 3.3750000000000003e-06,
      "loss": 0.6557,
      "step": 1865
    },
    {
      "epoch": 2.3943221460363286,
      "grad_norm": 0.15144115686416626,
      "learning_rate": 3.3500000000000005e-06,
      "loss": 0.778,
      "step": 1866
    },
    {
      "epoch": 2.3956052768755765,
      "grad_norm": 0.1890507936477661,
      "learning_rate": 3.3250000000000004e-06,
      "loss": 0.6539,
      "step": 1867
    },
    {
      "epoch": 2.3968884077148243,
      "grad_norm": 0.21001359820365906,
      "learning_rate": 3.3e-06,
      "loss": 0.6438,
      "step": 1868
    },
    {
      "epoch": 2.398171538554072,
      "grad_norm": 0.18896803259849548,
      "learning_rate": 3.2750000000000004e-06,
      "loss": 0.6833,
      "step": 1869
    },
    {
      "epoch": 2.39945466939332,
      "grad_norm": 0.12959522008895874,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.7507,
      "step": 1870
    },
    {
      "epoch": 2.4007378002325677,
      "grad_norm": 0.14917847514152527,
      "learning_rate": 3.225e-06,
      "loss": 0.6294,
      "step": 1871
    },
    {
      "epoch": 2.4020209310718155,
      "grad_norm": 0.13187511265277863,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.6521,
      "step": 1872
    },
    {
      "epoch": 2.403304061911063,
      "grad_norm": 0.22098731994628906,
      "learning_rate": 3.175e-06,
      "loss": 0.6875,
      "step": 1873
    },
    {
      "epoch": 2.4045871927503106,
      "grad_norm": 0.23417702317237854,
      "learning_rate": 3.1500000000000003e-06,
      "loss": 0.5511,
      "step": 1874
    },
    {
      "epoch": 2.4058703235895584,
      "grad_norm": 0.1343628615140915,
      "learning_rate": 3.125e-06,
      "loss": 0.7236,
      "step": 1875
    },
    {
      "epoch": 2.4071534544288062,
      "grad_norm": 0.27076542377471924,
      "learning_rate": 3.1e-06,
      "loss": 0.6547,
      "step": 1876
    },
    {
      "epoch": 2.408436585268054,
      "grad_norm": 0.1615552008152008,
      "learning_rate": 3.075e-06,
      "loss": 0.6922,
      "step": 1877
    },
    {
      "epoch": 2.409719716107302,
      "grad_norm": 0.13084520399570465,
      "learning_rate": 3.05e-06,
      "loss": 0.6826,
      "step": 1878
    },
    {
      "epoch": 2.4110028469465496,
      "grad_norm": 0.13955563306808472,
      "learning_rate": 3.0250000000000003e-06,
      "loss": 0.6913,
      "step": 1879
    },
    {
      "epoch": 2.4122859777857975,
      "grad_norm": 0.1469963937997818,
      "learning_rate": 3e-06,
      "loss": 0.7598,
      "step": 1880
    },
    {
      "epoch": 2.4135691086250453,
      "grad_norm": 0.22338779270648956,
      "learning_rate": 2.975e-06,
      "loss": 0.6145,
      "step": 1881
    },
    {
      "epoch": 2.414852239464293,
      "grad_norm": 0.19863590598106384,
      "learning_rate": 2.95e-06,
      "loss": 0.6727,
      "step": 1882
    },
    {
      "epoch": 2.4161353703035404,
      "grad_norm": 0.1779351532459259,
      "learning_rate": 2.9250000000000004e-06,
      "loss": 0.6556,
      "step": 1883
    },
    {
      "epoch": 2.4174185011427882,
      "grad_norm": 0.15160447359085083,
      "learning_rate": 2.9e-06,
      "loss": 0.6584,
      "step": 1884
    },
    {
      "epoch": 2.418701631982036,
      "grad_norm": 0.16567489504814148,
      "learning_rate": 2.8750000000000004e-06,
      "loss": 0.7227,
      "step": 1885
    },
    {
      "epoch": 2.419984762821284,
      "grad_norm": 0.1918822079896927,
      "learning_rate": 2.8500000000000002e-06,
      "loss": 0.7061,
      "step": 1886
    },
    {
      "epoch": 2.4212678936605316,
      "grad_norm": 0.17254500091075897,
      "learning_rate": 2.825e-06,
      "loss": 0.6857,
      "step": 1887
    },
    {
      "epoch": 2.4225510244997794,
