{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 5198,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 4.277699683024943,
      "learning_rate": 1.282051282051282e-07,
      "loss": 1.162,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.084184527772327,
      "learning_rate": 2.564102564102564e-07,
      "loss": 0.8929,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.835270989789079,
      "learning_rate": 3.846153846153847e-07,
      "loss": 1.3735,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.41566914038926,
      "learning_rate": 5.128205128205128e-07,
      "loss": 1.3783,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.989355083718749,
      "learning_rate": 6.41025641025641e-07,
      "loss": 1.3759,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.414822860652152,
      "learning_rate": 7.692307692307694e-07,
      "loss": 1.3155,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.160143515075171,
      "learning_rate": 8.974358974358975e-07,
      "loss": 1.4908,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.05191776915416,
      "learning_rate": 1.0256410256410257e-06,
      "loss": 1.5016,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.072193439315778,
      "learning_rate": 1.153846153846154e-06,
      "loss": 0.8405,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.200486248257892,
      "learning_rate": 1.282051282051282e-06,
      "loss": 1.3301,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.056162795465168,
      "learning_rate": 1.4102564102564104e-06,
      "loss": 1.3738,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.9156020138698344,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 1.3789,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8523480759799018,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.8522,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5222822989570153,
      "learning_rate": 1.794871794871795e-06,
      "loss": 1.3809,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.83283660881569,
      "learning_rate": 1.9230769230769234e-06,
      "loss": 1.3508,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.7158440606355647,
      "learning_rate": 2.0512820512820513e-06,
      "loss": 1.2531,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5861367699176743,
      "learning_rate": 2.1794871794871797e-06,
      "loss": 1.2983,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9986906892830132,
      "learning_rate": 2.307692307692308e-06,
      "loss": 1.2038,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9727324622133897,
      "learning_rate": 2.435897435897436e-06,
      "loss": 1.2086,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.865035436916153,
      "learning_rate": 2.564102564102564e-06,
      "loss": 1.2118,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7470957336793895,
      "learning_rate": 2.6923076923076923e-06,
      "loss": 1.2199,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7404614930772573,
      "learning_rate": 2.8205128205128207e-06,
      "loss": 1.1871,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7940739537248716,
      "learning_rate": 2.948717948717949e-06,
      "loss": 1.1623,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.92896042284595,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 1.1925,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4508884173629237,
      "learning_rate": 3.205128205128206e-06,
      "loss": 0.8227,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.692663077642002,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.1359,
      "step": 26
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7544531500536447,
      "learning_rate": 3.4615384615384617e-06,
      "loss": 1.1644,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7683556722244518,
      "learning_rate": 3.58974358974359e-06,
      "loss": 1.1936,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2629450239865763,
      "learning_rate": 3.7179487179487184e-06,
      "loss": 0.6474,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8228399561454582,
      "learning_rate": 3.846153846153847e-06,
      "loss": 1.1723,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8417768649230883,
      "learning_rate": 3.974358974358974e-06,
      "loss": 1.144,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6967592180779638,
      "learning_rate": 4.102564102564103e-06,
      "loss": 1.0897,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6100313222165838,
      "learning_rate": 4.230769230769231e-06,
      "loss": 1.1145,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4033807543730634,
      "learning_rate": 4.358974358974359e-06,
      "loss": 0.6996,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1611582967401501,
      "learning_rate": 4.487179487179488e-06,
      "loss": 0.717,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.311695197341542,
      "learning_rate": 4.615384615384616e-06,
      "loss": 0.6557,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4240931498439688,
      "learning_rate": 4.743589743589744e-06,
      "loss": 1.0539,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4330380925939514,
      "learning_rate": 4.871794871794872e-06,
      "loss": 1.0703,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.667248858372498,
      "learning_rate": 5e-06,
      "loss": 1.1586,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1875214312056313,
      "learning_rate": 5.128205128205128e-06,
      "loss": 0.7067,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5453475454341925,
      "learning_rate": 5.256410256410257e-06,
      "loss": 1.1474,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.445278702965791,
      "learning_rate": 5.384615384615385e-06,
      "loss": 1.0919,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4724553988479436,
      "learning_rate": 5.512820512820514e-06,
      "loss": 1.0732,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1598583866188583,
      "learning_rate": 5.641025641025641e-06,
      "loss": 0.6908,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4237317398005958,
      "learning_rate": 5.769230769230769e-06,
      "loss": 1.1044,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4037178295481687,
      "learning_rate": 5.897435897435898e-06,
      "loss": 1.0881,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.448428834551802,
      "learning_rate": 6.025641025641026e-06,
      "loss": 1.0686,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3027577305177145,
      "learning_rate": 6.153846153846155e-06,
      "loss": 1.0687,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2714078538841642,
      "learning_rate": 6.282051282051282e-06,
      "loss": 1.085,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3248561113683808,
      "learning_rate": 6.410256410256412e-06,
      "loss": 1.0541,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6485326768304982,
      "learning_rate": 6.538461538461539e-06,
      "loss": 1.064,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3109432056795018,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.0926,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7188201423984326,
      "learning_rate": 6.794871794871796e-06,
      "loss": 1.084,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2809203671716967,
      "learning_rate": 6.923076923076923e-06,
      "loss": 1.0448,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5975728272597338,
      "learning_rate": 7.051282051282053e-06,
      "loss": 1.1346,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.155929413053281,
      "learning_rate": 7.17948717948718e-06,
      "loss": 1.0294,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4513291792563625,
      "learning_rate": 7.307692307692308e-06,
      "loss": 1.0149,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3620776202562028,
      "learning_rate": 7.435897435897437e-06,
      "loss": 1.1046,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4926533620920819,
      "learning_rate": 7.564102564102564e-06,
      "loss": 1.0312,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3628096996347814,
      "learning_rate": 7.692307692307694e-06,
      "loss": 1.0393,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1357298216470477,
      "learning_rate": 7.820512820512822e-06,
      "loss": 0.6606,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0143192999309687,
      "learning_rate": 7.948717948717949e-06,
      "loss": 0.7055,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4343860743574053,
      "learning_rate": 8.076923076923077e-06,
      "loss": 1.0684,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3271937923948027,
      "learning_rate": 8.205128205128205e-06,
      "loss": 1.0955,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.252546213088867,
      "learning_rate": 8.333333333333334e-06,
      "loss": 1.0676,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3356149208020338,
      "learning_rate": 8.461538461538462e-06,
      "loss": 1.1125,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4746678757589515,
      "learning_rate": 8.58974358974359e-06,
      "loss": 1.0419,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2022735827280107,
      "learning_rate": 8.717948717948719e-06,
      "loss": 1.0641,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.407994478405718,
      "learning_rate": 8.846153846153847e-06,
      "loss": 1.0409,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2907596347323367,
      "learning_rate": 8.974358974358976e-06,
      "loss": 1.0585,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3714184841781836,
      "learning_rate": 9.102564102564104e-06,
      "loss": 0.9997,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3831035875397737,
      "learning_rate": 9.230769230769232e-06,
      "loss": 1.105,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2555889666125897,
      "learning_rate": 9.358974358974359e-06,
      "loss": 1.044,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.374656293254763,
      "learning_rate": 9.487179487179487e-06,
      "loss": 1.0742,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.345567226122562,
      "learning_rate": 9.615384615384616e-06,
      "loss": 1.0486,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2915992333866926,
      "learning_rate": 9.743589743589744e-06,
      "loss": 1.0428,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3680970720463175,
      "learning_rate": 9.871794871794872e-06,
      "loss": 1.0175,
      "step": 77
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5116081717928493,
      "learning_rate": 1e-05,
      "loss": 1.0716,
      "step": 78
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0425423637258402,
      "learning_rate": 1.012820512820513e-05,
      "loss": 0.6606,
      "step": 79
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4115948534426401,
      "learning_rate": 1.0256410256410256e-05,
      "loss": 1.0581,
      "step": 80
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3761858506837072,
      "learning_rate": 1.0384615384615386e-05,
      "loss": 1.0144,
      "step": 81
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4344251135931279,
      "learning_rate": 1.0512820512820514e-05,
      "loss": 1.116,
      "step": 82
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1067764208923738,
      "learning_rate": 1.0641025641025643e-05,
      "loss": 0.6988,
      "step": 83
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2961485785889528,
      "learning_rate": 1.076923076923077e-05,
      "loss": 1.051,
      "step": 84
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3817634779193582,
      "learning_rate": 1.0897435897435898e-05,
      "loss": 1.09,
      "step": 85
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2025025117064378,
      "learning_rate": 1.1025641025641028e-05,
      "loss": 1.0165,
      "step": 86
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2952768916526318,
      "learning_rate": 1.1153846153846154e-05,
      "loss": 1.0241,
      "step": 87
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.302806691567461,
      "learning_rate": 1.1282051282051283e-05,
      "loss": 1.0304,
      "step": 88
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3387455269360813,
      "learning_rate": 1.1410256410256411e-05,
      "loss": 1.0822,
      "step": 89
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3500729094680997,
      "learning_rate": 1.1538461538461538e-05,
      "loss": 1.0156,
      "step": 90
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2734618319924973,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 1.0412,
      "step": 91
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.344748216169347,
      "learning_rate": 1.1794871794871796e-05,
      "loss": 1.0665,
      "step": 92
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2893160819842224,
      "learning_rate": 1.1923076923076925e-05,
      "loss": 1.0343,
      "step": 93
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3104308316110391,
      "learning_rate": 1.2051282051282051e-05,
      "loss": 1.0833,
      "step": 94
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1716118819171766,
      "learning_rate": 1.217948717948718e-05,
      "loss": 1.0504,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1596915549841091,
      "learning_rate": 1.230769230769231e-05,
      "loss": 1.007,
      "step": 96
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2321326272295587,
      "learning_rate": 1.2435897435897436e-05,
      "loss": 1.0508,
      "step": 97
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2575154575721244,
      "learning_rate": 1.2564102564102565e-05,
      "loss": 0.7,
      "step": 98
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4711401576280203,
      "learning_rate": 1.2692307692307693e-05,
      "loss": 1.0387,
      "step": 99
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2974078670014715,
      "learning_rate": 1.2820512820512823e-05,
      "loss": 0.9929,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2139034511338962,
      "learning_rate": 1.294871794871795e-05,
      "loss": 0.9968,
      "step": 101
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3350595645009493,
      "learning_rate": 1.3076923076923078e-05,
      "loss": 1.0468,
      "step": 102
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9634818700222381,
      "learning_rate": 1.3205128205128207e-05,
      "loss": 0.6941,
      "step": 103
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4820218953906699,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 1.0143,
      "step": 104
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3305130571641823,
      "learning_rate": 1.3461538461538463e-05,
      "loss": 1.0049,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3112365994118695,
      "learning_rate": 1.3589743589743592e-05,
      "loss": 1.0538,
      "step": 106
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2337495867643105,
      "learning_rate": 1.3717948717948718e-05,
      "loss": 1.0444,
      "step": 107
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2694224100040652,
      "learning_rate": 1.3846153846153847e-05,
      "loss": 1.0634,
      "step": 108
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5722161480718564,
      "learning_rate": 1.3974358974358975e-05,
      "loss": 1.0828,
      "step": 109
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.343490350250062,
      "learning_rate": 1.4102564102564105e-05,
      "loss": 1.0209,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5373320987706547,
      "learning_rate": 1.4230769230769232e-05,
      "loss": 1.0264,
      "step": 111
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2060565896347428,
      "learning_rate": 1.435897435897436e-05,
      "loss": 1.0133,
      "step": 112
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2643497548821678,
      "learning_rate": 1.4487179487179489e-05,
      "loss": 1.0298,
      "step": 113
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4937109636411856,
      "learning_rate": 1.4615384615384615e-05,
      "loss": 1.0434,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.514083174538761,
      "learning_rate": 1.4743589743589745e-05,
      "loss": 1.0255,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4194954462356992,
      "learning_rate": 1.4871794871794874e-05,
      "loss": 1.0411,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9754715762730609,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.6663,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3872261787900404,
      "learning_rate": 1.5128205128205129e-05,
      "loss": 1.0301,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2707315749294292,
      "learning_rate": 1.5256410256410257e-05,
      "loss": 1.0099,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3612794503098629,
      "learning_rate": 1.5384615384615387e-05,
      "loss": 1.0142,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2330955615241932,
      "learning_rate": 1.5512820512820516e-05,
      "loss": 1.0191,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2576117914454041,
      "learning_rate": 1.5641025641025644e-05,
      "loss": 1.0656,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3376188332780075,
      "learning_rate": 1.576923076923077e-05,
      "loss": 1.0722,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3206463735866516,
      "learning_rate": 1.5897435897435897e-05,
      "loss": 1.106,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2024655561036557,
      "learning_rate": 1.602564102564103e-05,
      "loss": 1.0297,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.362179071179234,
      "learning_rate": 1.6153846153846154e-05,
      "loss": 1.0837,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.311656985581043,
      "learning_rate": 1.6282051282051282e-05,
      "loss": 1.0047,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1466273512305565,
      "learning_rate": 1.641025641025641e-05,
      "loss": 1.0364,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3564232343136133,
      "learning_rate": 1.653846153846154e-05,
      "loss": 1.0987,
      "step": 129
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2414505747485556,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.0498,
      "step": 130
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2057146554445115,
      "learning_rate": 1.6794871794871796e-05,
      "loss": 1.0251,
      "step": 131
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3234303305481263,
      "learning_rate": 1.6923076923076924e-05,
      "loss": 0.9988,
      "step": 132
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3331890978523722,
      "learning_rate": 1.7051282051282053e-05,
      "loss": 1.0894,
      "step": 133
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2199805887007658,
      "learning_rate": 1.717948717948718e-05,
      "loss": 1.0111,
      "step": 134
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1262332548396743,
      "learning_rate": 1.730769230769231e-05,
      "loss": 0.7324,
      "step": 135
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1988418439957678,
      "learning_rate": 1.7435897435897438e-05,
      "loss": 0.9985,
      "step": 136
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3347024590625933,
      "learning_rate": 1.7564102564102566e-05,
      "loss": 1.0402,
      "step": 137
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2816015137864232,
      "learning_rate": 1.7692307692307694e-05,
      "loss": 1.0719,
      "step": 138
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2954501181102769,
      "learning_rate": 1.7820512820512823e-05,
      "loss": 1.0442,
      "step": 139
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3451063882616643,
      "learning_rate": 1.794871794871795e-05,
      "loss": 0.9897,
      "step": 140
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.256426322559422,
      "learning_rate": 1.807692307692308e-05,
      "loss": 1.0053,
      "step": 141
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3301286644023298,
      "learning_rate": 1.8205128205128208e-05,
      "loss": 1.0507,
      "step": 142
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2581661353644473,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 1.032,
      "step": 143
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3435473041425825,
      "learning_rate": 1.8461538461538465e-05,
      "loss": 1.0775,
      "step": 144
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.96901269108284,
      "learning_rate": 1.8589743589743593e-05,
      "loss": 0.6585,
      "step": 145
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1358881528166347,
      "learning_rate": 1.8717948717948718e-05,
      "loss": 0.984,
      "step": 146
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4898055920242697,
      "learning_rate": 1.8846153846153846e-05,
      "loss": 1.1,
      "step": 147
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.568640515822483,
      "learning_rate": 1.8974358974358975e-05,
      "loss": 1.0902,
      "step": 148
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.27553641291474,
      "learning_rate": 1.9102564102564106e-05,
      "loss": 1.0205,
      "step": 149
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2625663299430345,
      "learning_rate": 1.923076923076923e-05,
      "loss": 1.0236,
      "step": 150
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1258519414703054,
      "learning_rate": 1.935897435897436e-05,
      "loss": 1.0242,
      "step": 151
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9605723943496727,
      "learning_rate": 1.9487179487179488e-05,
      "loss": 0.6699,
      "step": 152
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1683560992150026,
      "learning_rate": 1.9615384615384617e-05,
      "loss": 1.0363,
      "step": 153
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1315098792530456,
      "learning_rate": 1.9743589743589745e-05,
      "loss": 1.013,
      "step": 154
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2982203249957822,
      "learning_rate": 1.9871794871794873e-05,
      "loss": 1.0703,
      "step": 155
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1110122494474952,
      "learning_rate": 2e-05,
      "loss": 0.9805,
      "step": 156
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2661259573681831,
      "learning_rate": 1.9999998058827844e-05,
      "loss": 1.04,
      "step": 157
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.164910788441104,
      "learning_rate": 1.9999992235312136e-05,
      "loss": 1.0019,
      "step": 158
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3852285931425516,
      "learning_rate": 1.9999982529455127e-05,
      "loss": 1.0576,
      "step": 159
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2297756952918304,
      "learning_rate": 1.9999968941260596e-05,
      "loss": 1.0465,
      "step": 160
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.375256681732146,
      "learning_rate": 1.9999951470733808e-05,
      "loss": 1.0126,
      "step": 161
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.190958500196254,
      "learning_rate": 1.9999930117881548e-05,
      "loss": 0.997,
      "step": 162
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4033590369831237,
      "learning_rate": 1.9999904882712115e-05,
      "loss": 1.108,
      "step": 163
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2568550327009382,
      "learning_rate": 1.99998757652353e-05,
      "loss": 0.9994,
      "step": 164
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2920117760598406,
      "learning_rate": 1.9999842765462403e-05,
      "loss": 1.1099,
      "step": 165
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.168939452805966,
      "learning_rate": 1.999980588340624e-05,
      "loss": 0.9777,
      "step": 166
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.254322303282855,
      "learning_rate": 1.9999765119081132e-05,
      "loss": 1.105,
      "step": 167
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2426551993263806,
      "learning_rate": 1.9999720472502902e-05,
      "loss": 1.016,
      "step": 168
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2945826019117477,
      "learning_rate": 1.9999671943688885e-05,
      "loss": 1.0619,
      "step": 169
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1642466303869614,
      "learning_rate": 1.9999619532657915e-05,
      "loss": 1.0543,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0743736454522137,
      "learning_rate": 1.9999563239430352e-05,
      "loss": 0.6452,
      "step": 171
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2165573741266287,