      "grad_norm": 0.16150803864002228,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.6484,
      "step": 1888
    },
    {
      "epoch": 2.4238341553390272,
      "grad_norm": 0.11683553457260132,
      "learning_rate": 2.775e-06,
      "loss": 0.6445,
      "step": 1889
    },
    {
      "epoch": 2.425117286178275,
      "grad_norm": 0.2578047513961792,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.628,
      "step": 1890
    },
    {
      "epoch": 2.426400417017523,
      "grad_norm": 0.1274745613336563,
      "learning_rate": 2.725e-06,
      "loss": 0.7083,
      "step": 1891
    },
    {
      "epoch": 2.4276835478567707,
      "grad_norm": 0.11964628845453262,
      "learning_rate": 2.7e-06,
      "loss": 0.6475,
      "step": 1892
    },
    {
      "epoch": 2.4289666786960185,
      "grad_norm": 0.20247456431388855,
      "learning_rate": 2.6750000000000002e-06,
      "loss": 0.6906,
      "step": 1893
    },
    {
      "epoch": 2.4302498095352663,
      "grad_norm": 0.12243557721376419,
      "learning_rate": 2.65e-06,
      "loss": 0.661,
      "step": 1894
    },
    {
      "epoch": 2.431532940374514,
      "grad_norm": 0.20099373161792755,
      "learning_rate": 2.625e-06,
      "loss": 0.7208,
      "step": 1895
    },
    {
      "epoch": 2.4328160712137614,
      "grad_norm": 0.18498243391513824,
      "learning_rate": 2.6e-06,
      "loss": 0.7024,
      "step": 1896
    },
    {
      "epoch": 2.4340992020530092,
      "grad_norm": 0.13421030342578888,
      "learning_rate": 2.575e-06,
      "loss": 0.7228,
      "step": 1897
    },
    {
      "epoch": 2.435382332892257,
      "grad_norm": 0.14675818383693695,
      "learning_rate": 2.55e-06,
      "loss": 0.6144,
      "step": 1898
    },
    {
      "epoch": 2.436665463731505,
      "grad_norm": 0.11477550119161606,
      "learning_rate": 2.5250000000000004e-06,
      "loss": 0.6578,
      "step": 1899
    },
    {
      "epoch": 2.4379485945707526,
      "grad_norm": 0.14763018488883972,
      "learning_rate": 2.5e-06,
      "loss": 0.6612,
      "step": 1900
    },
    {
      "epoch": 2.4392317254100004,
      "grad_norm": 0.14850638806819916,
      "learning_rate": 2.4750000000000004e-06,
      "loss": 0.6331,
      "step": 1901
    },
    {
      "epoch": 2.4405148562492482,
      "grad_norm": 0.16140273213386536,
      "learning_rate": 2.4500000000000003e-06,
      "loss": 0.7171,
      "step": 1902
    },
    {
      "epoch": 2.441797987088496,
      "grad_norm": 0.16006286442279816,
      "learning_rate": 2.425e-06,
      "loss": 0.6971,
      "step": 1903
    },
    {
      "epoch": 2.443081117927744,
      "grad_norm": 0.15166538953781128,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.7377,
      "step": 1904
    },
    {
      "epoch": 2.444364248766991,
      "grad_norm": 0.21980290114879608,
      "learning_rate": 2.375e-06,
      "loss": 0.6612,
      "step": 1905
    },
    {
      "epoch": 2.445647379606239,
      "grad_norm": 0.1631053239107132,
      "learning_rate": 2.35e-06,
      "loss": 0.5741,
      "step": 1906
    },
    {
      "epoch": 2.446930510445487,
      "grad_norm": 0.1300109177827835,
      "learning_rate": 2.325e-06,
      "loss": 0.6286,
      "step": 1907
    },
    {
      "epoch": 2.4482136412847346,
      "grad_norm": 0.14739301800727844,
      "learning_rate": 2.3e-06,
      "loss": 0.695,
      "step": 1908
    },
    {
      "epoch": 2.4494967721239824,
      "grad_norm": 0.13381710648536682,
      "learning_rate": 2.2750000000000002e-06,
      "loss": 0.6737,
      "step": 1909
    },
    {
      "epoch": 2.4507799029632302,
      "grad_norm": 0.11684217303991318,
      "learning_rate": 2.25e-06,
      "loss": 0.6168,
      "step": 1910
    },
    {
      "epoch": 2.452063033802478,
      "grad_norm": 0.204594686627388,