      "learning_rate": 1.9999503064028043e-05,
      "loss": 1.005,
      "step": 172
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1865016974510059,
      "learning_rate": 1.999943900647435e-05,
      "loss": 1.034,
      "step": 173
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.207656532503002,
      "learning_rate": 1.9999371066794146e-05,
      "loss": 1.1014,
      "step": 174
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1785622370994182,
      "learning_rate": 1.9999299245013805e-05,
      "loss": 0.9696,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.226516797951566,
      "learning_rate": 1.999922354116121e-05,
      "loss": 1.0324,
      "step": 176
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0567076325984537,
      "learning_rate": 1.999914395526575e-05,
      "loss": 1.0009,
      "step": 177
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1721781370368198,
      "learning_rate": 1.9999060487358333e-05,
      "loss": 1.0324,
      "step": 178
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1217198947297418,
      "learning_rate": 1.9998973137471352e-05,
      "loss": 1.0179,
      "step": 179
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2495829803014054,
      "learning_rate": 1.9998881905638727e-05,
      "loss": 1.0087,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2559769554625189,
      "learning_rate": 1.9998786791895874e-05,
      "loss": 1.0529,
      "step": 181
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3939122342472547,
      "learning_rate": 1.999868779627972e-05,
      "loss": 1.0352,
      "step": 182
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1994465608065603,
      "learning_rate": 1.9998584918828695e-05,
      "loss": 1.0429,
      "step": 183
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4226358634091871,
      "learning_rate": 1.9998478159582747e-05,
      "loss": 1.0336,
      "step": 184
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.214491195632602,
      "learning_rate": 1.999836751858332e-05,
      "loss": 0.9648,
      "step": 185
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2902644777698586,
      "learning_rate": 1.9998252995873367e-05,
      "loss": 1.0793,
      "step": 186
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0726843299362894,
      "learning_rate": 1.999813459149735e-05,
      "loss": 1.0074,
      "step": 187
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2434332257293461,
      "learning_rate": 1.9998012305501243e-05,
      "loss": 1.0166,
      "step": 188
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2045266923868916,
      "learning_rate": 1.999788613793251e-05,
      "loss": 1.0616,
      "step": 189
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4995917345794574,
      "learning_rate": 1.999775608884015e-05,
      "loss": 1.0271,
      "step": 190
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3022971279140907,
      "learning_rate": 1.9997622158274635e-05,
      "loss": 1.0835,
      "step": 191
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.334030975069622,
      "learning_rate": 1.9997484346287973e-05,
      "loss": 0.9837,
      "step": 192
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2653352820114794,
      "learning_rate": 1.9997342652933668e-05,
      "loss": 0.9884,
      "step": 193
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2212194222565886,
      "learning_rate": 1.9997197078266723e-05,
      "loss": 1.0354,
      "step": 194
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2404212375353658,
      "learning_rate": 1.999704762234366e-05,
      "loss": 1.0426,
      "step": 195
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.294325091211646,
      "learning_rate": 1.99968942852225e-05,
      "loss": 1.0677,
      "step": 196
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3283980318575808,
      "learning_rate": 1.9996737066962778e-05,
      "loss": 1.0878,
      "step": 197
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1811872779123977,
      "learning_rate": 1.9996575967625525e-05,
      "loss": 1.0001,
      "step": 198
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2317181150052352,
      "learning_rate": 1.999641098727329e-05,
      "loss": 1.0523,
      "step": 199
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2437655513835912,
      "learning_rate": 1.999624212597013e-05,
      "loss": 1.0267,
      "step": 200
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.210610004331806,
      "learning_rate": 1.9996069383781587e-05,
      "loss": 1.08,
      "step": 201
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1831076590895204,
      "learning_rate": 1.9995892760774738e-05,
      "loss": 1.0385,
      "step": 202
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0591947298601423,
      "learning_rate": 1.9995712257018153e-05,
      "loss": 0.6685,
      "step": 203
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3142103832619394,
      "learning_rate": 1.9995527872581903e-05,
      "loss": 0.9795,
      "step": 204
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2581719831066767,
      "learning_rate": 1.9995339607537578e-05,
      "loss": 0.9679,
      "step": 205
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2060485550110618,
      "learning_rate": 1.9995147461958267e-05,
      "loss": 1.0125,
      "step": 206
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2911413790850526,
      "learning_rate": 1.999495143591857e-05,
      "loss": 1.0164,
      "step": 207
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3007254026498445,
      "learning_rate": 1.999475152949459e-05,
      "loss": 1.0826,
      "step": 208
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4464800854680355,
      "learning_rate": 1.9994547742763935e-05,
      "loss": 1.0305,
      "step": 209
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1938524792189584,
      "learning_rate": 1.9994340075805724e-05,
      "loss": 1.0383,
      "step": 210
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6171832792945737,
      "learning_rate": 1.9994128528700583e-05,
      "loss": 1.0249,
      "step": 211
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0817700683656553,
      "learning_rate": 1.9993913101530635e-05,
      "loss": 0.6618,
      "step": 212
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1180304455777836,
      "learning_rate": 1.9993693794379525e-05,
      "loss": 0.9885,
      "step": 213
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2743529937260334,
      "learning_rate": 1.9993470607332387e-05,
      "loss": 1.0296,
      "step": 214
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.163262318241892,
      "learning_rate": 1.999324354047588e-05,
      "loss": 1.0625,
      "step": 215
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.205057588368276,
      "learning_rate": 1.9993012593898146e-05,
      "loss": 1.0523,
      "step": 216
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9610002179099281,
      "learning_rate": 1.9992777767688857e-05,
      "loss": 0.6909,
      "step": 217
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4085207430828912,
      "learning_rate": 1.9992539061939175e-05,
      "loss": 1.0438,
      "step": 218
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1011268882249723,
      "learning_rate": 1.999229647674178e-05,
      "loss": 1.0153,
      "step": 219
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.34351315115978,
      "learning_rate": 1.9992050012190845e-05,
      "loss": 1.0173,
      "step": 220
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1843901404423351,
      "learning_rate": 1.9991799668382058e-05,
      "loss": 0.9968,
      "step": 221
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1190205485901576,
      "learning_rate": 1.9991545445412614e-05,
      "loss": 1.0193,
      "step": 222
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9569764822255843,
      "learning_rate": 1.9991287343381208e-05,
      "loss": 0.659,
      "step": 223
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2378184929121554,
      "learning_rate": 1.9991025362388044e-05,
      "loss": 1.037,
      "step": 224
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.404072205049165,
      "learning_rate": 1.9990759502534835e-05,
      "loss": 1.0831,
      "step": 225
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.120756311175063,
      "learning_rate": 1.9990489763924796e-05,
      "loss": 1.0098,
      "step": 226
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9160829872097374,
      "learning_rate": 1.9990216146662648e-05,
      "loss": 0.6666,
      "step": 227
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8771135978812833,
      "learning_rate": 1.9989938650854618e-05,
      "loss": 0.6793,
      "step": 228
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2580529585261977,
      "learning_rate": 1.998965727660844e-05,
      "loss": 1.0801,
      "step": 229
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.391065529599239,
      "learning_rate": 1.9989372024033352e-05,
      "loss": 1.0646,
      "step": 230
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.325249589301873,
      "learning_rate": 1.99890828932401e-05,
      "loss": 1.0586,
      "step": 231
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9758926325740938,
      "learning_rate": 1.9988789884340938e-05,
      "loss": 0.6381,
      "step": 232
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2507847066327569,
      "learning_rate": 1.9988492997449615e-05,
      "loss": 1.0402,
      "step": 233
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9676088182053438,
      "learning_rate": 1.9988192232681398e-05,
      "loss": 0.6957,
      "step": 234
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0110725632280988,
      "learning_rate": 1.9987887590153055e-05,
      "loss": 0.7002,
      "step": 235
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.335281320718869,
      "learning_rate": 1.9987579069982856e-05,
      "loss": 0.9513,
      "step": 236
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2198060386242948,
      "learning_rate": 1.9987266672290577e-05,
      "loss": 1.0265,
      "step": 237
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.136356296383523,
      "learning_rate": 1.9986950397197503e-05,
      "loss": 0.7066,
      "step": 238
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1790406762238466,
      "learning_rate": 1.9986630244826425e-05,
      "loss": 1.0257,
      "step": 239
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.608876126418028,
      "learning_rate": 1.998630621530164e-05,
      "loss": 1.0822,
      "step": 240
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2735680885435927,
      "learning_rate": 1.998597830874894e-05,
      "loss": 1.0936,
      "step": 241
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1109142429463235,
      "learning_rate": 1.9985646525295634e-05,
      "loss": 0.9705,
      "step": 242
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2358653002995463,
      "learning_rate": 1.998531086507053e-05,
      "loss": 0.9671,
      "step": 243
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.085423616389646,
      "learning_rate": 1.9984971328203945e-05,
      "loss": 0.9817,
      "step": 244
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3002922772195475,
      "learning_rate": 1.9984627914827698e-05,
      "loss": 0.9652,
      "step": 245
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1986714702079795,
      "learning_rate": 1.9984280625075115e-05,
      "loss": 1.0375,
      "step": 246
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4366612021478478,
      "learning_rate": 1.9983929459081022e-05,
      "loss": 0.9834,
      "step": 247
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2309955659363117,
      "learning_rate": 1.998357441698176e-05,
      "loss": 1.056,
      "step": 248
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.143595992212968,
      "learning_rate": 1.998321549891516e-05,
      "loss": 1.0337,
      "step": 249
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3137752866863495,
      "learning_rate": 1.9982852705020572e-05,
      "loss": 1.0881,
      "step": 250
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.14354790405392,
      "learning_rate": 1.9982486035438848e-05,
      "loss": 1.0386,
      "step": 251
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1529751035111304,
      "learning_rate": 1.9982115490312334e-05,
      "loss": 1.0191,
      "step": 252
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1672371934330215,
      "learning_rate": 1.9981741069784894e-05,
      "loss": 1.0016,
      "step": 253
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1774369365565853,
      "learning_rate": 1.9981362774001886e-05,
      "loss": 1.0097,
      "step": 254
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1999847985617333,
      "learning_rate": 1.9980980603110185e-05,
      "loss": 0.9652,
      "step": 255
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2037093321523216,
      "learning_rate": 1.9980594557258158e-05,
      "loss": 1.0612,
      "step": 256
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2983818890239955,
      "learning_rate": 1.9980204636595682e-05,
      "loss": 1.0232,
      "step": 257
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1457653839426352,
      "learning_rate": 1.9979810841274135e-05,
      "loss": 1.0071,
      "step": 258
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0433067355003978,
      "learning_rate": 1.9979413171446403e-05,
      "loss": 0.9719,
      "step": 259
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1781389105510598,
      "learning_rate": 1.9979011627266884e-05,
      "loss": 1.0003,
      "step": 260
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.212244090507404,
      "learning_rate": 1.997860620889146e-05,
      "loss": 0.972,
      "step": 261
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1287404630785913,
      "learning_rate": 1.997819691647753e-05,
      "loss": 0.9631,
      "step": 262
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3164064171406105,
      "learning_rate": 1.9977783750184e-05,
      "loss": 1.0673,
      "step": 263
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1141710418930393,
      "learning_rate": 1.9977366710171274e-05,
      "loss": 0.9906,
      "step": 264
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8770475910172495,
      "learning_rate": 1.9976945796601258e-05,
      "loss": 0.6246,
      "step": 265
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3207234804234333,
      "learning_rate": 1.9976521009637366e-05,
      "loss": 1.0403,
      "step": 266
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3322355171096172,
      "learning_rate": 1.997609234944452e-05,
      "loss": 1.0672,
      "step": 267
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2169794331067292,
      "learning_rate": 1.9975659816189137e-05,
      "loss": 1.0841,
      "step": 268
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1621644864518117,
      "learning_rate": 1.997522341003914e-05,
      "loss": 0.982,
      "step": 269
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1563757553372076,
      "learning_rate": 1.9974783131163957e-05,
      "loss": 1.0395,
      "step": 270
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1551981314543878,
      "learning_rate": 1.9974338979734523e-05,
      "loss": 1.0882,
      "step": 271
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.223158378484621,
      "learning_rate": 1.997389095592327e-05,
      "loss": 1.0499,
      "step": 272
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0852294711020098,
      "learning_rate": 1.9973439059904133e-05,
      "loss": 0.9914,
      "step": 273
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.750092772702699,
      "learning_rate": 1.9972983291852565e-05,
      "loss": 1.0704,
      "step": 274
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.314540746752706,
      "learning_rate": 1.9972523651945496e-05,
      "loss": 1.0705,
      "step": 275
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0915760318502712,
      "learning_rate": 1.9972060140361384e-05,
      "loss": 0.9724,
      "step": 276
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3923297847234302,
      "learning_rate": 1.997159275728018e-05,
      "loss": 1.1707,
      "step": 277
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2475753663609628,
      "learning_rate": 1.9971121502883332e-05,
      "loss": 1.018,
      "step": 278
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9066893950078213,
      "learning_rate": 1.9970646377353802e-05,
      "loss": 0.6231,
      "step": 279
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1261679365960522,
      "learning_rate": 1.997016738087605e-05,
      "loss": 1.0101,
      "step": 280
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1937490914033009,
      "learning_rate": 1.9969684513636035e-05,
      "loss": 0.9501,
      "step": 281
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0588010933735585,
      "learning_rate": 1.9969197775821227e-05,
      "loss": 1.0087,
      "step": 282
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1713176653672386,
      "learning_rate": 1.9968707167620593e-05,
      "loss": 1.0264,
      "step": 283
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2016909080749465,
      "learning_rate": 1.9968212689224603e-05,
      "loss": 0.9314,
      "step": 284
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1905777297780082,
      "learning_rate": 1.996771434082523e-05,
      "loss": 1.0205,
      "step": 285
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1312325278666358,
      "learning_rate": 1.9967212122615958e-05,
      "loss": 1.0291,
      "step": 286
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.126479084287003,
      "learning_rate": 1.9966706034791752e-05,
      "loss": 0.9568,
      "step": 287
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8748595378310983,
      "learning_rate": 1.9966196077549106e-05,
      "loss": 0.6852,
      "step": 288
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3136827377961422,
      "learning_rate": 1.996568225108599e-05,
      "loss": 1.0486,
      "step": 289
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3606443196250464,
      "learning_rate": 1.99651645556019e-05,
      "loss": 1.0608,
      "step": 290
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.285694678038863,
      "learning_rate": 1.9964642991297817e-05,
      "loss": 1.0271,
      "step": 291
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.245196179520899,
      "learning_rate": 1.996411755837623e-05,
      "loss": 1.0332,
      "step": 292
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1909460713119506,
      "learning_rate": 1.9963588257041137e-05,
      "loss": 0.9962,
      "step": 293
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1695504896636553,
      "learning_rate": 1.996305508749802e-05,
      "loss": 1.0208,
      "step": 294
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1665862001713312,
      "learning_rate": 1.9962518049953887e-05,
      "loss": 1.0261,
      "step": 295
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.310065382308234,
      "learning_rate": 1.9961977144617225e-05,
      "loss": 1.0571,
      "step": 296
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3718718541338442,
      "learning_rate": 1.996143237169803e-05,
      "loss": 1.0495,
      "step": 297
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2735879390880234,
      "learning_rate": 1.996088373140781e-05,
      "loss": 0.988,
      "step": 298
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.230687404876456,
      "learning_rate": 1.9960331223959564e-05,
      "loss": 1.054,
      "step": 299
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1743451753371696,
      "learning_rate": 1.995977484956779e-05,
      "loss": 1.0443,
      "step": 300
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9775023289141906,
      "learning_rate": 1.9959214608448495e-05,
      "loss": 0.6924,
      "step": 301
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.303750654394259,
      "learning_rate": 1.9958650500819183e-05,
      "loss": 1.0267,
      "step": 302
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1746733469965458,
      "learning_rate": 1.995808252689886e-05,
      "loss": 1.0224,
      "step": 303
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.315984145360078,
      "learning_rate": 1.9957510686908034e-05,
      "loss": 1.0305,
      "step": 304
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0647219747096082,
      "learning_rate": 1.9956934981068713e-05,
      "loss": 0.9811,
      "step": 305
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.14845673105461,
      "learning_rate": 1.9956355409604402e-05,
      "loss": 1.015,
      "step": 306
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1917618621429902,
      "learning_rate": 1.9955771972740118e-05,
      "loss": 1.0078,
      "step": 307
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.251063715972033,
      "learning_rate": 1.9955184670702363e-05,
      "loss": 1.0665,
      "step": 308
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1966418330040836,
      "learning_rate": 1.995459350371915e-05,
      "loss": 1.0877,
      "step": 309
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.19221860831285,
      "learning_rate": 1.9953998472019996e-05,
      "loss": 1.0437,
      "step": 310
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2833857235839703,
      "learning_rate": 1.995339957583591e-05,
      "loss": 1.0336,
      "step": 311
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.127838113915208,
      "learning_rate": 1.9952796815399403e-05,
      "loss": 1.0699,
      "step": 312
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9917678729418531,
      "learning_rate": 1.9952190190944484e-05,
      "loss": 0.7057,
      "step": 313
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.125719376461344,
      "learning_rate": 1.9951579702706668e-05,
      "loss": 1.0246,
      "step": 314
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9584457442049862,
      "learning_rate": 1.9950965350922975e-05,
      "loss": 0.6548,
      "step": 315
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0982056270411296,
      "learning_rate": 1.9950347135831907e-05,
      "loss": 1.0576,
      "step": 316
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0947247551112778,
      "learning_rate": 1.994972505767348e-05,
      "loss": 1.0022,
      "step": 317
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1151304006543707,
      "learning_rate": 1.994909911668921e-05,
      "loss": 1.0431,
      "step": 318
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1991003306924624,
      "learning_rate": 1.99484693131221e-05,
      "loss": 1.0257,
      "step": 319
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1064321236443022,
      "learning_rate": 1.994783564721667e-05,
      "loss": 0.9624,
      "step": 320
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0838987378925824,
      "learning_rate": 1.9947198119218924e-05,
      "loss": 0.9989,
      "step": 321
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1208665495855121,
      "learning_rate": 1.994655672937638e-05,
      "loss": 1.0041,
      "step": 322
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9974377862877718,
      "learning_rate": 1.9945911477938044e-05,
      "loss": 0.6777,
      "step": 323
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3198657516185301,
      "learning_rate": 1.994526236515442e-05,
      "loss": 1.0085,
      "step": 324
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1398189786042345,
      "learning_rate": 1.994460939127753e-05,
      "loss": 0.6464,
      "step": 325
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1593727527099227,
      "learning_rate": 1.9943952556560863e-05,
      "loss": 0.9965,
      "step": 326
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8682137364882782,
      "learning_rate": 1.9943291861259433e-05,
      "loss": 0.6756,
      "step": 327
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2493474751013585,
      "learning_rate": 1.9942627305629747e-05,
      "loss": 1.0607,
      "step": 328
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.216420294090127,
      "learning_rate": 1.9941958889929808e-05,
      "loss": 1.0394,
      "step": 329
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0727379428299124,
      "learning_rate": 1.9941286614419113e-05,
      "loss": 1.1106,
      "step": 330
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2766507522556465,
      "learning_rate": 1.994061047935867e-05,
      "loss": 1.0766,
      "step": 331
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.050398723734925,
      "learning_rate": 1.9939930485010968e-05,
      "loss": 1.0338,
      "step": 332
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1567494412790003,
      "learning_rate": 1.9939246631640014e-05,
      "loss": 0.9632,
      "step": 333
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2903734805338969,
      "learning_rate": 1.99385589195113e-05,
      "loss": 1.092,
      "step": 334
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0737225468054052,
      "learning_rate": 1.9937867348891815e-05,
      "loss": 1.0207,
      "step": 335
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0910272098961473,
      "learning_rate": 1.9937171920050057e-05,
      "loss": 1.0285,
      "step": 336
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.056721336301862,
      "learning_rate": 1.9936472633256012e-05,
      "loss": 1.0081,
      "step": 337
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1836671086513042,
      "learning_rate": 1.9935769488781167e-05,
      "loss": 0.9595,
      "step": 338
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1209344283399645,
      "learning_rate": 1.993506248689851e-05,
      "loss": 1.0143,
      "step": 339
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.076816421422508,
      "learning_rate": 1.993435162788252e-05,
      "loss": 1.0363,
      "step": 340
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.150905786138878,
      "learning_rate": 1.993363691200918e-05,
      "loss": 1.0265,
      "step": 341
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1862771649919515,
      "learning_rate": 1.9932918339555965e-05,
      "loss": 1.0301,
      "step": 342
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0710416662907125,
      "learning_rate": 1.9932195910801848e-05,
      "loss": 0.9666,
      "step": 343
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3968221085970989,
      "learning_rate": 1.9931469626027305e-05,
      "loss": 0.9921,
      "step": 344
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0776316320068509,
      "learning_rate": 1.9930739485514304e-05,
      "loss": 1.0657,
      "step": 345
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2121469379988759,
      "learning_rate": 1.9930005489546308e-05,
      "loss": 0.9591,
      "step": 346
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2150370524162464,
      "learning_rate": 1.9929267638408277e-05,
      "loss": 1.0258,
      "step": 347
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1299690849977624,
      "learning_rate": 1.9928525932386678e-05,
      "loss": 0.9682,
      "step": 348
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0397892369842434,
      "learning_rate": 1.9927780371769463e-05,
      "loss": 0.6577,
      "step": 349
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3011268186770435,
      "learning_rate": 1.9927030956846083e-05,
      "loss": 1.0109,
      "step": 350
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8629767236262657,
      "learning_rate": 1.992627768790749e-05,
      "loss": 0.6725,
      "step": 351
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5524063645594393,
      "learning_rate": 1.9925520565246125e-05,
      "loss": 1.108,
      "step": 352
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9549775855655449,
      "learning_rate": 1.9924759589155932e-05,
      "loss": 0.6795,
      "step": 353
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9458674440341421,
      "learning_rate": 1.9923994759932344e-05,
      "loss": 0.6523,
      "step": 354
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1941553908880491,
      "learning_rate": 1.9923226077872296e-05,
      "loss": 1.0752,
      "step": 355
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1648205979387105,
      "learning_rate": 1.9922453543274223e-05,
      "loss": 1.014,
      "step": 356
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9375308456412874,
      "learning_rate": 1.9921677156438044e-05,
      "loss": 0.7296,
      "step": 357
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2104166706659572,
      "learning_rate": 1.9920896917665178e-05,
      "loss": 1.0606,
      "step": 358
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1223448513155516,
      "learning_rate": 1.992011282725854e-05,
      "loss": 1.0314,
      "step": 359
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0726616437336007,
      "learning_rate": 1.9919324885522548e-05,
      "loss": 0.961,
      "step": 360
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0533525263494994,
      "learning_rate": 1.99185330927631e-05,
      "loss": 0.9672,
      "step": 361
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0726525198454155,
      "learning_rate": 1.99177374492876e-05,
      "loss": 1.059,
      "step": 362
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1181839774577995,
      "learning_rate": 1.991693795540494e-05,
      "loss": 1.0038,
      "step": 363
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1571948381091357,
      "learning_rate": 1.9916134611425522e-05,
      "loss": 1.0345,
      "step": 364
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.213825678940458,
      "learning_rate": 1.9915327417661226e-05,
      "loss": 1.106,
      "step": 365
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0109489600288917,
      "learning_rate": 1.991451637442543e-05,
      "loss": 0.6628,
      "step": 366
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2757376863858139,
      "learning_rate": 1.9913701482033008e-05,
      "loss": 1.0666,
      "step": 367
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0449824874892224,
      "learning_rate": 1.9912882740800336e-05,
      "loss": 0.9541,
      "step": 368
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1588196000335518,
      "learning_rate": 1.9912060151045273e-05,
      "loss": 1.013,
      "step": 369
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1627061858102101,
      "learning_rate": 1.9911233713087172e-05,
      "loss": 0.9816,
      "step": 370
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9674442777548118,
      "learning_rate": 1.9910403427246895e-05,
      "loss": 0.6664,
      "step": 371
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.031977397192623,
      "learning_rate": 1.990956929384678e-05,
      "loss": 0.951,
      "step": 372
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.17519212079324,
      "learning_rate": 1.990873131321067e-05,
      "loss": 1.0609,
      "step": 373
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1019228811163704,
      "learning_rate": 1.9907889485663897e-05,
      "loss": 1.0259,
      "step": 374
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.106811263430531,
      "learning_rate": 1.9907043811533283e-05,
      "loss": 0.9501,
      "step": 375
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1366130121774456,
      "learning_rate": 1.9906194291147155e-05,
      "loss": 1.0159,
      "step": 376
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0361194440076327,
      "learning_rate": 1.9905340924835322e-05,
      "loss": 0.983,
      "step": 377
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3065004546201835,
      "learning_rate": 1.9904483712929094e-05,
      "loss": 1.0081,
      "step": 378
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3448192000771564,
      "learning_rate": 1.9903622655761267e-05,
      "loss": 1.0234,
      "step": 379
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0872054900365151,
      "learning_rate": 1.990275775366613e-05,
      "loss": 0.6722,
      "step": 380
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0906342332327523,
      "learning_rate": 1.9901889006979473e-05,
      "loss": 0.9732,
      "step": 381
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0628459767923493,
      "learning_rate": 1.990101641603857e-05,
      "loss": 0.9672,
      "step": 382
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.10554733740508,
      "learning_rate": 1.9900139981182193e-05,
      "loss": 1.0309,
      "step": 383
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1196117370693148,
      "learning_rate": 1.9899259702750604e-05,
      "loss": 1.015,
      "step": 384
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0371812399967362,
      "learning_rate": 1.9898375581085555e-05,
      "loss": 0.9944,
      "step": 385
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3693503642271023,
      "learning_rate": 1.9897487616530296e-05,
      "loss": 1.0083,
      "step": 386
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0520005709713638,
      "learning_rate": 1.9896595809429565e-05,
      "loss": 0.9984,
      "step": 387
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4389682940599324,
      "learning_rate": 1.9895700160129593e-05,
      "loss": 0.9264,
      "step": 388
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3555450816469927,
      "learning_rate": 1.9894800668978095e-05,
      "loss": 1.0713,
      "step": 389
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.147310810413707,
      "learning_rate": 1.9893897336324292e-05,
      "loss": 0.9986,
      "step": 390
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2284093314849063,
      "learning_rate": 1.9892990162518884e-05,
      "loss": 0.982,
      "step": 391
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2948230438506168,
      "learning_rate": 1.9892079147914072e-05,
      "loss": 1.0322,
      "step": 392
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2873291868431802,
      "learning_rate": 1.9891164292863537e-05,
      "loss": 1.1023,
      "step": 393
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1978309525312638,
      "learning_rate": 1.9890245597722465e-05,
      "loss": 1.014,
      "step": 394
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9048282066989525,
      "learning_rate": 1.9889323062847516e-05,
      "loss": 0.6249,
      "step": 395
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1765580906095114,
      "learning_rate": 1.988839668859686e-05,
      "loss": 1.0003,
      "step": 396
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1633486607159191,
      "learning_rate": 1.988746647533014e-05,
      "loss": 0.9825,
      "step": 397
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1013599976448514,
      "learning_rate": 1.9886532423408495e-05,
      "loss": 0.9931,
      "step": 398
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1501303365776714,
      "learning_rate": 1.9885594533194564e-05,
      "loss": 0.9872,
      "step": 399
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0902219176264103,
      "learning_rate": 1.9884652805052465e-05,
      "loss": 1.0221,
      "step": 400
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2013279119573128,
      "learning_rate": 1.9883707239347804e-05,
      "loss": 1.0055,
      "step": 401
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0144238628469815,
      "learning_rate": 1.988275783644769e-05,
      "loss": 1.0318,
      "step": 402
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1306569108837328,
      "learning_rate": 1.988180459672071e-05,
      "loss": 0.9959,
      "step": 403
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1032350263588389,
      "learning_rate": 1.988084752053695e-05,
      "loss": 0.9653,
      "step": 404
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0440999761556382,
      "learning_rate": 1.9879886608267967e-05,
      "loss": 1.0024,
      "step": 405
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.210565552674335,
      "learning_rate": 1.9878921860286832e-05,
      "loss": 1.028,
      "step": 406
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1119762015991823,
      "learning_rate": 1.9877953276968088e-05,
      "loss": 1.0298,
      "step": 407
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0226298751009009,
      "learning_rate": 1.9876980858687777e-05,
      "loss": 0.9851,
      "step": 408
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1027348804210906,
      "learning_rate": 1.9876004605823417e-05,
      "loss": 1.031,
      "step": 409
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.434369375673915,
      "learning_rate": 1.987502451875403e-05,
      "loss": 1.0467,
      "step": 410
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2292109357882002,
      "learning_rate": 1.987404059786012e-05,
      "loss": 1.0059,
      "step": 411
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0876744396243434,
      "learning_rate": 1.9873052843523676e-05,
      "loss": 1.0347,
      "step": 412
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0451368902817968,
      "learning_rate": 1.987206125612818e-05,
      "loss": 1.0017,
      "step": 413
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.103120607276679,
      "learning_rate": 1.98710658360586e-05,
      "loss": 0.9903,
      "step": 414
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1251670223749302,
      "learning_rate": 1.987006658370139e-05,
      "loss": 1.0472,
      "step": 415
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3335475731854358,
      "learning_rate": 1.9869063499444495e-05,
      "loss": 0.9921,
      "step": 416
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1493994558762628,
      "learning_rate": 1.9868056583677346e-05,
      "loss": 1.0353,
      "step": 417
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8200522001286487,
      "learning_rate": 1.9867045836790867e-05,
      "loss": 0.6552,
      "step": 418
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.18571053923601,
      "learning_rate": 1.9866031259177463e-05,
      "loss": 0.9839,
      "step": 419
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0947015319850046,
      "learning_rate": 1.9865012851231022e-05,
      "loss": 0.9813,
      "step": 420
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8398087688302335,
      "learning_rate": 1.9863990613346936e-05,
      "loss": 0.6394,
      "step": 421
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2052139158577766,
      "learning_rate": 1.986296454592206e-05,
      "loss": 0.969,
      "step": 422
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.196272633108867,
      "learning_rate": 1.9861934649354763e-05,
      "loss": 1.04,
      "step": 423
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1840712588105338,
      "learning_rate": 1.9860900924044873e-05,
      "loss": 1.0595,
      "step": 424
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1526412410141569,
      "learning_rate": 1.9859863370393726e-05,
      "loss": 1.0194,
      "step": 425
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9320924793257543,
      "learning_rate": 1.9858821988804132e-05,
      "loss": 0.6488,
      "step": 426
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3604525949965975,
      "learning_rate": 1.9857776779680393e-05,
      "loss": 1.0322,
      "step": 427
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2406360499537916,
      "learning_rate": 1.98567277434283e-05,
      "loss": 1.0178,
      "step": 428
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.140540899338587,
      "learning_rate": 1.9855674880455115e-05,
      "loss": 1.0341,
      "step": 429
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0996023174909753,
      "learning_rate": 1.98546181911696e-05,
      "loss": 1.0394,
      "step": 430
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1827971796331107,
      "learning_rate": 1.9853557675982e-05,
      "loss": 1.0251,
      "step": 431
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1718595494801514,
      "learning_rate": 1.985249333530404e-05,
      "loss": 1.022,
      "step": 432
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1783374478700117,
      "learning_rate": 1.9851425169548938e-05,
      "loss": 1.0332,
      "step": 433
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8846481863189851,
      "learning_rate": 1.9850353179131392e-05,
      "loss": 0.68,
      "step": 434
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0898126854964207,
      "learning_rate": 1.9849277364467585e-05,
      "loss": 0.9978,
      "step": 435
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0311943046158367,
      "learning_rate": 1.984819772597518e-05,
      "loss": 1.0169,
      "step": 436
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.034270588299244,
      "learning_rate": 1.9847114264073336e-05,
      "loss": 1.0238,
      "step": 437
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2318034162813114,
      "learning_rate": 1.984602697918269e-05,
      "loss": 0.9908,
      "step": 438
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0631143310565254,
      "learning_rate": 1.9844935871725363e-05,
      "loss": 1.0221,
      "step": 439
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1337468507418322,
      "learning_rate": 1.9843840942124956e-05,
      "loss": 1.0094,
      "step": 440
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2021792862603873,
      "learning_rate": 1.9842742190806566e-05,
      "loss": 1.0198,
      "step": 441
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0728820976823166,
      "learning_rate": 1.984163961819676e-05,
      "loss": 1.0217,
      "step": 442
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.212356385033231,
      "learning_rate": 1.9840533224723595e-05,
      "loss": 0.9431,
      "step": 443
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.046173048521547,
      "learning_rate": 1.9839423010816616e-05,
      "loss": 1.0028,
      "step": 444
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1073520794408251,
      "learning_rate": 1.983830897690684e-05,
      "loss": 1.0177,
      "step": 445
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1741036281811201,
      "learning_rate": 1.9837191123426777e-05,
      "loss": 1.0741,
      "step": 446
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4061483167602178,
      "learning_rate": 1.983606945081042e-05,
      "loss": 0.9651,
      "step": 447
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0965682866523248,
      "learning_rate": 1.983494395949323e-05,
      "loss": 1.0061,
      "step": 448
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0589261056726549,
      "learning_rate": 1.983381464991217e-05,
      "loss": 1.0057,
      "step": 449
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.432700621171798,
      "learning_rate": 1.9832681522505676e-05,
      "loss": 1.0449,
      "step": 450
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1094456190774105,
      "learning_rate": 1.9831544577713663e-05,
      "loss": 1.0151,
      "step": 451
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.238894337323993,
      "learning_rate": 1.983040381597754e-05,
      "loss": 1.0489,
      "step": 452
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1609600187613502,
      "learning_rate": 1.982925923774018e-05,
      "loss": 0.9799,
      "step": 453
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9482762262361668,
      "learning_rate": 1.9828110843445954e-05,
      "loss": 0.6974,
      "step": 454
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9224403956454439,
      "learning_rate": 1.982695863354071e-05,
      "loss": 0.6252,
      "step": 455
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0912998190576126,
      "learning_rate": 1.9825802608471767e-05,
      "loss": 0.9939,
      "step": 456
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9514143737962832,
      "learning_rate": 1.982464276868794e-05,
      "loss": 0.7311,
      "step": 457
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2361960193599248,
      "learning_rate": 1.982347911463952e-05,
      "loss": 1.0484,
      "step": 458
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2441884919424802,
      "learning_rate": 1.9822311646778277e-05,
      "loss": 1.0076,
      "step": 459
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0814610823767576,
      "learning_rate": 1.982114036555746e-05,
      "loss": 1.0095,
      "step": 460
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0154212124797914,
      "learning_rate": 1.9819965271431797e-05,
      "loss": 1.0242,
      "step": 461
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2949240182481982,
      "learning_rate": 1.9818786364857506e-05,
      "loss": 1.0176,
      "step": 462
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2660234953118676,
      "learning_rate": 1.9817603646292278e-05,
      "loss": 1.0005,
      "step": 463
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2390586741500829,
      "learning_rate": 1.9816417116195287e-05,
      "loss": 1.063,
      "step": 464
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3185629765155795,
      "learning_rate": 1.9815226775027182e-05,
      "loss": 0.9873,
      "step": 465
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2561168883105813,
      "learning_rate": 1.9814032623250093e-05,
      "loss": 1.0146,
      "step": 466
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.264848528317526,
      "learning_rate": 1.9812834661327632e-05,
      "loss": 1.0148,
      "step": 467
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2104188323582903,
      "learning_rate": 1.9811632889724888e-05,
      "loss": 1.0215,
      "step": 468
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1050117254788077,
      "learning_rate": 1.9810427308908437e-05,
      "loss": 1.0517,
      "step": 469
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1222205928766007,
      "learning_rate": 1.9809217919346318e-05,
      "loss": 0.9902,
      "step": 470
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1495346011803484,
      "learning_rate": 1.980800472150806e-05,
      "loss": 1.0178,
      "step": 471
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1728131183132968,
      "learning_rate": 1.9806787715864674e-05,
      "loss": 1.0095,
      "step": 472
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3043256581955789,
      "learning_rate": 1.9805566902888637e-05,
      "loss": 1.0608,
      "step": 473
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1377517426293582,
      "learning_rate": 1.9804342283053916e-05,
      "loss": 1.001,
      "step": 474
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3406944153283122,
      "learning_rate": 1.980311385683594e-05,
      "loss": 1.0234,
      "step": 475
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1494560294807197,
      "learning_rate": 1.980188162471164e-05,
      "loss": 0.9859,
      "step": 476
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0595226162841522,
      "learning_rate": 1.98006455871594e-05,
      "loss": 1.0087,
      "step": 477
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0272436770862128,
      "learning_rate": 1.97994057446591e-05,
      "loss": 0.6409,
      "step": 478
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9737139618285231,
      "learning_rate": 1.979816209769209e-05,
      "loss": 0.6924,
      "step": 479
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1213882110254898,
      "learning_rate": 1.9796914646741187e-05,
      "loss": 1.0072,
      "step": 480
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9470060602452911,
      "learning_rate": 1.9795663392290702e-05,
      "loss": 0.6836,
      "step": 481
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2226008172806921,
      "learning_rate": 1.9794408334826415e-05,
      "loss": 1.0164,
      "step": 482
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.06175418997942,
      "learning_rate": 1.979314947483558e-05,
      "loss": 1.0039,
      "step": 483
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1871792272192652,
      "learning_rate": 1.9791886812806932e-05,
      "loss": 1.0446,
      "step": 484
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.117292051241518,
      "learning_rate": 1.9790620349230676e-05,
      "loss": 1.0165,
      "step": 485
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.090749136362263,
      "learning_rate": 1.9789350084598504e-05,
      "loss": 0.9995,
      "step": 486
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0436543946073351,
      "learning_rate": 1.9788076019403565e-05,
      "loss": 1.0022,
      "step": 487
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.181660483327411,
      "learning_rate": 1.9786798154140507e-05,
      "loss": 1.0502,
      "step": 488
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.107522560854139,
      "learning_rate": 1.9785516489305437e-05,
      "loss": 0.9873,
      "step": 489
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0822272807416475,