      "learning_rate": 2.225e-06,
      "loss": 0.7289,
      "step": 1911
    },
    {
      "epoch": 2.453346164641726,
      "grad_norm": 0.13649746775627136,
      "learning_rate": 2.2e-06,
      "loss": 0.6781,
      "step": 1912
    },
    {
      "epoch": 2.4546292954809736,
      "grad_norm": 0.16303804516792297,
      "learning_rate": 2.175e-06,
      "loss": 0.667,
      "step": 1913
    },
    {
      "epoch": 2.4559124263202214,
      "grad_norm": 0.17752477526664734,
      "learning_rate": 2.1499999999999997e-06,
      "loss": 0.6896,
      "step": 1914
    },
    {
      "epoch": 2.4571955571594692,
      "grad_norm": 0.16628795862197876,
      "learning_rate": 2.1250000000000004e-06,
      "loss": 0.5952,
      "step": 1915
    },
    {
      "epoch": 2.458478687998717,
      "grad_norm": 0.15011388063430786,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.812,
      "step": 1916
    },
    {
      "epoch": 2.459761818837965,
      "grad_norm": 0.13481688499450684,
      "learning_rate": 2.075e-06,
      "loss": 0.7192,
      "step": 1917
    },
    {
      "epoch": 2.461044949677212,
      "grad_norm": 0.17045043408870697,
      "learning_rate": 2.0500000000000003e-06,
      "loss": 0.5998,
      "step": 1918
    },
    {
      "epoch": 2.46232808051646,
      "grad_norm": 0.13013005256652832,
      "learning_rate": 2.025e-06,
      "loss": 0.5981,
      "step": 1919
    },
    {
      "epoch": 2.463611211355708,
      "grad_norm": 0.14957274496555328,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.5284,
      "step": 1920
    },
    {
      "epoch": 2.4648943421949556,
      "grad_norm": 0.17074356973171234,
      "learning_rate": 1.975e-06,
      "loss": 0.6578,
      "step": 1921
    },
    {
      "epoch": 2.4661774730342034,
      "grad_norm": 0.14697015285491943,
      "learning_rate": 1.95e-06,
      "loss": 0.7056,
      "step": 1922
    },
    {
      "epoch": 2.4674606038734512,
      "grad_norm": 0.1444215029478073,
      "learning_rate": 1.925e-06,
      "loss": 0.7171,
      "step": 1923
    },
    {
      "epoch": 2.468743734712699,
      "grad_norm": 0.18043532967567444,
      "learning_rate": 1.9e-06,
      "loss": 0.6667,
      "step": 1924
    },
    {
      "epoch": 2.470026865551947,
      "grad_norm": 0.14275479316711426,
      "learning_rate": 1.875e-06,
      "loss": 0.6797,
      "step": 1925
    },
    {
      "epoch": 2.4713099963911946,
      "grad_norm": 0.12070459127426147,
      "learning_rate": 1.85e-06,
      "loss": 0.7065,
      "step": 1926
    },
    {
      "epoch": 2.4725931272304424,
      "grad_norm": 0.1490575224161148,
      "learning_rate": 1.8249999999999999e-06,
      "loss": 0.57,
      "step": 1927
    },
    {
      "epoch": 2.47387625806969,
      "grad_norm": 0.15245544910430908,
      "learning_rate": 1.8e-06,
      "loss": 0.7561,
      "step": 1928
    },
    {
      "epoch": 2.4751593889089376,
      "grad_norm": 0.12037181109189987,
      "learning_rate": 1.775e-06,
      "loss": 0.631,
      "step": 1929
    },
    {
      "epoch": 2.4764425197481854,
      "grad_norm": 0.14657901227474213,
      "learning_rate": 1.7500000000000002e-06,
      "loss": 0.5922,
      "step": 1930
    },
    {
      "epoch": 2.477725650587433,
      "grad_norm": 0.14087170362472534,
      "learning_rate": 1.7250000000000002e-06,
      "loss": 0.6448,
      "step": 1931
    },
    {
      "epoch": 2.479008781426681,
      "grad_norm": 0.1591077446937561,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 0.7789,
      "step": 1932
    },
    {
      "epoch": 2.480291912265929,
      "grad_norm": 0.12104818224906921,
      "learning_rate": 1.6750000000000003e-06,
      "loss": 0.6817,
      "step": 1933
    },
    {
      "epoch": 2.4815750431051766,
      "grad_norm": 0.17551586031913757,