      "learning_rate": 1.9784231025395936e-05,
      "loss": 0.9725,
      "step": 490
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1677181112916948,
      "learning_rate": 1.9782941762911075e-05,
      "loss": 1.0145,
      "step": 491
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2610649551516802,
      "learning_rate": 1.9781648702351383e-05,
      "loss": 1.0252,
      "step": 492
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1515925751405238,
      "learning_rate": 1.9780351844218874e-05,
      "loss": 0.9592,
      "step": 493
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0698537467718887,
      "learning_rate": 1.977905118901703e-05,
      "loss": 1.0019,
      "step": 494
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1492309760753596,
      "learning_rate": 1.977774673725081e-05,
      "loss": 1.0185,
      "step": 495
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.196706470537198,
      "learning_rate": 1.977643848942665e-05,
      "loss": 1.0063,
      "step": 496
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1592546686219256,
      "learning_rate": 1.977512644605246e-05,
      "loss": 1.0088,
      "step": 497
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.181933241466453,
      "learning_rate": 1.9773810607637612e-05,
      "loss": 0.947,
      "step": 498
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.041185381178923,
      "learning_rate": 1.9772490974692962e-05,
      "loss": 0.9965,
      "step": 499
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.064858231101432,
      "learning_rate": 1.9771167547730844e-05,
      "loss": 1.006,
      "step": 500
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.248411299440421,
      "learning_rate": 1.976984032726505e-05,
      "loss": 1.0994,
      "step": 501
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1240266680567472,
      "learning_rate": 1.976850931381086e-05,
      "loss": 0.9869,
      "step": 502
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.237869755396576,
      "learning_rate": 1.976717450788501e-05,
      "loss": 1.0155,
      "step": 503
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1428058600068203,
      "learning_rate": 1.9765835910005726e-05,
      "loss": 0.982,
      "step": 504
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.119135610188904,
      "learning_rate": 1.9764493520692685e-05,
      "loss": 0.9776,
      "step": 505
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.140153850247005,
      "learning_rate": 1.9763147340467067e-05,
      "loss": 0.688,
      "step": 506
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1369592048326431,
      "learning_rate": 1.9761797369851498e-05,
      "loss": 1.0011,
      "step": 507
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3871111507707883,
      "learning_rate": 1.9760443609370074e-05,
      "loss": 1.0147,
      "step": 508
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0552670229969787,
      "learning_rate": 1.975908605954838e-05,
      "loss": 0.9375,
      "step": 509
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0686006601542548,
      "learning_rate": 1.9757724720913466e-05,
      "loss": 1.0252,
      "step": 510
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0871177867570132,
      "learning_rate": 1.9756359593993845e-05,
      "loss": 1.0101,
      "step": 511
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.128015867466672,
      "learning_rate": 1.975499067931951e-05,
      "loss": 1.0532,
      "step": 512
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.304070394945391,
      "learning_rate": 1.975361797742192e-05,
      "loss": 1.0345,
      "step": 513
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.173796235585996,
      "learning_rate": 1.9752241488834002e-05,
      "loss": 0.9935,
      "step": 514
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1494864519605992,
      "learning_rate": 1.975086121409016e-05,
      "loss": 0.9846,
      "step": 515
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0572281046482088,
      "learning_rate": 1.974947715372626e-05,
      "loss": 1.034,
      "step": 516
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1951231060405196,
      "learning_rate": 1.974808930827965e-05,
      "loss": 1.0069,
      "step": 517
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2590005236828914,
      "learning_rate": 1.9746697678289128e-05,
      "loss": 1.0258,
      "step": 518
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.261575854258293,
      "learning_rate": 1.9745302264294982e-05,
      "loss": 0.9622,
      "step": 519
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8209053681507833,
      "learning_rate": 1.9743903066838954e-05,
      "loss": 0.6576,
      "step": 520
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1031970829239706,
      "learning_rate": 1.9742500086464266e-05,
      "loss": 0.9539,
      "step": 521
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1010504732722235,
      "learning_rate": 1.9741093323715597e-05,
      "loss": 0.9707,
      "step": 522
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.922208216245474,
      "learning_rate": 1.9739682779139107e-05,
      "loss": 0.6055,
      "step": 523
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1537277943655269,
      "learning_rate": 1.9738268453282414e-05,
      "loss": 0.9952,
      "step": 524
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3575954010737852,
      "learning_rate": 1.9736850346694608e-05,
      "loss": 1.0335,
      "step": 525
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8245557921055601,
      "learning_rate": 1.973542845992625e-05,
      "loss": 0.6615,
      "step": 526
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1838651409305179,
      "learning_rate": 1.9734002793529362e-05,
      "loss": 1.0271,
      "step": 527
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3210644289619886,
      "learning_rate": 1.9732573348057437e-05,
      "loss": 1.0112,
      "step": 528
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8972417757877911,
      "learning_rate": 1.973114012406544e-05,
      "loss": 0.6578,
      "step": 529
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.189584156098533,
      "learning_rate": 1.9729703122109788e-05,
      "loss": 1.0698,
      "step": 530
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2313561951792507,
      "learning_rate": 1.9728262342748384e-05,
      "loss": 1.0436,
      "step": 531
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0879196038144716,
      "learning_rate": 1.9726817786540584e-05,
      "loss": 0.9901,
      "step": 532
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1853332568637829,
      "learning_rate": 1.9725369454047215e-05,
      "loss": 1.0291,
      "step": 533
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.156912194982515,
      "learning_rate": 1.9723917345830568e-05,
      "loss": 1.0283,
      "step": 534
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1491536335209158,
      "learning_rate": 1.9722461462454405e-05,
      "loss": 0.9793,
      "step": 535
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4400416908773082,
      "learning_rate": 1.9721001804483947e-05,
      "loss": 1.0389,
      "step": 536
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0936408650497305,
      "learning_rate": 1.9719538372485887e-05,
      "loss": 0.9857,
      "step": 537
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1165973258642907,
      "learning_rate": 1.9718071167028376e-05,
      "loss": 0.9982,
      "step": 538
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1195238478356644,
      "learning_rate": 1.9716600188681038e-05,
      "loss": 1.0022,
      "step": 539
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1597183588022335,
      "learning_rate": 1.971512543801495e-05,
      "loss": 1.025,
      "step": 540
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1180105152857505,
      "learning_rate": 1.9713646915602663e-05,
      "loss": 0.9915,
      "step": 541
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1366540609384075,
      "learning_rate": 1.9712164622018197e-05,
      "loss": 0.9662,
      "step": 542
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0690715181997745,
      "learning_rate": 1.9710678557837024e-05,
      "loss": 0.977,
      "step": 543
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1358104883716253,
      "learning_rate": 1.9709188723636088e-05,
      "loss": 0.9663,
      "step": 544
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1025410610751882,
      "learning_rate": 1.970769511999379e-05,
      "loss": 0.996,
      "step": 545
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.09951282203518,
      "learning_rate": 1.9706197747490004e-05,
      "loss": 1.0304,
      "step": 546
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0871392460859381,
      "learning_rate": 1.9704696606706055e-05,
      "loss": 0.9941,
      "step": 547
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.273914618569329,
      "learning_rate": 1.9703191698224742e-05,
      "loss": 1.0679,
      "step": 548
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1392189146438605,
      "learning_rate": 1.9701683022630323e-05,
      "loss": 1.0285,
      "step": 549
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2308992323495005,
      "learning_rate": 1.9700170580508514e-05,
      "loss": 1.0171,
      "step": 550
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2532648172051464,
      "learning_rate": 1.9698654372446495e-05,
      "loss": 1.0808,
      "step": 551
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9972185288362714,
      "learning_rate": 1.969713439903292e-05,
      "loss": 1.0083,
      "step": 552
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.026002828538484,
      "learning_rate": 1.9695610660857886e-05,
      "loss": 1.0147,
      "step": 553
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1003204683886363,
      "learning_rate": 1.9694083158512965e-05,
      "loss": 0.9363,
      "step": 554
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8614587229322429,
      "learning_rate": 1.9692551892591185e-05,
      "loss": 0.6859,
      "step": 555
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0748935110450648,
      "learning_rate": 1.9691016863687037e-05,
      "loss": 0.9958,
      "step": 556
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2342742402715063,
      "learning_rate": 1.968947807239647e-05,
      "loss": 0.9742,
      "step": 557
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.998161724792691,
      "learning_rate": 1.9687935519316897e-05,
      "loss": 0.9975,
      "step": 558
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8883901957920165,
      "learning_rate": 1.9686389205047186e-05,
      "loss": 0.6564,
      "step": 559
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0987372632707026,
      "learning_rate": 1.9684839130187678e-05,
      "loss": 0.9906,
      "step": 560
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.254101931820387,
      "learning_rate": 1.968328529534016e-05,
      "loss": 0.9786,
      "step": 561
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1923300647862538,
      "learning_rate": 1.9681727701107885e-05,
      "loss": 0.9826,
      "step": 562
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.019951965820635,
      "learning_rate": 1.9680166348095568e-05,
      "loss": 0.9579,
      "step": 563
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1478287910726352,
      "learning_rate": 1.967860123690937e-05,
      "loss": 1.0167,
      "step": 564
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1636148005262925,
      "learning_rate": 1.9677032368156934e-05,
      "loss": 1.0305,
      "step": 565
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0898380333868452,
      "learning_rate": 1.967545974244734e-05,
      "loss": 1.0031,
      "step": 566
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2129113386369579,
      "learning_rate": 1.9673883360391138e-05,
      "loss": 1.0524,
      "step": 567
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0912223060682098,
      "learning_rate": 1.9672303222600333e-05,
      "loss": 1.0139,
      "step": 568
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1316876853306586,
      "learning_rate": 1.967071932968839e-05,
      "loss": 0.9917,
      "step": 569
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2722569124218044,
      "learning_rate": 1.9669131682270232e-05,
      "loss": 1.0193,
      "step": 570
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9791312709209121,
      "learning_rate": 1.9667540280962235e-05,
      "loss": 0.9461,
      "step": 571
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0514714821773956,
      "learning_rate": 1.966594512638224e-05,
      "loss": 1.024,
      "step": 572
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1378495850562729,
      "learning_rate": 1.9664346219149538e-05,
      "loss": 1.0308,
      "step": 573
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2048209103664353,
      "learning_rate": 1.966274355988488e-05,
      "loss": 1.0576,
      "step": 574
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.874738414520189,
      "learning_rate": 1.9661137149210473e-05,
      "loss": 0.6645,
      "step": 575
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.139744466803008,
      "learning_rate": 1.9659526987749987e-05,
      "loss": 1.086,
      "step": 576
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0092442790383949,
      "learning_rate": 1.9657913076128532e-05,
      "loss": 1.0029,
      "step": 577
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.201914305086405,
      "learning_rate": 1.965629541497269e-05,
      "loss": 1.0471,
      "step": 578
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9199780454110293,
      "learning_rate": 1.9654674004910493e-05,
      "loss": 0.6508,
      "step": 579
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.178527522651339,
      "learning_rate": 1.9653048846571427e-05,
      "loss": 0.9896,
      "step": 580
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3719388036728228,
      "learning_rate": 1.9651419940586437e-05,
      "loss": 1.0463,
      "step": 581
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1283628936426937,
      "learning_rate": 1.964978728758791e-05,
      "loss": 1.0004,
      "step": 582
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3501613175971166,
      "learning_rate": 1.9648150888209715e-05,
      "loss": 1.042,
      "step": 583
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1429097118305438,
      "learning_rate": 1.9646510743087144e-05,
      "loss": 0.9692,
      "step": 584
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0414931273942138,
      "learning_rate": 1.964486685285697e-05,
      "loss": 1.0295,
      "step": 585
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2041727777445497,
      "learning_rate": 1.9643219218157395e-05,
      "loss": 1.0341,
      "step": 586
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2133867831521365,
      "learning_rate": 1.9641567839628092e-05,
      "loss": 0.9645,
      "step": 587
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2782675236423815,
      "learning_rate": 1.963991271791019e-05,
      "loss": 0.9725,
      "step": 588
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0752064427571475,
      "learning_rate": 1.9638253853646255e-05,
      "loss": 0.9915,
      "step": 589
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1240117821681554,
      "learning_rate": 1.9636591247480323e-05,
      "loss": 1.0181,
      "step": 590
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0639517050435618,
      "learning_rate": 1.9634924900057867e-05,
      "loss": 0.984,
      "step": 591
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8236589544872727,
      "learning_rate": 1.963325481202583e-05,
      "loss": 0.6463,
      "step": 592
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1147589526142851,
      "learning_rate": 1.963158098403259e-05,
      "loss": 0.9651,
      "step": 593
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0282842923238045,
      "learning_rate": 1.9629903416727987e-05,
      "loss": 0.6785,
      "step": 594
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1310909934599178,
      "learning_rate": 1.962822211076331e-05,
      "loss": 1.008,
      "step": 595
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2251742013316218,
      "learning_rate": 1.96265370667913e-05,
      "loss": 1.0203,
      "step": 596
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9453576723450402,
      "learning_rate": 1.9624848285466146e-05,
      "loss": 0.9989,
      "step": 597
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.142063711388488,
      "learning_rate": 1.9623155767443498e-05,
      "loss": 0.9646,
      "step": 598
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0493321721564262,
      "learning_rate": 1.9621459513380445e-05,
      "loss": 1.0043,
      "step": 599
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1259714066088795,
      "learning_rate": 1.9619759523935532e-05,
      "loss": 1.0264,
      "step": 600
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1859583794489132,
      "learning_rate": 1.9618055799768757e-05,
      "loss": 1.0681,
      "step": 601
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.309845997945121,
      "learning_rate": 1.961634834154156e-05,
      "loss": 1.0832,
      "step": 602
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2262070535250909,
      "learning_rate": 1.9614637149916834e-05,
      "loss": 1.009,
      "step": 603
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.02934095808106,
      "learning_rate": 1.9612922225558924e-05,
      "loss": 1.0194,
      "step": 604
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.055757766988287,
      "learning_rate": 1.961120356913363e-05,
      "loss": 0.9179,
      "step": 605
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0521224472568742,
      "learning_rate": 1.960948118130818e-05,
      "loss": 0.6495,
      "step": 606
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.054451891402437,
      "learning_rate": 1.9607755062751273e-05,
      "loss": 1.0015,
      "step": 607
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1554977567219125,
      "learning_rate": 1.9606025214133046e-05,
      "loss": 1.0071,
      "step": 608
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.054981292539609,
      "learning_rate": 1.9604291636125084e-05,
      "loss": 1.0188,
      "step": 609
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8204666725102241,
      "learning_rate": 1.960255432940043e-05,
      "loss": 0.6637,
      "step": 610
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.986546869146954,
      "learning_rate": 1.9600813294633552e-05,
      "loss": 0.94,
      "step": 611
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9973851956242396,
      "learning_rate": 1.9599068532500394e-05,
      "loss": 0.9563,
      "step": 612
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1528355308780118,
      "learning_rate": 1.9597320043678322e-05,
      "loss": 1.0371,
      "step": 613
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0980941515807308,
      "learning_rate": 1.9595567828846166e-05,
      "loss": 0.9994,
      "step": 614
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0740650231731592,
      "learning_rate": 1.9593811888684192e-05,
      "loss": 1.0285,
      "step": 615
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1234239054205186,
      "learning_rate": 1.9592052223874115e-05,
      "loss": 1.0579,
      "step": 616
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2432939902775024,
      "learning_rate": 1.959028883509911e-05,
      "loss": 0.9637,
      "step": 617
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.106042968703749,
      "learning_rate": 1.9588521723043764e-05,
      "loss": 1.004,
      "step": 618
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.075140830200565,
      "learning_rate": 1.958675088839415e-05,
      "loss": 0.9666,
      "step": 619
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.172641810632296,
      "learning_rate": 1.9584976331837758e-05,
      "loss": 1.0229,
      "step": 620
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.959559089465734,
      "learning_rate": 1.9583198054063535e-05,
      "loss": 0.6489,
      "step": 621
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8296834108895985,
      "learning_rate": 1.9581416055761865e-05,
      "loss": 0.6248,
      "step": 622
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1198169202692307,
      "learning_rate": 1.9579630337624585e-05,
      "loss": 1.0318,
      "step": 623
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0888796076472518,
      "learning_rate": 1.9577840900344974e-05,
      "loss": 0.912,
      "step": 624
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2434650028422325,
      "learning_rate": 1.9576047744617752e-05,
      "loss": 1.0187,
      "step": 625
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9673675637282378,
      "learning_rate": 1.957425087113908e-05,
      "loss": 0.9561,
      "step": 626
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9532981655648701,
      "learning_rate": 1.9572450280606568e-05,
      "loss": 0.6584,
      "step": 627
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.284729832486718,
      "learning_rate": 1.9570645973719273e-05,
      "loss": 1.0433,
      "step": 628
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2150390863150302,
      "learning_rate": 1.9568837951177677e-05,
      "loss": 0.9821,
      "step": 629
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2206290667003332,
      "learning_rate": 1.9567026213683728e-05,
      "loss": 1.0126,
      "step": 630
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2719134362725537,
      "learning_rate": 1.9565210761940798e-05,
      "loss": 1.0115,
      "step": 631
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0949510477953497,
      "learning_rate": 1.956339159665371e-05,
      "loss": 1.0145,
      "step": 632
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0797729748786318,
      "learning_rate": 1.956156871852873e-05,
      "loss": 1.0195,
      "step": 633
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1577864664576945,
      "learning_rate": 1.9559742128273558e-05,
      "loss": 1.032,
      "step": 634
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.15385451953906,
      "learning_rate": 1.9557911826597337e-05,
      "loss": 1.0108,
      "step": 635
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0742305597826542,
      "learning_rate": 1.9556077814210662e-05,
      "loss": 0.9993,
      "step": 636
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8691646152035732,
      "learning_rate": 1.955424009182555e-05,
      "loss": 0.6401,
      "step": 637
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2278224633858275,
      "learning_rate": 1.955239866015547e-05,
      "loss": 1.0752,
      "step": 638
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0682153259246445,
      "learning_rate": 1.9550553519915335e-05,
      "loss": 0.969,
      "step": 639
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8770137786598924,
      "learning_rate": 1.954870467182149e-05,
      "loss": 0.6961,
      "step": 640
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1585465899135639,
      "learning_rate": 1.954685211659172e-05,
      "loss": 1.011,
      "step": 641
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.187642574966496,
      "learning_rate": 1.9544995854945248e-05,
      "loss": 0.9875,
      "step": 642
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8309726442055383,
      "learning_rate": 1.954313588760274e-05,
      "loss": 0.651,
      "step": 643
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.098498141511842,
      "learning_rate": 1.9541272215286304e-05,
      "loss": 1.0419,
      "step": 644
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1714168781629288,
      "learning_rate": 1.9539404838719477e-05,
      "loss": 0.9849,
      "step": 645
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0601499092942712,
      "learning_rate": 1.9537533758627242e-05,
      "loss": 0.9974,
      "step": 646
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.199403943446363,
      "learning_rate": 1.953565897573601e-05,
      "loss": 0.985,
      "step": 647
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.263521065697018,
      "learning_rate": 1.9533780490773645e-05,
      "loss": 1.0205,
      "step": 648
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0760380775128533,
      "learning_rate": 1.9531898304469435e-05,
      "loss": 1.0144,
      "step": 649
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.258025483793657,
      "learning_rate": 1.953001241755411e-05,
      "loss": 1.0002,
      "step": 650
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.17853920534148,
      "learning_rate": 1.952812283075984e-05,
      "loss": 0.9377,
      "step": 651
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8266236547285263,
      "learning_rate": 1.952622954482022e-05,
      "loss": 0.7048,
      "step": 652
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0154112081906197,
      "learning_rate": 1.9524332560470293e-05,
      "loss": 0.9701,
      "step": 653
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0846827061851358,
      "learning_rate": 1.9522431878446536e-05,
      "loss": 0.9861,
      "step": 654
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9718598941490708,
      "learning_rate": 1.9520527499486856e-05,
      "loss": 0.9472,
      "step": 655
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.019340526300405,
      "learning_rate": 1.95186194243306e-05,
      "loss": 0.9885,
      "step": 656
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1912727789299598,
      "learning_rate": 1.9516707653718546e-05,
      "loss": 1.0578,
      "step": 657
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1107050966972596,
      "learning_rate": 1.9514792188392914e-05,
      "loss": 1.0041,
      "step": 658
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.123657934798847,
      "learning_rate": 1.9512873029097347e-05,
      "loss": 0.9905,
      "step": 659
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0716173413180539,
      "learning_rate": 1.9510950176576933e-05,
      "loss": 1.0159,
      "step": 660
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1419150663982867,
      "learning_rate": 1.950902363157819e-05,
      "loss": 0.984,
      "step": 661
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2259429010948193,
      "learning_rate": 1.950709339484907e-05,
      "loss": 1.042,
      "step": 662
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1532875053542726,
      "learning_rate": 1.9505159467138954e-05,
      "loss": 1.0475,
      "step": 663
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9631444705720756,
      "learning_rate": 1.9503221849198655e-05,
      "loss": 0.9864,
      "step": 664
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1082685203287062,
      "learning_rate": 1.9501280541780435e-05,
      "loss": 0.7049,
      "step": 665
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.197688230968111,
      "learning_rate": 1.9499335545637968e-05,
      "loss": 0.9978,
      "step": 666
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0826522720383989,
      "learning_rate": 1.949738686152637e-05,
      "loss": 1.026,
      "step": 667
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1150573965417234,
      "learning_rate": 1.9495434490202188e-05,
      "loss": 1.0143,
      "step": 668
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.064574962279569,
      "learning_rate": 1.94934784324234e-05,
      "loss": 1.0002,
      "step": 669
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0631176004043297,
      "learning_rate": 1.9491518688949417e-05,
      "loss": 1.0112,
      "step": 670
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2491836029464825,
      "learning_rate": 1.9489555260541074e-05,
      "loss": 0.9742,
      "step": 671
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2130095543349981,
      "learning_rate": 1.948758814796064e-05,
      "loss": 0.9871,
      "step": 672
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0915728353664416,
      "learning_rate": 1.9485617351971827e-05,
      "loss": 1.0147,
      "step": 673
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2603766811267278,
      "learning_rate": 1.9483642873339753e-05,
      "loss": 1.0442,
      "step": 674
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0725679552212752,
      "learning_rate": 1.9481664712830987e-05,
      "loss": 0.9712,
      "step": 675
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0552641016525823,
      "learning_rate": 1.9479682871213515e-05,
      "loss": 0.998,
      "step": 676
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1781019468322362,
      "learning_rate": 1.9477697349256756e-05,
      "loss": 1.0123,
      "step": 677
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.052206258174652,
      "learning_rate": 1.947570814773156e-05,
      "loss": 0.9843,
      "step": 678
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.099332231623052,
      "learning_rate": 1.9473715267410206e-05,
      "loss": 1.0292,
      "step": 679
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0551122853055181,
      "learning_rate": 1.9471718709066392e-05,
      "loss": 1.0126,
      "step": 680
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1993032711644802,
      "learning_rate": 1.9469718473475256e-05,
      "loss": 1.0838,
      "step": 681
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.032355203976376,
      "learning_rate": 1.9467714561413358e-05,
      "loss": 0.9668,
      "step": 682
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8175582643316551,
      "learning_rate": 1.9465706973658683e-05,
      "loss": 0.6772,
      "step": 683
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0811341992181307,
      "learning_rate": 1.9463695710990648e-05,
      "loss": 1.0202,
      "step": 684
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.988659937165547,
      "learning_rate": 1.946168077419009e-05,
      "loss": 0.9673,
      "step": 685
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9878192083519568,
      "learning_rate": 1.9459662164039283e-05,
      "loss": 0.9893,
      "step": 686
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.040915680434638,
      "learning_rate": 1.9457639881321917e-05,
      "loss": 0.9434,
      "step": 687
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0127852410527503,
      "learning_rate": 1.9455613926823115e-05,
      "loss": 1.0106,
      "step": 688
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3040473480733275,
      "learning_rate": 1.945358430132942e-05,
      "loss": 1.0286,
      "step": 689
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.108764797006152,
      "learning_rate": 1.9451551005628803e-05,
      "loss": 1.0082,
      "step": 690
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0759821231651268,
      "learning_rate": 1.9449514040510654e-05,
      "loss": 0.9636,
      "step": 691
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2878448339851327,
      "learning_rate": 1.9447473406765803e-05,
      "loss": 1.0464,
      "step": 692
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2360812843981916,
      "learning_rate": 1.9445429105186487e-05,
      "loss": 1.007,
      "step": 693
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8704428682746779,
      "learning_rate": 1.9443381136566382e-05,
      "loss": 0.6598,
      "step": 694
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0903239491184942,
      "learning_rate": 1.9441329501700568e-05,
      "loss": 0.9978,
      "step": 695
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1740702091452073,
      "learning_rate": 1.943927420138557e-05,
      "loss": 0.9603,
      "step": 696
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.106424330413936,
      "learning_rate": 1.9437215236419322e-05,
      "loss": 1.0251,
      "step": 697
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1595899489408938,
      "learning_rate": 1.9435152607601187e-05,
      "loss": 1.0096,
      "step": 698
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0441016799590308,
      "learning_rate": 1.943308631573195e-05,
      "loss": 0.9668,
      "step": 699
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1875571650663403,
      "learning_rate": 1.9431016361613816e-05,
      "loss": 1.0572,
      "step": 700
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.095776242691168,
      "learning_rate": 1.9428942746050406e-05,
      "loss": 0.9831,
      "step": 701
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1245748934473108,
      "learning_rate": 1.9426865469846773e-05,
      "loss": 1.0387,
      "step": 702
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1001729342614044,
      "learning_rate": 1.9424784533809393e-05,
      "loss": 1.016,
      "step": 703
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1580542638395377,
      "learning_rate": 1.942269993874615e-05,
      "loss": 0.9561,
      "step": 704
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0955115795155745,
      "learning_rate": 1.9420611685466358e-05,
      "loss": 1.0481,
      "step": 705
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0936252687849564,
      "learning_rate": 1.9418519774780748e-05,
      "loss": 1.0581,
      "step": 706
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8180090301118058,
      "learning_rate": 1.9416424207501474e-05,
      "loss": 0.6701,
      "step": 707
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1246204240858388,
      "learning_rate": 1.9414324984442102e-05,
      "loss": 1.0488,
      "step": 708
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1095359042919097,
      "learning_rate": 1.9412222106417632e-05,
      "loss": 1.0251,
      "step": 709
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1992397252152707,
      "learning_rate": 1.9410115574244462e-05,
      "loss": 1.0007,
      "step": 710
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1152902555119553,
      "learning_rate": 1.9408005388740433e-05,
      "loss": 0.9237,
      "step": 711
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.123856495226336,
      "learning_rate": 1.9405891550724778e-05,
      "loss": 0.9603,
      "step": 712
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.350020204464345,
      "learning_rate": 1.940377406101817e-05,
      "loss": 0.9467,
      "step": 713
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0211656968658969,
      "learning_rate": 1.9401652920442694e-05,
      "loss": 0.9909,
      "step": 714
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.183752818403217,
      "learning_rate": 1.9399528129821842e-05,
      "loss": 0.9432,
      "step": 715
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1300030863140718,
      "learning_rate": 1.939739968998054e-05,
      "loss": 0.9534,
      "step": 716
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1383254825206541,
      "learning_rate": 1.939526760174511e-05,
      "loss": 1.0121,
      "step": 717
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3066886882680673,
      "learning_rate": 1.939313186594331e-05,
      "loss": 1.0105,
      "step": 718
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0894929013918904,
      "learning_rate": 1.9390992483404308e-05,
      "loss": 0.9595,
      "step": 719
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.960805639179041,
      "learning_rate": 1.938884945495868e-05,
      "loss": 0.7093,
      "step": 720
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0647613735269439,
      "learning_rate": 1.9386702781438425e-05,
      "loss": 1.0264,
      "step": 721
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.052288628313723,
      "learning_rate": 1.938455246367696e-05,
      "loss": 0.9809,
      "step": 722
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3150582874237855,
      "learning_rate": 1.9382398502509107e-05,
      "loss": 1.0709,
      "step": 723
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0479749400731122,
      "learning_rate": 1.938024089877111e-05,
      "loss": 0.9741,
      "step": 724
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1846616378403072,
      "learning_rate": 1.9378079653300624e-05,
      "loss": 1.0408,
      "step": 725
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1171969905405141,
      "learning_rate": 1.9375914766936723e-05,
      "loss": 1.0236,
      "step": 726
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0573748608584401,
      "learning_rate": 1.9373746240519884e-05,
      "loss": 1.0231,
      "step": 727
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0841113901041644,
      "learning_rate": 1.937157407489201e-05,
      "loss": 1.0133,
      "step": 728
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2000618376460364,
      "learning_rate": 1.9369398270896403e-05,
      "loss": 0.9864,
      "step": 729
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.031078441643056,
      "learning_rate": 1.936721882937779e-05,
      "loss": 1.0473,
      "step": 730
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1004764304393384,
      "learning_rate": 1.9365035751182307e-05,
      "loss": 1.0257,
      "step": 731
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0743345781215818,
      "learning_rate": 1.93628490371575e-05,
      "loss": 0.9878,
      "step": 732
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0647810376211475,
      "learning_rate": 1.9360658688152322e-05,
      "loss": 0.6808,
      "step": 733
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0614148996246184,
      "learning_rate": 1.9358464705017143e-05,
      "loss": 1.0143,
      "step": 734
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.125647626672716,
      "learning_rate": 1.9356267088603745e-05,
      "loss": 1.0025,
      "step": 735
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9918368013782787,
      "learning_rate": 1.9354065839765316e-05,
      "loss": 0.9489,
      "step": 736
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7980029398517164,
      "learning_rate": 1.9351860959356462e-05,
      "loss": 0.6328,
      "step": 737
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.187214480209861,
      "learning_rate": 1.9349652448233187e-05,
      "loss": 0.9697,
      "step": 738
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8103474612993053,
      "learning_rate": 1.934744030725291e-05,
      "loss": 0.6559,
      "step": 739
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.092504814555596,
      "learning_rate": 1.934522453727447e-05,
      "loss": 0.9965,
      "step": 740
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0042110754494824,
      "learning_rate": 1.93430051391581e-05,
      "loss": 0.968,
      "step": 741
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1229117084099849,
      "learning_rate": 1.934078211376544e-05,
      "loss": 1.0026,
      "step": 742
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0441166911707316,
      "learning_rate": 1.9338555461959554e-05,
      "loss": 0.9591,
      "step": 743
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9840214348375484,
      "learning_rate": 1.93363251846049e-05,
      "loss": 1.007,
      "step": 744
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2160468932137862,
      "learning_rate": 1.9334091282567352e-05,
      "loss": 0.9865,
      "step": 745
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0429005245600043,
      "learning_rate": 1.9331853756714185e-05,
      "loss": 0.9724,
      "step": 746
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8530856540011683,
      "learning_rate": 1.9329612607914088e-05,
      "loss": 0.6706,
      "step": 747
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4524155262429288,
      "learning_rate": 1.9327367837037142e-05,
      "loss": 1.0146,
      "step": 748
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9477158194371559,
      "learning_rate": 1.9325119444954855e-05,
      "loss": 0.9995,
      "step": 749
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.142800132261686,
      "learning_rate": 1.9322867432540126e-05,
      "loss": 1.0558,
      "step": 750
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.156131000855066,
      "learning_rate": 1.9320611800667268e-05,
      "loss": 0.9402,
      "step": 751
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1026999431795848,
      "learning_rate": 1.9318352550211986e-05,
      "loss": 1.0514,
      "step": 752
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0501733501857546,
      "learning_rate": 1.9316089682051403e-05,
      "loss": 1.0046,
      "step": 753
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0573093088581553,
      "learning_rate": 1.9313823197064042e-05,
      "loss": 0.9153,
      "step": 754
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9064562789896379,
      "learning_rate": 1.9311553096129835e-05,
      "loss": 0.6636,
      "step": 755
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1186555710857962,
      "learning_rate": 1.9309279380130112e-05,
      "loss": 1.0097,
      "step": 756
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0547019878028168,
      "learning_rate": 1.93070020499476e-05,
      "loss": 0.9871,
      "step": 757
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1004533152568925,
      "learning_rate": 1.930472110646645e-05,
      "loss": 0.9614,
      "step": 758
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0161719402656864,
      "learning_rate": 1.9302436550572187e-05,
      "loss": 0.9858,
      "step": 759
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.208279714382971,
      "learning_rate": 1.930014838315177e-05,
      "loss": 1.0472,
      "step": 760
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1074592739178306,
      "learning_rate": 1.9297856605093534e-05,
      "loss": 0.9478,
      "step": 761
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1129682698040253,
      "learning_rate": 1.9295561217287226e-05,
      "loss": 1.0,
      "step": 762
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0648283734474329,
      "learning_rate": 1.9293262220624002e-05,
      "loss": 0.9881,
      "step": 763
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0276564069741714,
      "learning_rate": 1.9290959615996407e-05,
      "loss": 0.9858,
      "step": 764
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0353612037858166,
      "learning_rate": 1.9288653404298392e-05,
      "loss": 0.9922,
      "step": 765
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.237572143163481,
      "learning_rate": 1.9286343586425307e-05,
      "loss": 1.0561,
      "step": 766
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0837130638767616,
      "learning_rate": 1.9284030163273907e-05,
      "loss": 0.9646,
      "step": 767
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9780040836758322,
      "learning_rate": 1.9281713135742333e-05,
      "loss": 0.9746,
      "step": 768
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0317618224748069,
      "learning_rate": 1.9279392504730147e-05,
      "loss": 1.0341,
      "step": 769
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9634727470025451,
      "learning_rate": 1.9277068271138287e-05,
      "loss": 0.6803,
      "step": 770
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0838793796088397,
      "learning_rate": 1.9274740435869107e-05,
      "loss": 0.9825,
      "step": 771
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0920270536840568,
      "learning_rate": 1.927240899982635e-05,
      "loss": 0.9809,
      "step": 772
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1136629031832495,
      "learning_rate": 1.9270073963915162e-05,
      "loss": 1.0611,
      "step": 773
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0758474843518953,
      "learning_rate": 1.9267735329042086e-05,
      "loss": 0.9861,
      "step": 774
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9635654704809378,
      "learning_rate": 1.9265393096115056e-05,
      "loss": 0.6999,
      "step": 775
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9863105647251854,
      "learning_rate": 1.926304726604341e-05,
      "loss": 0.9547,
      "step": 776
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0641601910774467,
      "learning_rate": 1.9260697839737875e-05,
      "loss": 1.0243,
      "step": 777
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0941956340270163,
      "learning_rate": 1.925834481811059e-05,
      "loss": 1.0264,
      "step": 778
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.974399140718357,
      "learning_rate": 1.9255988202075065e-05,
      "loss": 0.9985,
      "step": 779
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0782976551081762,
      "learning_rate": 1.925362799254623e-05,
      "loss": 0.9858,
      "step": 780
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9203171578078526,
      "learning_rate": 1.9251264190440398e-05,
      "loss": 0.693,
      "step": 781
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0169672924219646,
      "learning_rate": 1.9248896796675277e-05,
      "loss": 0.9721,
      "step": 782
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0582750802379122,
      "learning_rate": 1.924652581216997e-05,
      "loss": 1.0031,
      "step": 783
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9999727984916021,
      "learning_rate": 1.9244151237844975e-05,
      "loss": 0.9549,
      "step": 784
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0859288733726324,
      "learning_rate": 1.9241773074622182e-05,
      "loss": 0.9621,
      "step": 785
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1640182593118549,
      "learning_rate": 1.923939132342488e-05,
      "loss": 1.0563,
      "step": 786
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1198600454762888,
      "learning_rate": 1.923700598517775e-05,
      "loss": 0.969,
      "step": 787
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0884534262443948,
      "learning_rate": 1.923461706080685e-05,
      "loss": 1.008,
      "step": 788
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0794255184646173,
      "learning_rate": 1.923222455123965e-05,
      "loss": 1.0122,
      "step": 789
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.072666320580728,
      "learning_rate": 1.9229828457405005e-05,
      "loss": 1.0486,
      "step": 790
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.237059993717423,
      "learning_rate": 1.9227428780233162e-05,
      "loss": 0.9436,
      "step": 791
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9737794733594864,
      "learning_rate": 1.922502552065576e-05,
      "loss": 0.9778,
      "step": 792
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.192785408324741,
      "learning_rate": 1.922261867960582e-05,
      "loss": 0.9405,
      "step": 793
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2147989874286884,
      "learning_rate": 1.9220208258017763e-05,
      "loss": 1.0324,
      "step": 794
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.124588013005867,
      "learning_rate": 1.92177942568274e-05,
      "loss": 0.9925,
      "step": 795
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0866887424778664,
      "learning_rate": 1.921537667697193e-05,
      "loss": 0.9878,
      "step": 796
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.259831770557628,
      "learning_rate": 1.9212955519389938e-05,
      "loss": 1.0118,
      "step": 797
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.02274366011414,
      "learning_rate": 1.9210530785021405e-05,
      "loss": 1.0194,
      "step": 798
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9996138643027563,
      "learning_rate": 1.9208102474807692e-05,
      "loss": 0.9813,
      "step": 799
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1337657526858136,
      "learning_rate": 1.920567058969155e-05,
      "loss": 0.9995,
      "step": 800
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0255953845440977,
      "learning_rate": 1.920323513061713e-05,
      "loss": 1.0284,
      "step": 801
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1361962183342107,
      "learning_rate": 1.9200796098529956e-05,
      "loss": 0.9676,
      "step": 802
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9742479786660095,
      "learning_rate": 1.919835349437694e-05,
      "loss": 0.9602,
      "step": 803
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0890065845932266,
      "learning_rate": 1.9195907319106394e-05,
      "loss": 0.9844,
      "step": 804
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0658117248789674,
      "learning_rate": 1.9193457573667996e-05,
      "loss": 0.9948,
      "step": 805
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9818716489657592,
      "learning_rate": 1.919100425901283e-05,
      "loss": 0.7091,
      "step": 806
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.067563894046494,
      "learning_rate": 1.9188547376093355e-05,
      "loss": 1.0085,
      "step": 807