      "learning_rate": 1.65e-06,
      "loss": 0.7015,
      "step": 1934
    },
    {
      "epoch": 2.4828581739444244,
      "grad_norm": 0.1548716276884079,
      "learning_rate": 1.6250000000000001e-06,
      "loss": 0.6708,
      "step": 1935
    },
    {
      "epoch": 2.4841413047836722,
      "grad_norm": 0.16079238057136536,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.6582,
      "step": 1936
    },
    {
      "epoch": 2.48542443562292,
      "grad_norm": 0.1451188176870346,
      "learning_rate": 1.5750000000000002e-06,
      "loss": 0.6439,
      "step": 1937
    },
    {
      "epoch": 2.486707566462168,
      "grad_norm": 0.12615858018398285,
      "learning_rate": 1.55e-06,
      "loss": 0.6176,
      "step": 1938
    },
    {
      "epoch": 2.4879906973014156,
      "grad_norm": 0.14888960123062134,
      "learning_rate": 1.525e-06,
      "loss": 0.6462,
      "step": 1939
    },
    {
      "epoch": 2.4892738281406634,
      "grad_norm": 0.16563569009304047,
      "learning_rate": 1.5e-06,
      "loss": 0.6951,
      "step": 1940
    },
    {
      "epoch": 2.490556958979911,
      "grad_norm": 0.1359473466873169,
      "learning_rate": 1.475e-06,
      "loss": 0.6866,
      "step": 1941
    },
    {
      "epoch": 2.4918400898191586,
      "grad_norm": 0.15769274532794952,
      "learning_rate": 1.45e-06,
      "loss": 0.6707,
      "step": 1942
    },
    {
      "epoch": 2.4931232206584064,
      "grad_norm": 0.14069971442222595,
      "learning_rate": 1.4250000000000001e-06,
      "loss": 0.6501,
      "step": 1943
    },
    {
      "epoch": 2.494406351497654,
      "grad_norm": 0.13464029133319855,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.6009,
      "step": 1944
    },
    {
      "epoch": 2.495689482336902,
      "grad_norm": 0.16317002475261688,
      "learning_rate": 1.3750000000000002e-06,
      "loss": 0.6338,
      "step": 1945
    },
    {
      "epoch": 2.49697261317615,
      "grad_norm": 0.14009489119052887,
      "learning_rate": 1.35e-06,
      "loss": 0.6176,
      "step": 1946
    },
    {
      "epoch": 2.4982557440153976,
      "grad_norm": 0.14106228947639465,
      "learning_rate": 1.325e-06,
      "loss": 0.7253,
      "step": 1947
    },
    {
      "epoch": 2.4995388748546454,
      "grad_norm": 0.13814334571361542,
      "learning_rate": 1.3e-06,
      "loss": 0.6528,
      "step": 1948
    },
    {
      "epoch": 2.5008220056938932,
      "grad_norm": 0.13895253837108612,
      "learning_rate": 1.275e-06,
      "loss": 0.6229,
      "step": 1949
    },
    {
      "epoch": 2.5021051365331406,
      "grad_norm": 0.12528285384178162,
      "learning_rate": 1.25e-06,
      "loss": 0.5489,
      "step": 1950
    },
    {
      "epoch": 2.5033882673723884,
      "grad_norm": 0.12458769977092743,
      "learning_rate": 1.2250000000000001e-06,
      "loss": 0.7325,
      "step": 1951
    },
    {
      "epoch": 2.504671398211636,
      "grad_norm": 0.1379503756761551,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.6028,
      "step": 1952
    },
    {
      "epoch": 2.505954529050884,
      "grad_norm": 0.20622695982456207,
      "learning_rate": 1.175e-06,
      "loss": 0.6328,
      "step": 1953
    },
    {
      "epoch": 2.507237659890132,
      "grad_norm": 0.1834592968225479,
      "learning_rate": 1.15e-06,
      "loss": 0.7393,
      "step": 1954
    },
    {
      "epoch": 2.5085207907293796,
      "grad_norm": 0.1329461634159088,
      "learning_rate": 1.125e-06,
      "loss": 0.6305,
      "step": 1955
    },
    {
      "epoch": 2.5098039215686274,
      "grad_norm": 0.15132160484790802,
      "learning_rate": 1.1e-06,
      "loss": 0.6468,
      "step": 1956
    },
    {
      "epoch": 2.511087052407875,
      "grad_norm": 0.14850351214408875,