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0935499188156859,
      "learning_rate": 1.918608692586342e-05,
      "loss": 1.0017,
      "step": 808
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0160138042769862,
      "learning_rate": 1.918362290927825e-05,
      "loss": 0.6738,
      "step": 809
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0056962339263986,
      "learning_rate": 1.9181155327294468e-05,
      "loss": 0.9774,
      "step": 810
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.13029953800152,
      "learning_rate": 1.9178684180870072e-05,
      "loss": 1.0571,
      "step": 811
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.121430244770439,
      "learning_rate": 1.9176209470964446e-05,
      "loss": 0.9878,
      "step": 812
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.14455102758603,
      "learning_rate": 1.9173731198538354e-05,
      "loss": 0.9535,
      "step": 813
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0371023346685324,
      "learning_rate": 1.9171249364553956e-05,
      "loss": 0.94,
      "step": 814
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1459036628496073,
      "learning_rate": 1.9168763969974773e-05,
      "loss": 1.023,
      "step": 815
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1242733110287901,
      "learning_rate": 1.916627501576573e-05,
      "loss": 1.0515,
      "step": 816
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0941490269286946,
      "learning_rate": 1.916378250289312e-05,
      "loss": 0.9877,
      "step": 817
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.988445529835027,
      "learning_rate": 1.9161286432324628e-05,
      "loss": 0.9339,
      "step": 818
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1318600370624265,
      "learning_rate": 1.9158786805029307e-05,
      "loss": 0.988,
      "step": 819
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0321109382387492,
      "learning_rate": 1.9156283621977603e-05,
      "loss": 0.9474,
      "step": 820
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0813452356904474,
      "learning_rate": 1.9153776884141336e-05,
      "loss": 1.0275,
      "step": 821
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.138095020614241,
      "learning_rate": 1.915126659249371e-05,
      "loss": 1.0716,
      "step": 822
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9518489652794843,
      "learning_rate": 1.9148752748009304e-05,
      "loss": 0.9052,
      "step": 823
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.036335692803142,
      "learning_rate": 1.914623535166408e-05,
      "loss": 0.8842,
      "step": 824
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.099933979084468,
      "learning_rate": 1.9143714404435382e-05,
      "loss": 0.9957,
      "step": 825
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2236325015853888,
      "learning_rate": 1.9141189907301922e-05,
      "loss": 0.9801,
      "step": 826
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8916765658824924,
      "learning_rate": 1.9138661861243802e-05,
      "loss": 0.9592,
      "step": 827
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.125924408830532,
      "learning_rate": 1.913613026724249e-05,
      "loss": 0.9816,
      "step": 828
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0405577063773637,
      "learning_rate": 1.9133595126280848e-05,
      "loss": 1.0655,
      "step": 829
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.378306699215027,
      "learning_rate": 1.9131056439343095e-05,
      "loss": 0.9795,
      "step": 830
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0578274022624035,
      "learning_rate": 1.9128514207414838e-05,
      "loss": 0.9488,
      "step": 831
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1712451567426934,
      "learning_rate": 1.9125968431483068e-05,
      "loss": 1.0032,
      "step": 832
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1272811417442323,
      "learning_rate": 1.9123419112536132e-05,
      "loss": 1.0021,
      "step": 833
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2464505405311004,
      "learning_rate": 1.912086625156377e-05,
      "loss": 1.0675,
      "step": 834
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9803607715462651,
      "learning_rate": 1.911830984955709e-05,
      "loss": 1.0224,
      "step": 835
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0537427902047505,
      "learning_rate": 1.911574990750857e-05,
      "loss": 1.031,
      "step": 836
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.926249562944616,
      "learning_rate": 1.9113186426412073e-05,
      "loss": 0.6528,
      "step": 837
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0922193784522845,
      "learning_rate": 1.9110619407262828e-05,
      "loss": 1.0638,
      "step": 838
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.977356741744155,
      "learning_rate": 1.9108048851057447e-05,
      "loss": 1.0027,
      "step": 839
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0934973169407463,
      "learning_rate": 1.9105474758793897e-05,
      "loss": 0.9858,
      "step": 840
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9539496758231832,
      "learning_rate": 1.9102897131471536e-05,
      "loss": 0.972,
      "step": 841
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1750580963444528,
      "learning_rate": 1.9100315970091088e-05,
      "loss": 0.9944,
      "step": 842
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0524786698766333,
      "learning_rate": 1.9097731275654645e-05,
      "loss": 0.988,
      "step": 843
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9854251013724802,
      "learning_rate": 1.909514304916568e-05,
      "loss": 0.9809,
      "step": 844
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1303920542048633,
      "learning_rate": 1.9092551291629026e-05,
      "loss": 1.0557,
      "step": 845
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.046254519813121,
      "learning_rate": 1.9089956004050893e-05,
      "loss": 1.0355,
      "step": 846
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1351281711721453,
      "learning_rate": 1.908735718743887e-05,
      "loss": 1.0482,
      "step": 847
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.056319941689377,
      "learning_rate": 1.908475484280189e-05,
      "loss": 0.9944,
      "step": 848
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1414806974614944,
      "learning_rate": 1.908214897115029e-05,
      "loss": 0.9941,
      "step": 849
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2777840922732038,
      "learning_rate": 1.907953957349575e-05,
      "loss": 1.0154,
      "step": 850
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2178038405559464,
      "learning_rate": 1.907692665085133e-05,
      "loss": 1.0431,
      "step": 851
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9887397257704198,
      "learning_rate": 1.9074310204231457e-05,
      "loss": 0.9668,
      "step": 852
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9382246635298883,
      "learning_rate": 1.9071690234651923e-05,
      "loss": 0.9769,
      "step": 853
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1330601016704596,
      "learning_rate": 1.9069066743129893e-05,
      "loss": 0.9564,
      "step": 854
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9610611661076243,
      "learning_rate": 1.90664397306839e-05,
      "loss": 0.965,
      "step": 855
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1828492029157143,
      "learning_rate": 1.9063809198333832e-05,
      "loss": 0.995,
      "step": 856
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2186054751935311,
      "learning_rate": 1.9061175147100957e-05,
      "loss": 1.0228,
      "step": 857
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1413471323457272,
      "learning_rate": 1.905853757800791e-05,
      "loss": 0.9972,
      "step": 858
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0254559404142953,
      "learning_rate": 1.9055896492078675e-05,
      "loss": 0.9948,
      "step": 859
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9986608969370687,
      "learning_rate": 1.905325189033862e-05,
      "loss": 0.9986,
      "step": 860
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.078217069048036,
      "learning_rate": 1.905060377381447e-05,
      "loss": 0.9927,
      "step": 861
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2178209558621143,
      "learning_rate": 1.904795214353431e-05,
      "loss": 0.9971,
      "step": 862
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.033577044147273,
      "learning_rate": 1.90452970005276e-05,
      "loss": 1.0157,
      "step": 863
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9789502057784726,
      "learning_rate": 1.9042638345825155e-05,
      "loss": 0.9521,
      "step": 864
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.088275740668406,
      "learning_rate": 1.9039976180459158e-05,
      "loss": 1.0204,
      "step": 865
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9471133854531825,
      "learning_rate": 1.9037310505463153e-05,
      "loss": 0.9734,
      "step": 866
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0219050462278074,
      "learning_rate": 1.9034641321872043e-05,
      "loss": 1.0078,
      "step": 867
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.023265220227587,
      "learning_rate": 1.9031968630722104e-05,
      "loss": 0.9882,
      "step": 868
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0494051436903085,
      "learning_rate": 1.902929243305096e-05,
      "loss": 1.0345,
      "step": 869
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8708043205648761,
      "learning_rate": 1.902661272989761e-05,
      "loss": 0.3376,
      "step": 870
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.12239475172517,
      "learning_rate": 1.9023929522302394e-05,
      "loss": 0.9716,
      "step": 871
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.163560912055775,
      "learning_rate": 1.9021242811307044e-05,
      "loss": 0.9961,
      "step": 872
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0958776373275347,
      "learning_rate": 1.901855259795462e-05,
      "loss": 1.0311,
      "step": 873
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0505144057855806,
      "learning_rate": 1.9015858883289556e-05,
      "loss": 0.9617,
      "step": 874
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1184643829365217,
      "learning_rate": 1.9013161668357655e-05,
      "loss": 0.9907,
      "step": 875
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1241556827433885,
      "learning_rate": 1.901046095420606e-05,
      "loss": 0.9622,
      "step": 876
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0880898855179153,
      "learning_rate": 1.9007756741883284e-05,
      "loss": 0.9824,
      "step": 877
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1317093141061167,
      "learning_rate": 1.9005049032439193e-05,
      "loss": 1.0114,
      "step": 878
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9958752056974334,
      "learning_rate": 1.9002337826925012e-05,
      "loss": 1.009,
      "step": 879
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2559537051238168,
      "learning_rate": 1.899962312639333e-05,
      "loss": 0.9321,
      "step": 880
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1900845685328778,
      "learning_rate": 1.8996904931898085e-05,
      "loss": 0.9781,
      "step": 881
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0602290755722663,
      "learning_rate": 1.899418324449457e-05,
      "loss": 0.9628,
      "step": 882
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3297870491839658,
      "learning_rate": 1.8991458065239444e-05,
      "loss": 0.7221,
      "step": 883
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0512567207966261,
      "learning_rate": 1.8988729395190712e-05,
      "loss": 1.0072,
      "step": 884
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9402345678248444,
      "learning_rate": 1.8985997235407735e-05,
      "loss": 0.6605,
      "step": 885
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8663197546347378,
      "learning_rate": 1.898326158695124e-05,
      "loss": 0.6376,
      "step": 886
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.900945427100467,
      "learning_rate": 1.8980522450883287e-05,
      "loss": 0.9755,
      "step": 887
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1799064795749368,
      "learning_rate": 1.8977779828267314e-05,
      "loss": 1.0161,
      "step": 888
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0621131589828432,
      "learning_rate": 1.8975033720168094e-05,
      "loss": 0.9424,
      "step": 889
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1150408153449998,
      "learning_rate": 1.897228412765177e-05,
      "loss": 0.9755,
      "step": 890
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0125748522230045,
      "learning_rate": 1.896953105178582e-05,
      "loss": 1.0386,
      "step": 891
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.973223855608447,
      "learning_rate": 1.8966774493639084e-05,
      "loss": 0.9662,
      "step": 892
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3299215777105,
      "learning_rate": 1.896401445428176e-05,
      "loss": 0.9612,
      "step": 893
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1256942953529494,
      "learning_rate": 1.896125093478538e-05,
      "loss": 1.0004,
      "step": 894
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0309156363239185,
      "learning_rate": 1.895848393622284e-05,
      "loss": 1.0022,
      "step": 895
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0285084392490005,
      "learning_rate": 1.895571345966839e-05,
      "loss": 0.9916,
      "step": 896
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2326159962905403,
      "learning_rate": 1.8952939506197622e-05,
      "loss": 0.9811,
      "step": 897
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0333875967382515,
      "learning_rate": 1.8950162076887477e-05,
      "loss": 0.9593,
      "step": 898
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0397574694987426,
      "learning_rate": 1.894738117281625e-05,
      "loss": 1.0205,
      "step": 899
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.221009489143249,
      "learning_rate": 1.8944596795063584e-05,
      "loss": 0.9834,
      "step": 900
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.088023994684154,
      "learning_rate": 1.894180894471047e-05,
      "loss": 1.0592,
      "step": 901
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9932766832228249,
      "learning_rate": 1.8939017622839253e-05,
      "loss": 0.969,
      "step": 902
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0989673190028246,
      "learning_rate": 1.8936222830533613e-05,
      "loss": 1.008,
      "step": 903
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0484565259941812,
      "learning_rate": 1.8933424568878586e-05,
      "loss": 0.7155,
      "step": 904
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1384021687444026,
      "learning_rate": 1.8930622838960555e-05,
      "loss": 0.9947,
      "step": 905
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0426040537074721,
      "learning_rate": 1.8927817641867244e-05,
      "loss": 1.0073,
      "step": 906
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0674795061588178,
      "learning_rate": 1.8925008978687737e-05,
      "loss": 1.0379,
      "step": 907
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0713261795221007,
      "learning_rate": 1.8922196850512446e-05,
      "loss": 1.016,
      "step": 908
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9644845133766935,
      "learning_rate": 1.8919381258433135e-05,
      "loss": 0.967,
      "step": 909
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0982347871540223,
      "learning_rate": 1.8916562203542916e-05,
      "loss": 1.0226,
      "step": 910
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.021657907901574,
      "learning_rate": 1.8913739686936244e-05,
      "loss": 1.0127,
      "step": 911
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.143610206122845,
      "learning_rate": 1.8910913709708918e-05,
      "loss": 0.9611,
      "step": 912
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.075217291934735,
      "learning_rate": 1.8908084272958077e-05,
      "loss": 0.9825,
      "step": 913
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.157356896082768,
      "learning_rate": 1.8905251377782206e-05,
      "loss": 1.0143,
      "step": 914
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0481546394515056,
      "learning_rate": 1.8902415025281136e-05,
      "loss": 1.0242,
      "step": 915
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1772541626733217,
      "learning_rate": 1.889957521655603e-05,
      "loss": 1.063,
      "step": 916
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.14925313145459,
      "learning_rate": 1.8896731952709408e-05,
      "loss": 1.0056,
      "step": 917
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5272146329548728,
      "learning_rate": 1.8893885234845117e-05,
      "loss": 0.9157,
      "step": 918
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.029530704266639,
      "learning_rate": 1.8891035064068354e-05,
      "loss": 0.9116,
      "step": 919
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9666425430053295,
      "learning_rate": 1.888818144148565e-05,
      "loss": 1.0293,
      "step": 920
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.07207177957899,
      "learning_rate": 1.888532436820488e-05,
      "loss": 1.0586,
      "step": 921
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3086722885946112,
      "learning_rate": 1.8882463845335263e-05,
      "loss": 0.9687,
      "step": 922
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.253163100775817,
      "learning_rate": 1.8879599873987343e-05,
      "loss": 1.0334,
      "step": 923
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1043178797511468,
      "learning_rate": 1.8876732455273022e-05,
      "loss": 0.9946,
      "step": 924
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9625548847143204,
      "learning_rate": 1.8873861590305527e-05,
      "loss": 0.9295,
      "step": 925
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0917494988356105,
      "learning_rate": 1.8870987280199428e-05,
      "loss": 1.0114,
      "step": 926
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9372593538557502,
      "learning_rate": 1.886810952607063e-05,
      "loss": 0.9414,
      "step": 927
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2142072049486587,
      "learning_rate": 1.8865228329036372e-05,
      "loss": 0.9662,
      "step": 928
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8896582536961908,
      "learning_rate": 1.886234369021524e-05,
      "loss": 0.6378,
      "step": 929
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1090290191602372,
      "learning_rate": 1.885945561072715e-05,
      "loss": 1.0039,
      "step": 930
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0224361669441908,
      "learning_rate": 1.885656409169335e-05,
      "loss": 0.9585,
      "step": 931
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0169855598631556,
      "learning_rate": 1.885366913423643e-05,
      "loss": 0.984,
      "step": 932
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0894725741746116,
      "learning_rate": 1.8850770739480312e-05,
      "loss": 0.9739,
      "step": 933
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8712637655474066,
      "learning_rate": 1.8847868908550252e-05,
      "loss": 0.6642,
      "step": 934
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9931867131788884,
      "learning_rate": 1.8844963642572837e-05,
      "loss": 0.9776,
      "step": 935
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1158691604304447,
      "learning_rate": 1.8842054942676e-05,
      "loss": 0.9764,
      "step": 936
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9930588744826317,
      "learning_rate": 1.8839142809988987e-05,
      "loss": 0.9924,
      "step": 937
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8780026186765971,
      "learning_rate": 1.88362272456424e-05,
      "loss": 0.654,
      "step": 938
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0114616820908613,
      "learning_rate": 1.8833308250768153e-05,
      "loss": 0.9788,
      "step": 939
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8520840210702224,
      "learning_rate": 1.8830385826499507e-05,
      "loss": 0.6494,
      "step": 940
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8775822349859148,
      "learning_rate": 1.882745997397104e-05,
      "loss": 0.7195,
      "step": 941
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1011605297651363,
      "learning_rate": 1.8824530694318675e-05,
      "loss": 1.0045,
      "step": 942
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.041032829240915,
      "learning_rate": 1.882159798867966e-05,
      "loss": 0.9843,
      "step": 943
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1356629098380477,
      "learning_rate": 1.8818661858192562e-05,
      "loss": 0.9756,
      "step": 944
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1162610756133986,
      "learning_rate": 1.88157223039973e-05,
      "loss": 1.012,
      "step": 945
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0635844870784337,
      "learning_rate": 1.8812779327235106e-05,
      "loss": 1.0357,
      "step": 946
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9513406924394022,
      "learning_rate": 1.880983292904854e-05,
      "loss": 0.9528,
      "step": 947
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8977725807894252,
      "learning_rate": 1.88068831105815e-05,
      "loss": 0.6616,
      "step": 948
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9926683665569548,
      "learning_rate": 1.8803929872979214e-05,
      "loss": 0.9787,
      "step": 949
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1523039968551831,
      "learning_rate": 1.8800973217388215e-05,
      "loss": 1.0199,
      "step": 950
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1027582909787448,
      "learning_rate": 1.879801314495639e-05,
      "loss": 1.0436,
      "step": 951
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.02836299669084,
      "learning_rate": 1.879504965683294e-05,
      "loss": 1.0345,
      "step": 952
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.063658321758672,
      "learning_rate": 1.8792082754168385e-05,
      "loss": 1.0263,
      "step": 953
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1071258187871345,
      "learning_rate": 1.878911243811459e-05,
      "loss": 1.0277,
      "step": 954
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.060812948371124,
      "learning_rate": 1.8786138709824726e-05,
      "loss": 0.9497,
      "step": 955
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3439381754948594,
      "learning_rate": 1.8783161570453295e-05,
      "loss": 1.0743,
      "step": 956
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.033304418635494,
      "learning_rate": 1.878018102115614e-05,
      "loss": 0.997,
      "step": 957
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.112556414905837,
      "learning_rate": 1.8777197063090394e-05,
      "loss": 0.9626,
      "step": 958
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2522211442568314,
      "learning_rate": 1.877420969741454e-05,
      "loss": 0.9952,
      "step": 959
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.018964164535544,
      "learning_rate": 1.877121892528838e-05,
      "loss": 1.0153,
      "step": 960
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8811266360442912,
      "learning_rate": 1.876822474787303e-05,
      "loss": 0.7024,
      "step": 961
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1225856381178907,
      "learning_rate": 1.8765227166330933e-05,
      "loss": 0.9512,
      "step": 962
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0722467531782647,
      "learning_rate": 1.8762226181825857e-05,
      "loss": 1.0332,
      "step": 963
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2375151267837043,
      "learning_rate": 1.875922179552288e-05,
      "loss": 1.0074,
      "step": 964
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.000609847791507,
      "learning_rate": 1.875621400858842e-05,
      "loss": 1.0028,
      "step": 965
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8748120434407705,
      "learning_rate": 1.875320282219019e-05,
      "loss": 0.6751,
      "step": 966
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.142861939499267,