      "learning_rate": 1.0749999999999999e-06,
      "loss": 0.7207,
      "step": 1957
    },
    {
      "epoch": 2.512370183247123,
      "grad_norm": 0.20022115111351013,
      "learning_rate": 1.0500000000000001e-06,
      "loss": 0.6579,
      "step": 1958
    },
    {
      "epoch": 2.513653314086371,
      "grad_norm": 0.13997414708137512,
      "learning_rate": 1.0250000000000001e-06,
      "loss": 0.6329,
      "step": 1959
    },
    {
      "epoch": 2.5149364449256186,
      "grad_norm": 0.1177024245262146,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.6416,
      "step": 1960
    },
    {
      "epoch": 2.5162195757648664,
      "grad_norm": 0.17202909290790558,
      "learning_rate": 9.75e-07,
      "loss": 0.6973,
      "step": 1961
    },
    {
      "epoch": 2.5175027066041142,
      "grad_norm": 0.14593887329101562,
      "learning_rate": 9.5e-07,
      "loss": 0.65,
      "step": 1962
    },
    {
      "epoch": 2.518785837443362,
      "grad_norm": 0.24783790111541748,
      "learning_rate": 9.25e-07,
      "loss": 0.5378,
      "step": 1963
    },
    {
      "epoch": 2.52006896828261,
      "grad_norm": 0.1566145122051239,
      "learning_rate": 9e-07,
      "loss": 0.6586,
      "step": 1964
    },
    {
      "epoch": 2.521352099121857,
      "grad_norm": 0.14079132676124573,
      "learning_rate": 8.750000000000001e-07,
      "loss": 0.6432,
      "step": 1965
    },
    {
      "epoch": 2.522635229961105,
      "grad_norm": 0.1519237905740738,
      "learning_rate": 8.500000000000001e-07,
      "loss": 0.7309,
      "step": 1966
    },
    {
      "epoch": 2.523918360800353,
      "grad_norm": 0.14740775525569916,
      "learning_rate": 8.25e-07,
      "loss": 0.6675,
      "step": 1967
    },
    {
      "epoch": 2.5252014916396006,
      "grad_norm": 0.14846907556056976,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.5475,
      "step": 1968
    },
    {
      "epoch": 2.5264846224788484,
      "grad_norm": 0.1568073183298111,
      "learning_rate": 7.75e-07,
      "loss": 0.6293,
      "step": 1969
    },
    {
      "epoch": 2.5277677533180962,
      "grad_norm": 0.13279815018177032,
      "learning_rate": 7.5e-07,
      "loss": 0.6441,
      "step": 1970
    },
    {
      "epoch": 2.529050884157344,
      "grad_norm": 0.16034287214279175,
      "learning_rate": 7.25e-07,
      "loss": 0.6756,
      "step": 1971
    },
    {
      "epoch": 2.530334014996592,
      "grad_norm": 0.13962702453136444,
      "learning_rate": 7.000000000000001e-07,
      "loss": 0.7309,
      "step": 1972
    },
    {
      "epoch": 2.531617145835839,
      "grad_norm": 0.15083888173103333,
      "learning_rate": 6.75e-07,
      "loss": 0.6594,
      "step": 1973
    },
    {
      "epoch": 2.532900276675087,
      "grad_norm": 0.13922539353370667,
      "learning_rate": 6.5e-07,
      "loss": 0.6845,
      "step": 1974
    },
    {
      "epoch": 2.534183407514335,
      "grad_norm": 0.18370391428470612,
      "learning_rate": 6.25e-07,
      "loss": 0.7015,
      "step": 1975
    },
    {
      "epoch": 2.5354665383535826,
      "grad_norm": 0.1504078209400177,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.6972,
      "step": 1976
    },
    {
      "epoch": 2.5367496691928304,
      "grad_norm": 0.143167182803154,
      "learning_rate": 5.75e-07,
      "loss": 0.6797,
      "step": 1977
    },
    {
      "epoch": 2.538032800032078,
      "grad_norm": 0.1535852700471878,
      "learning_rate": 5.5e-07,
      "loss": 0.6688,
      "step": 1978
    },
    {
      "epoch": 2.539315930871326,
      "grad_norm": 0.14264197647571564,
      "learning_rate": 5.250000000000001e-07,
      "loss": 0.7081,
      "step": 1979
    },
    {
      "epoch": 2.540599061710574,
      "grad_norm": 0.1670532375574112,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.5783,