      "learning_rate": 1.8750188237497247e-05,
      "loss": 0.9922,
      "step": 967
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8334553948306042,
      "learning_rate": 1.874717025567995e-05,
      "loss": 0.652,
      "step": 968
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0448454231743973,
      "learning_rate": 1.874414887790999e-05,
      "loss": 0.9798,
      "step": 969
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9013790592721936,
      "learning_rate": 1.8741124105360363e-05,
      "loss": 0.6684,
      "step": 970
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.178001172139392,
      "learning_rate": 1.873809593920539e-05,
      "loss": 0.9722,
      "step": 971
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.017574625823801,
      "learning_rate": 1.8735064380620717e-05,
      "loss": 0.9658,
      "step": 972
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1258537088155889,
      "learning_rate": 1.873202943078329e-05,
      "loss": 1.0533,
      "step": 973
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9647297456412104,
      "learning_rate": 1.8728991090871387e-05,
      "loss": 1.0075,
      "step": 974
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.940385450033749,
      "learning_rate": 1.8725949362064596e-05,
      "loss": 0.6975,
      "step": 975
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0570022314394072,
      "learning_rate": 1.8722904245543817e-05,
      "loss": 0.9565,
      "step": 976
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.182133180838242,
      "learning_rate": 1.871985574249127e-05,
      "loss": 0.9735,
      "step": 977
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1911513014731339,
      "learning_rate": 1.8716803854090495e-05,
      "loss": 0.9926,
      "step": 978
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.048298254374087,
      "learning_rate": 1.8713748581526334e-05,
      "loss": 1.0269,
      "step": 979
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0676762101503923,
      "learning_rate": 1.871068992598495e-05,
      "loss": 0.9584,
      "step": 980
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0856033917982955,
      "learning_rate": 1.8707627888653816e-05,
      "loss": 1.0426,
      "step": 981
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.029887013374347,
      "learning_rate": 1.8704562470721728e-05,
      "loss": 0.9963,
      "step": 982
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9292970759670975,
      "learning_rate": 1.870149367337878e-05,
      "loss": 0.6778,
      "step": 983
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0365839995127493,
      "learning_rate": 1.8698421497816386e-05,
      "loss": 0.9559,
      "step": 984
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9097885583261294,
      "learning_rate": 1.869534594522727e-05,
      "loss": 1.0301,
      "step": 985
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8517259352902651,
      "learning_rate": 1.8692267016805473e-05,
      "loss": 0.681,
      "step": 986
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0792594275284184,
      "learning_rate": 1.8689184713746333e-05,
      "loss": 1.0234,
      "step": 987
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.015729438129916,
      "learning_rate": 1.868609903724651e-05,
      "loss": 0.9821,
      "step": 988
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9665101464224178,
      "learning_rate": 1.8683009988503972e-05,
      "loss": 0.967,
      "step": 989
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.074481703912726,
      "learning_rate": 1.867991756871799e-05,
      "loss": 0.9911,
      "step": 990
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.207017721791673,
      "learning_rate": 1.867682177908915e-05,
      "loss": 1.0831,
      "step": 991
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1187599455901494,
      "learning_rate": 1.867372262081934e-05,
      "loss": 0.9854,
      "step": 992
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8432832462785647,
      "learning_rate": 1.8670620095111766e-05,
      "loss": 0.6613,
      "step": 993
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0253704142849356,
      "learning_rate": 1.8667514203170934e-05,
      "loss": 0.9457,
      "step": 994
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0177504102826624,
      "learning_rate": 1.8664404946202658e-05,
      "loss": 0.9888,
      "step": 995
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9545926331823233,
      "learning_rate": 1.8661292325414058e-05,
      "loss": 0.9627,
      "step": 996
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9695563847703843,
      "learning_rate": 1.865817634201356e-05,
      "loss": 0.9475,
      "step": 997
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0366591436686365,
      "learning_rate": 1.8655056997210893e-05,
      "loss": 1.0247,
      "step": 998
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1386506064736792,
      "learning_rate": 1.8651934292217097e-05,
      "loss": 1.0248,
      "step": 999
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.151966490252456,
      "learning_rate": 1.864880822824452e-05,
      "loss": 0.9836,
      "step": 1000
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0936688558033854,
      "learning_rate": 1.8645678806506795e-05,
      "loss": 0.9748,
      "step": 1001
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0272550331544166,
      "learning_rate": 1.864254602821888e-05,
      "loss": 0.9473,
      "step": 1002
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.018204575363229,
      "learning_rate": 1.8639409894597026e-05,
      "loss": 0.9512,
      "step": 1003
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0277574532515552,
      "learning_rate": 1.8636270406858786e-05,
      "loss": 1.0012,
      "step": 1004
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.087029203792295,
      "learning_rate": 1.8633127566223023e-05,
      "loss": 0.9777,
      "step": 1005
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0624654040901376,
      "learning_rate": 1.862998137390989e-05,
      "loss": 1.0224,
      "step": 1006
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0914232757014508,
      "learning_rate": 1.8626831831140845e-05,
      "loss": 0.9842,
      "step": 1007
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1019692965490249,
      "learning_rate": 1.8623678939138652e-05,
      "loss": 1.0282,
      "step": 1008
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0993880598446397,
      "learning_rate": 1.8620522699127374e-05,
      "loss": 1.0129,
      "step": 1009
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9965848052422186,
      "learning_rate": 1.8617363112332376e-05,
      "loss": 0.6785,
      "step": 1010
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1515269879391823,
      "learning_rate": 1.8614200179980307e-05,
      "loss": 0.9495,
      "step": 1011
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0172488694440311,
      "learning_rate": 1.8611033903299136e-05,
      "loss": 0.9535,
      "step": 1012
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0605063584352963,
      "learning_rate": 1.8607864283518116e-05,
      "loss": 1.0151,
      "step": 1013
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1276123428204312,
      "learning_rate": 1.8604691321867804e-05,
      "loss": 0.9941,
      "step": 1014
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8312464060338206,
      "learning_rate": 1.8601515019580053e-05,
      "loss": 0.6786,
      "step": 1015
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2050605667155112,
      "learning_rate": 1.8598335377888012e-05,
      "loss": 1.0164,
      "step": 1016
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.021882607520291,
      "learning_rate": 1.8595152398026128e-05,
      "loss": 1.0006,
      "step": 1017
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0293604975423856,
      "learning_rate": 1.8591966081230142e-05,
      "loss": 0.9643,
      "step": 1018
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9966974252359234,
      "learning_rate": 1.8588776428737095e-05,
      "loss": 0.9952,
      "step": 1019
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0941016926047444,
      "learning_rate": 1.858558344178532e-05,
      "loss": 1.019,
      "step": 1020
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1713245473934402,
      "learning_rate": 1.8582387121614437e-05,
      "loss": 0.9382,
      "step": 1021
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0286243642204316,
      "learning_rate": 1.857918746946538e-05,
      "loss": 0.9905,
      "step": 1022
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.243337464057125,
      "learning_rate": 1.8575984486580353e-05,
      "loss": 0.9898,
      "step": 1023
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9788195388880007,
      "learning_rate": 1.857277817420287e-05,
      "loss": 0.6752,
      "step": 1024
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.097569721008369,
      "learning_rate": 1.8569568533577727e-05,
      "loss": 1.0003,
      "step": 1025
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1719061641721629,
      "learning_rate": 1.8566355565951023e-05,
      "loss": 1.0631,
      "step": 1026
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.033025916348958,
      "learning_rate": 1.8563139272570142e-05,
      "loss": 0.9538,
      "step": 1027
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.018207393576156,
      "learning_rate": 1.8559919654683756e-05,
      "loss": 0.9822,
      "step": 1028
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0673532532039374,
      "learning_rate": 1.8556696713541833e-05,
      "loss": 0.9663,
      "step": 1029
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1292266069413106,
      "learning_rate": 1.855347045039563e-05,
      "loss": 0.9967,
      "step": 1030
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9308444639595252,
      "learning_rate": 1.8550240866497697e-05,
      "loss": 1.0204,
      "step": 1031
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0437961857474032,
      "learning_rate": 1.854700796310186e-05,
      "loss": 0.9986,
      "step": 1032
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1493742568465206,
      "learning_rate": 1.8543771741463254e-05,
      "loss": 0.9894,
      "step": 1033
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1899852012836822,
      "learning_rate": 1.8540532202838286e-05,
      "loss": 0.9479,
      "step": 1034
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9815704192423077,
      "learning_rate": 1.8537289348484658e-05,
      "loss": 0.9543,
      "step": 1035
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0243522302187713,
      "learning_rate": 1.8534043179661357e-05,
      "loss": 1.0107,
      "step": 1036
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1810463705616934,
      "learning_rate": 1.8530793697628658e-05,
      "loss": 1.0105,
      "step": 1037
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.002523393107215,
      "learning_rate": 1.8527540903648122e-05,
      "loss": 1.0143,
      "step": 1038
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1622326743407192,
      "learning_rate": 1.8524284798982595e-05,
      "loss": 1.0416,
      "step": 1039
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1055702867988229,
      "learning_rate": 1.852102538489621e-05,
      "loss": 0.959,
      "step": 1040
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0341634394501305,
      "learning_rate": 1.8517762662654383e-05,
      "loss": 0.9591,
      "step": 1041
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8593297805593312,
      "learning_rate": 1.851449663352381e-05,
      "loss": 0.6698,
      "step": 1042
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1491376152222195,
      "learning_rate": 1.851122729877249e-05,
      "loss": 1.0328,
      "step": 1043
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0210541218833007,
      "learning_rate": 1.8507954659669677e-05,
      "loss": 0.9747,
      "step": 1044
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8268070591760103,
      "learning_rate": 1.850467871748593e-05,
      "loss": 0.6263,
      "step": 1045
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0774537526615013,
      "learning_rate": 1.850139947349308e-05,
      "loss": 0.9686,
      "step": 1046
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7898517056117785,
      "learning_rate": 1.8498116928964244e-05,
      "loss": 0.6755,
      "step": 1047
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9950929777758715,
      "learning_rate": 1.849483108517381e-05,
      "loss": 0.9801,
      "step": 1048
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9560585445641867,
      "learning_rate": 1.849154194339747e-05,
      "loss": 1.0156,
      "step": 1049
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0950224780171736,
      "learning_rate": 1.8488249504912173e-05,
      "loss": 0.9659,
      "step": 1050
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.279445655193816,
      "learning_rate": 1.8484953770996163e-05,
      "loss": 0.9852,
      "step": 1051
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1097544026369661,
      "learning_rate": 1.848165474292895e-05,
      "loss": 1.0018,
      "step": 1052
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1444777056713509,
      "learning_rate": 1.8478352421991334e-05,
      "loss": 0.9788,
      "step": 1053
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9137980154128706,
      "learning_rate": 1.847504680946539e-05,
      "loss": 0.9373,
      "step": 1054
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0955069028577191,
      "learning_rate": 1.847173790663447e-05,
      "loss": 0.9419,
      "step": 1055
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1044551278616475,
      "learning_rate": 1.8468425714783206e-05,
      "loss": 1.019,
      "step": 1056
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3185411875359596,
      "learning_rate": 1.84651102351975e-05,
      "loss": 1.0292,
      "step": 1057
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9790320922713452,
      "learning_rate": 1.846179146916454e-05,
      "loss": 0.946,
      "step": 1058
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9746699077885115,
      "learning_rate": 1.8458469417972783e-05,
      "loss": 0.6871,
      "step": 1059
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.186226583362829,
      "learning_rate": 1.8455144082911965e-05,
      "loss": 0.9388,
      "step": 1060
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8395481651254743,
      "learning_rate": 1.8451815465273097e-05,
      "loss": 0.6802,
      "step": 1061
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.996743587685856,
      "learning_rate": 1.8448483566348456e-05,
      "loss": 0.9441,
      "step": 1062
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1049429494781504,
      "learning_rate": 1.8445148387431605e-05,
      "loss": 0.9789,
      "step": 1063
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0377428012755436,
      "learning_rate": 1.8441809929817382e-05,
      "loss": 0.9367,
      "step": 1064
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9955810150522155,
      "learning_rate": 1.8438468194801876e-05,
      "loss": 1.0026,
      "step": 1065
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1477309568741927,
      "learning_rate": 1.8435123183682475e-05,
      "loss": 0.9753,
      "step": 1066
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0743119464891275,
      "learning_rate": 1.8431774897757824e-05,
      "loss": 1.0141,
      "step": 1067
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1253139888416046,
      "learning_rate": 1.8428423338327847e-05,
      "loss": 1.0448,
      "step": 1068
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8525029179687951,
      "learning_rate": 1.8425068506693727e-05,
      "loss": 0.699,
      "step": 1069
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.133670685705529,
      "learning_rate": 1.842171040415793e-05,
      "loss": 1.0291,
      "step": 1070
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1298552811037093,
      "learning_rate": 1.8418349032024185e-05,
      "loss": 1.0104,
      "step": 1071
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1323812749189606,
      "learning_rate": 1.8414984391597492e-05,
      "loss": 1.0241,
      "step": 1072
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0512767903121172,
      "learning_rate": 1.8411616484184126e-05,
      "loss": 1.0086,
      "step": 1073
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0473023454683985,
      "learning_rate": 1.8408245311091618e-05,
      "loss": 0.9598,
      "step": 1074
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0597213503425966,
      "learning_rate": 1.8404870873628774e-05,
      "loss": 1.0525,
      "step": 1075
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1052535409092052,
      "learning_rate": 1.8401493173105675e-05,
      "loss": 0.9429,
      "step": 1076
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0208022717197054,
      "learning_rate": 1.8398112210833648e-05,
      "loss": 1.022,
      "step": 1077
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8314166972751915,
      "learning_rate": 1.8394727988125308e-05,
      "loss": 0.6496,
      "step": 1078
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2174862962468413,
      "learning_rate": 1.8391340506294524e-05,
      "loss": 1.0627,
      "step": 1079
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1438451283304099,
      "learning_rate": 1.8387949766656434e-05,
      "loss": 0.9871,
      "step": 1080
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9867848356331875,
      "learning_rate": 1.8384555770527438e-05,
      "loss": 0.9989,
      "step": 1081
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.119744499764927,
      "learning_rate": 1.8381158519225204e-05,
      "loss": 1.0495,
      "step": 1082
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0867124316493249,
      "learning_rate": 1.8377758014068662e-05,
      "loss": 0.6762,
      "step": 1083
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9626548486404748,
      "learning_rate": 1.8374354256378e-05,
      "loss": 1.0146,
      "step": 1084
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0716048963277134,
      "learning_rate": 1.837094724747468e-05,
      "loss": 0.9965,
      "step": 1085
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2567336304617314,
      "learning_rate": 1.8367536988681422e-05,
      "loss": 1.0048,
      "step": 1086
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4209280930206776,
      "learning_rate": 1.83641234813222e-05,
      "loss": 0.9419,
      "step": 1087
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9226729939283539,
      "learning_rate": 1.8360706726722253e-05,
      "loss": 0.7128,
      "step": 1088
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0361626519194873,
      "learning_rate": 1.835728672620809e-05,
      "loss": 0.9772,
      "step": 1089
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0111500229373864,
      "learning_rate": 1.8353863481107473e-05,
      "loss": 0.9939,
      "step": 1090
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8541406988936189,
      "learning_rate": 1.835043699274942e-05,
      "loss": 0.66,
      "step": 1091
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1715939535451876,
      "learning_rate": 1.8347007262464206e-05,
      "loss": 1.0195,
      "step": 1092
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9259264405594605,
      "learning_rate": 1.8343574291583385e-05,
      "loss": 0.9469,
      "step": 1093
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0648786195281554,
      "learning_rate": 1.8340138081439743e-05,
      "loss": 0.9752,
      "step": 1094
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9483499247243286,
      "learning_rate": 1.833669863336734e-05,
      "loss": 0.9516,
      "step": 1095
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9684310410114311,
      "learning_rate": 1.833325594870148e-05,
      "loss": 0.9859,
      "step": 1096
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.986658864467399,
      "learning_rate": 1.8329810028778747e-05,
      "loss": 0.7048,
      "step": 1097
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9509068333749117,
      "learning_rate": 1.8326360874936952e-05,
      "loss": 0.9925,
      "step": 1098
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0036019484923244,
      "learning_rate": 1.8322908488515182e-05,
      "loss": 0.9815,
      "step": 1099
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9904064224395981,
      "learning_rate": 1.8319452870853772e-05,
      "loss": 0.9488,
      "step": 1100
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0310303863310708,
      "learning_rate": 1.8315994023294306e-05,
      "loss": 1.0297,
      "step": 1101
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9702539828439992,
      "learning_rate": 1.8312531947179634e-05,
      "loss": 0.9528,
      "step": 1102
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.160530743941038,
      "learning_rate": 1.8309066643853854e-05,
      "loss": 0.9877,
      "step": 1103
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0406154195205612,
      "learning_rate": 1.8305598114662312e-05,
      "loss": 0.9885,
      "step": 1104
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1571763687676229,
      "learning_rate": 1.830212636095161e-05,
      "loss": 1.0056,
      "step": 1105
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0905228352617893,
      "learning_rate": 1.8298651384069605e-05,
      "loss": 0.91,
      "step": 1106
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1403088625519011,
      "learning_rate": 1.8295173185365405e-05,
      "loss": 1.0213,
      "step": 1107
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1141928490658557,
      "learning_rate": 1.829169176618936e-05,
      "loss": 0.9574,
      "step": 1108
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.13251280273165,
      "learning_rate": 1.828820712789308e-05,
      "loss": 0.9372,
      "step": 1109
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9803539727573486,
      "learning_rate": 1.828471927182942e-05,
      "loss": 0.9366,
      "step": 1110
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2011240998125534,
      "learning_rate": 1.828122819935249e-05,
      "loss": 0.9788,
      "step": 1111
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1660299309255484,
      "learning_rate": 1.8277733911817642e-05,
      "loss": 1.0398,
      "step": 1112
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9769354780335276,
      "learning_rate": 1.8274236410581478e-05,
      "loss": 0.7401,
      "step": 1113
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1713500178138387,
      "learning_rate": 1.827073569700185e-05,
      "loss": 0.9831,
      "step": 1114
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9446754058678665,
      "learning_rate": 1.8267231772437854e-05,
      "loss": 0.9294,
      "step": 1115
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.204821174731274,
      "learning_rate": 1.8263724638249834e-05,
      "loss": 1.0012,
      "step": 1116
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9153907750313441,
      "learning_rate": 1.8260214295799382e-05,
      "loss": 0.7098,
      "step": 1117
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.142314646166478,
      "learning_rate": 1.825670074644933e-05,
      "loss": 1.042,
      "step": 1118
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.106355987783129,
      "learning_rate": 1.8253183991563768e-05,
      "loss": 0.9888,
      "step": 1119
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0511818660775498,
      "learning_rate": 1.824966403250801e-05,
      "loss": 0.9614,
      "step": 1120
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.045335661431238,
      "learning_rate": 1.8246140870648633e-05,
      "loss": 0.9763,
      "step": 1121
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0673344804914555,
      "learning_rate": 1.8242614507353446e-05,
      "loss": 1.0067,
      "step": 1122
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0007365715944048,
      "learning_rate": 1.8239084943991507e-05,
      "loss": 0.9409,
      "step": 1123
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0580870330130436,
      "learning_rate": 1.823555218193311e-05,
      "loss": 1.0474,
      "step": 1124
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0104033793777256,