      "step": 1980
    },
    {
      "epoch": 2.5418821925498216,
      "grad_norm": 0.1279667615890503,
      "learning_rate": 4.75e-07,
      "loss": 0.6799,
      "step": 1981
    },
    {
      "epoch": 2.5431653233890694,
      "grad_norm": 0.16824696958065033,
      "learning_rate": 4.5e-07,
      "loss": 0.6167,
      "step": 1982
    },
    {
      "epoch": 2.5444484542283172,
      "grad_norm": 0.13737574219703674,
      "learning_rate": 4.2500000000000006e-07,
      "loss": 0.726,
      "step": 1983
    },
    {
      "epoch": 2.545731585067565,
      "grad_norm": 0.12598061561584473,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.6992,
      "step": 1984
    },
    {
      "epoch": 2.547014715906813,
      "grad_norm": 0.15870770812034607,
      "learning_rate": 3.75e-07,
      "loss": 0.5808,
      "step": 1985
    },
    {
      "epoch": 2.5482978467460606,
      "grad_norm": 0.1343153417110443,
      "learning_rate": 3.5000000000000004e-07,
      "loss": 0.6221,
      "step": 1986
    },
    {
      "epoch": 2.549580977585308,
      "grad_norm": 0.1357189118862152,
      "learning_rate": 3.25e-07,
      "loss": 0.6542,
      "step": 1987
    },
    {
      "epoch": 2.550864108424556,
      "grad_norm": 0.18828342854976654,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 0.6904,
      "step": 1988
    },
    {
      "epoch": 2.5521472392638036,
      "grad_norm": 0.15896464884281158,
      "learning_rate": 2.75e-07,
      "loss": 0.6045,
      "step": 1989
    },
    {
      "epoch": 2.5534303701030514,
      "grad_norm": 0.14014972746372223,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.7393,
      "step": 1990
    },
    {
      "epoch": 2.554713500942299,
      "grad_norm": 0.1694372594356537,
      "learning_rate": 2.25e-07,
      "loss": 0.6976,
      "step": 1991
    },
    {
      "epoch": 2.555996631781547,
      "grad_norm": 0.15549033880233765,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.6747,
      "step": 1992
    },
    {
      "epoch": 2.557279762620795,
      "grad_norm": 0.13465555012226105,
      "learning_rate": 1.7500000000000002e-07,
      "loss": 0.6928,
      "step": 1993
    },
    {
      "epoch": 2.5585628934600426,
      "grad_norm": 0.14585599303245544,
      "learning_rate": 1.5000000000000002e-07,
      "loss": 0.7058,
      "step": 1994
    },
    {
      "epoch": 2.5598460242992904,
      "grad_norm": 0.1730305701494217,
      "learning_rate": 1.2500000000000002e-07,
      "loss": 0.7004,
      "step": 1995
    },
    {
      "epoch": 2.561129155138538,
      "grad_norm": 0.21158845722675323,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 0.6539,
      "step": 1996
    },
    {
      "epoch": 2.5624122859777856,
      "grad_norm": 0.13504266738891602,
      "learning_rate": 7.500000000000001e-08,
      "loss": 0.6869,
      "step": 1997
    },
    {
      "epoch": 2.5636954168170334,
      "grad_norm": 0.14100311696529388,
      "learning_rate": 5.0000000000000004e-08,
      "loss": 0.7997,
      "step": 1998
    },
    {
      "epoch": 2.564978547656281,
      "grad_norm": 0.1221596822142601,
      "learning_rate": 2.5000000000000002e-08,
      "loss": 0.6062,
      "step": 1999
    },
    {
      "epoch": 2.566261678495529,
      "grad_norm": 0.13055212795734406,
      "learning_rate": 0.0,
      "loss": 0.5805,
      "step": 2000
    },
    {
      "epoch": 2.566261678495529,
      "step": 2000,
      "total_flos": 2.605973017460736e+18,
      "train_loss": 0.7921548962593079,
      "train_runtime": 150925.8962,
      "train_samples_per_second": 0.424,
      "train_steps_per_second": 0.013
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.605973017460736e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}