      "learning_rate": 1.8232016222549797e-05,
      "loss": 0.9873,
      "step": 1125
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2191609496363105,
      "learning_rate": 1.8228477067214352e-05,
      "loss": 0.9939,
      "step": 1126
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.063665308461004,
      "learning_rate": 1.8224934717300794e-05,
      "loss": 0.9468,
      "step": 1127
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1111577258264926,
      "learning_rate": 1.8221389174184385e-05,
      "loss": 1.0543,
      "step": 1128
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1263901713043643,
      "learning_rate": 1.8217840439241633e-05,
      "loss": 0.9622,
      "step": 1129
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.847167043932759,
      "learning_rate": 1.8214288513850267e-05,
      "loss": 0.698,
      "step": 1130
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8947778893002624,
      "learning_rate": 1.8210733399389277e-05,
      "loss": 0.9225,
      "step": 1131
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1571040555436225,
      "learning_rate": 1.820717509723888e-05,
      "loss": 1.0222,
      "step": 1132
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0337820297180729,
      "learning_rate": 1.8203613608780525e-05,
      "loss": 1.0264,
      "step": 1133
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1290992659258552,
      "learning_rate": 1.8200048935396908e-05,
      "loss": 0.8921,
      "step": 1134
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0293706099775441,
      "learning_rate": 1.819648107847196e-05,
      "loss": 0.9602,
      "step": 1135
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.232152969961051,
      "learning_rate": 1.8192910039390844e-05,
      "loss": 1.0082,
      "step": 1136
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.246435906463253,
      "learning_rate": 1.8189335819539963e-05,
      "loss": 1.0064,
      "step": 1137
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6443820068937406,
      "learning_rate": 1.8185758420306947e-05,
      "loss": 0.3615,
      "step": 1138
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3802213653400555,
      "learning_rate": 1.818217784308067e-05,
      "loss": 0.945,
      "step": 1139
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0634074469991015,
      "learning_rate": 1.817859408925123e-05,
      "loss": 1.0103,
      "step": 1140
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7726181915512361,
      "learning_rate": 1.817500716020997e-05,
      "loss": 0.638,
      "step": 1141
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0065069775833095,
      "learning_rate": 1.8171417057349457e-05,
      "loss": 0.9913,
      "step": 1142
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1427081076950472,
      "learning_rate": 1.816782378206349e-05,
      "loss": 0.9454,
      "step": 1143
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9405508982056013,
      "learning_rate": 1.8164227335747108e-05,
      "loss": 0.8973,
      "step": 1144
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7850463331204454,
      "learning_rate": 1.8160627719796568e-05,
      "loss": 0.6446,
      "step": 1145
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.016119797468945,
      "learning_rate": 1.815702493560937e-05,
      "loss": 0.9699,
      "step": 1146
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0619577030169363,
      "learning_rate": 1.8153418984584238e-05,
      "loss": 0.9519,
      "step": 1147
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9963370154689394,
      "learning_rate": 1.8149809868121125e-05,
      "loss": 0.9768,
      "step": 1148
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9087378445442664,
      "learning_rate": 1.8146197587621217e-05,
      "loss": 0.6398,
      "step": 1149
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0755567900159841,
      "learning_rate": 1.814258214448692e-05,
      "loss": 1.0083,
      "step": 1150
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1237495568844633,
      "learning_rate": 1.8138963540121878e-05,
      "loss": 1.0177,
      "step": 1151
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0540370654120492,
      "learning_rate": 1.813534177593096e-05,
      "loss": 0.941,
      "step": 1152
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9042130306790428,
      "learning_rate": 1.8131716853320254e-05,
      "loss": 0.9544,
      "step": 1153
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9707710403778986,
      "learning_rate": 1.8128088773697086e-05,
      "loss": 0.9946,
      "step": 1154
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.046250015092366,
      "learning_rate": 1.8124457538469996e-05,
      "loss": 0.9686,
      "step": 1155
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0642558308717869,
      "learning_rate": 1.8120823149048753e-05,
      "loss": 1.0081,
      "step": 1156
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8819019933422078,
      "learning_rate": 1.811718560684436e-05,
      "loss": 0.6717,
      "step": 1157
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.026513460826394,
      "learning_rate": 1.8113544913269025e-05,
      "loss": 0.9731,
      "step": 1158
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9369575246024786,
      "learning_rate": 1.8109901069736202e-05,
      "loss": 0.9211,
      "step": 1159
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9762134893938174,
      "learning_rate": 1.8106254077660552e-05,
      "loss": 1.012,
      "step": 1160
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.089719099518422,
      "learning_rate": 1.810260393845796e-05,
      "loss": 1.0622,
      "step": 1161
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.020483095038562,
      "learning_rate": 1.809895065354554e-05,
      "loss": 0.9611,
      "step": 1162
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0000014968670878,
      "learning_rate": 1.8095294224341622e-05,
      "loss": 0.9823,
      "step": 1163
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.35324851306824,
      "learning_rate": 1.8091634652265755e-05,
      "loss": 1.022,
      "step": 1164
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0804798834524314,
      "learning_rate": 1.8087971938738715e-05,
      "loss": 1.0244,
      "step": 1165
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0852286595849414,
      "learning_rate": 1.808430608518249e-05,
      "loss": 1.039,
      "step": 1166
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.020217464406593,
      "learning_rate": 1.808063709302029e-05,
      "loss": 0.9653,
      "step": 1167
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1083667567846829,
      "learning_rate": 1.807696496367655e-05,
      "loss": 1.0053,
      "step": 1168
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0049506378639483,
      "learning_rate": 1.8073289698576913e-05,
      "loss": 0.9308,
      "step": 1169
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8925091918489425,
      "learning_rate": 1.8069611299148236e-05,
      "loss": 0.6991,
      "step": 1170
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0694812710734107,
      "learning_rate": 1.8065929766818617e-05,
      "loss": 0.989,
      "step": 1171
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.089101249508806,
      "learning_rate": 1.806224510301734e-05,
      "loss": 1.0188,
      "step": 1172
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.26014790427264,
      "learning_rate": 1.8058557309174926e-05,
      "loss": 0.9458,
      "step": 1173
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0244109041876241,
      "learning_rate": 1.8054866386723096e-05,
      "loss": 0.9751,
      "step": 1174
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0555004481388384,
      "learning_rate": 1.80511723370948e-05,
      "loss": 0.9512,
      "step": 1175
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9645995207902138,
      "learning_rate": 1.804747516172419e-05,
      "loss": 0.9595,
      "step": 1176
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.071905662614851,
      "learning_rate": 1.8043774862046644e-05,
      "loss": 1.0257,
      "step": 1177
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.089082496144281,
      "learning_rate": 1.804007143949874e-05,
      "loss": 1.0352,
      "step": 1178
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.12143176578893,
      "learning_rate": 1.8036364895518272e-05,
      "loss": 0.9994,
      "step": 1179
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9130036079804614,
      "learning_rate": 1.8032655231544253e-05,
      "loss": 0.9441,
      "step": 1180
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2014022753495102,
      "learning_rate": 1.8028942449016903e-05,
      "loss": 1.0348,
      "step": 1181
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0740297865476178,
      "learning_rate": 1.8025226549377647e-05,
      "loss": 0.9621,
      "step": 1182
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3235461577413714,
      "learning_rate": 1.8021507534069133e-05,
      "loss": 1.0542,
      "step": 1183
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0461742306197352,
      "learning_rate": 1.8017785404535198e-05,
      "loss": 0.9695,
      "step": 1184
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.073692061638113,
      "learning_rate": 1.8014060162220916e-05,
      "loss": 0.9709,
      "step": 1185
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8419424382544164,
      "learning_rate": 1.801033180857254e-05,
      "loss": 0.613,
      "step": 1186
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.105283154832329,
      "learning_rate": 1.8006600345037558e-05,
      "loss": 0.9684,
      "step": 1187
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0823724742099206,
      "learning_rate": 1.8002865773064644e-05,
      "loss": 0.9735,
      "step": 1188
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0290822426489616,
      "learning_rate": 1.799912809410369e-05,
      "loss": 0.9516,
      "step": 1189
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1454014856841468,
      "learning_rate": 1.799538730960579e-05,
      "loss": 0.9768,
      "step": 1190
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0545644114335189,
      "learning_rate": 1.799164342102325e-05,
      "loss": 0.9541,
      "step": 1191
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0479837260485065,
      "learning_rate": 1.7987896429809573e-05,
      "loss": 0.9753,
      "step": 1192
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3123411307020354,
      "learning_rate": 1.798414633741947e-05,
      "loss": 1.0216,
      "step": 1193
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0679703462696215,
      "learning_rate": 1.7980393145308857e-05,
      "loss": 0.9521,
      "step": 1194
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.132009618177653,
      "learning_rate": 1.797663685493485e-05,
      "loss": 1.0512,
      "step": 1195
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.859621996226141,
      "learning_rate": 1.7972877467755777e-05,
      "loss": 0.6698,
      "step": 1196
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2087292367322768,
      "learning_rate": 1.7969114985231152e-05,
      "loss": 0.9977,
      "step": 1197
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1572879408803716,
      "learning_rate": 1.796534940882171e-05,
      "loss": 1.0074,
      "step": 1198
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.041651260416026,
      "learning_rate": 1.7961580739989365e-05,
      "loss": 1.0254,
      "step": 1199
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.05170266972945,
      "learning_rate": 1.795780898019726e-05,
      "loss": 1.0126,
      "step": 1200
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1470273609968125,
      "learning_rate": 1.795403413090971e-05,
      "loss": 1.0106,
      "step": 1201
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.027952689679604,
      "learning_rate": 1.7950256193592243e-05,
      "loss": 0.9859,
      "step": 1202
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9418497547068645,
      "learning_rate": 1.794647516971159e-05,
      "loss": 0.6485,
      "step": 1203
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1368455742353922,
      "learning_rate": 1.7942691060735666e-05,
      "loss": 0.9849,
      "step": 1204
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.131357622658366,
      "learning_rate": 1.79389038681336e-05,
      "loss": 0.9912,
      "step": 1205
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1224643502615812,
      "learning_rate": 1.7935113593375707e-05,
      "loss": 0.9518,
      "step": 1206
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.146068709664852,
      "learning_rate": 1.7931320237933503e-05,
      "loss": 0.9726,
      "step": 1207
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8373683401216165,
      "learning_rate": 1.79275238032797e-05,
      "loss": 0.6421,
      "step": 1208
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0177054005284187,
      "learning_rate": 1.7923724290888205e-05,
      "loss": 0.9489,
      "step": 1209
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0051219089697447,
      "learning_rate": 1.791992170223412e-05,
      "loss": 0.9723,
      "step": 1210
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9866917209102133,
      "learning_rate": 1.791611603879374e-05,
      "loss": 0.9354,
      "step": 1211
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0242661222942266,
      "learning_rate": 1.791230730204455e-05,
      "loss": 0.9796,
      "step": 1212
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.109429863249778,
      "learning_rate": 1.7908495493465236e-05,
      "loss": 0.652,
      "step": 1213
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0859847251482684,
      "learning_rate": 1.7904680614535675e-05,
      "loss": 1.0321,
      "step": 1214
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0306151361294165,
      "learning_rate": 1.7900862666736935e-05,
      "loss": 0.9866,
      "step": 1215
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0934843426184617,
      "learning_rate": 1.789704165155127e-05,
      "loss": 1.007,
      "step": 1216
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0733211280472437,
      "learning_rate": 1.7893217570462134e-05,
      "loss": 0.988,
      "step": 1217
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2106601008966218,
      "learning_rate": 1.7889390424954168e-05,
      "loss": 1.0197,
      "step": 1218
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0700048401257412,
      "learning_rate": 1.78855602165132e-05,
      "loss": 1.0277,
      "step": 1219
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0784241630122753,
      "learning_rate": 1.7881726946626244e-05,
      "loss": 0.9664,
      "step": 1220
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0035481735083565,
      "learning_rate": 1.787789061678151e-05,
      "loss": 0.9999,
      "step": 1221
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9948615692506332,
      "learning_rate": 1.78740512284684e-05,
      "loss": 0.9797,
      "step": 1222
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0756995454904994,
      "learning_rate": 1.787020878317749e-05,
      "loss": 0.9362,
      "step": 1223
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9796413244388256,
      "learning_rate": 1.7866363282400555e-05,
      "loss": 0.9447,
      "step": 1224
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.147253940660453,
      "learning_rate": 1.7862514727630543e-05,
      "loss": 0.6607,
      "step": 1225
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1031506633255481,
      "learning_rate": 1.7858663120361597e-05,
      "loss": 0.9888,
      "step": 1226
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1315458570288586,
      "learning_rate": 1.785480846208905e-05,
      "loss": 1.0158,
      "step": 1227
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0018007804448947,
      "learning_rate": 1.7850950754309405e-05,
      "loss": 1.0237,
      "step": 1228
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0700316164991968,
      "learning_rate": 1.7847089998520365e-05,
      "loss": 0.8947,
      "step": 1229
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.035093741284013,
      "learning_rate": 1.7843226196220803e-05,
      "loss": 0.9872,
      "step": 1230
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.026637279985333,
      "learning_rate": 1.783935934891078e-05,
      "loss": 0.9995,
      "step": 1231
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.116524822947894,
      "learning_rate": 1.7835489458091544e-05,
      "loss": 0.9809,
      "step": 1232
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8258651920581417,
      "learning_rate": 1.7831616525265515e-05,
      "loss": 0.7064,
      "step": 1233
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9748109074487589,
      "learning_rate": 1.7827740551936296e-05,
      "loss": 0.9413,
      "step": 1234
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9445043705060986,
      "learning_rate": 1.7823861539608686e-05,
      "loss": 0.9384,
      "step": 1235
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0687209564405002,
      "learning_rate": 1.7819979489788638e-05,
      "loss": 0.9663,
      "step": 1236
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9474297450146021,
      "learning_rate": 1.7816094403983298e-05,
      "loss": 0.9965,
      "step": 1237
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0273095424835021,
      "learning_rate": 1.7812206283701002e-05,
      "loss": 0.981,
      "step": 1238
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9801881925913157,
      "learning_rate": 1.7808315130451244e-05,
      "loss": 0.937,
      "step": 1239
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9087393869465046,
      "learning_rate": 1.78044209457447e-05,
      "loss": 0.9516,
      "step": 1240
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9503681224441483,
      "learning_rate": 1.7800523731093232e-05,
      "loss": 0.9506,
      "step": 1241
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0843971366556957,
      "learning_rate": 1.7796623488009875e-05,
      "loss": 0.9879,
      "step": 1242
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1033463980246137,
      "learning_rate": 1.7792720218008826e-05,
      "loss": 0.9981,
      "step": 1243
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.074659633327872,
      "learning_rate": 1.7788813922605488e-05,
      "loss": 1.02,
      "step": 1244
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1155216620306485,
      "learning_rate": 1.7784904603316402e-05,
      "loss": 1.0284,
      "step": 1245
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.050179286259498,
      "learning_rate": 1.7780992261659305e-05,
      "loss": 1.0149,
      "step": 1246
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1929481189092352,
      "learning_rate": 1.777707689915311e-05,
      "loss": 1.0129,
      "step": 1247
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.953115677690255,
      "learning_rate": 1.777315851731789e-05,
      "loss": 1.0007,
      "step": 1248
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9659450600388246,
      "learning_rate": 1.7769237117674893e-05,
      "loss": 1.0059,
      "step": 1249
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.084003559793353,
      "learning_rate": 1.7765312701746543e-05,
      "loss": 0.9363,
      "step": 1250
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1834954037582863,
      "learning_rate": 1.7761385271056436e-05,
      "loss": 0.9617,
      "step": 1251
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8529775033942943,
      "learning_rate": 1.7757454827129338e-05,
      "loss": 0.6572,
      "step": 1252
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0590786153404967,
      "learning_rate": 1.7753521371491174e-05,
      "loss": 0.9689,
      "step": 1253
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.149779496545526,
      "learning_rate": 1.7749584905669057e-05,
      "loss": 0.994,
      "step": 1254
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.204412191509424,
      "learning_rate": 1.774564543119125e-05,
      "loss": 0.9833,
      "step": 1255
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0708402945137847,
      "learning_rate": 1.7741702949587196e-05,
      "loss": 1.0065,
      "step": 1256
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.996265975821699,
      "learning_rate": 1.7737757462387507e-05,
      "loss": 0.9853,
      "step": 1257
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.378283184973666,
      "learning_rate": 1.7733808971123946e-05,
      "loss": 0.9579,
      "step": 1258
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0794861136369633,
      "learning_rate": 1.7729857477329463e-05,
      "loss": 1.0161,
      "step": 1259
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0724101489239823,
      "learning_rate": 1.7725902982538162e-05,
      "loss": 1.0363,
      "step": 1260
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1437294079936189,
      "learning_rate": 1.772194548828531e-05,
      "loss": 1.021,
      "step": 1261
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1220767611956566,
      "learning_rate": 1.7717984996107346e-05,
      "loss": 0.9601,
      "step": 1262
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0313843727393954,
      "learning_rate": 1.771402150754187e-05,
      "loss": 0.9554,
      "step": 1263
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1279614047755746,
      "learning_rate": 1.7710055024127637e-05,
      "loss": 1.0265,
      "step": 1264
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9644469878071749,
      "learning_rate": 1.7706085547404582e-05,
      "loss": 0.9164,
      "step": 1265
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8678752100301654,
      "learning_rate": 1.770211307891379e-05,
      "loss": 0.7136,
      "step": 1266
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9668529607418458,
      "learning_rate": 1.769813762019751e-05,
      "loss": 0.9838,
      "step": 1267
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9677730376495859,
      "learning_rate": 1.769415917279915e-05,
      "loss": 0.9933,
      "step": 1268
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0653375033717916,
      "learning_rate": 1.7690177738263284e-05,
      "loss": 1.0003,
      "step": 1269
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9375911824766042,
      "learning_rate": 1.7686193318135635e-05,
      "loss": 0.994,
      "step": 1270
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9873225837060927,
      "learning_rate": 1.76822059139631e-05,
      "loss": 1.0365,
      "step": 1271
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8230636671516361,
      "learning_rate": 1.7678215527293724e-05,
      "loss": 0.6466,
      "step": 1272
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0037183643759604,
      "learning_rate": 1.767422215967671e-05,
      "loss": 0.9343,
      "step": 1273
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7986694409459643,
      "learning_rate": 1.767022581266242e-05,
      "loss": 0.6899,
      "step": 1274
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0901892374649145,
      "learning_rate": 1.766622648780238e-05,
      "loss": 0.9763,
      "step": 1275
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0212813059513552,
      "learning_rate": 1.766222418664926e-05,
      "loss": 0.9494,
      "step": 1276
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0838381283067187,
      "learning_rate": 1.765821891075689e-05,
      "loss": 1.0371,
      "step": 1277
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1389506139539303,
      "learning_rate": 1.7654210661680263e-05,
      "loss": 1.0222,
      "step": 1278
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.988060486962096,
      "learning_rate": 1.765019944097551e-05,
      "loss": 0.9847,
      "step": 1279
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.094985619255203,
      "learning_rate": 1.7646185250199936e-05,
      "loss": 0.9357,
      "step": 1280
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1524243193278452,
      "learning_rate": 1.7642168090911976e-05,
      "loss": 0.9871,
      "step": 1281
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0058773930333935,
      "learning_rate": 1.763814796467124e-05,
      "loss": 0.993,
      "step": 1282
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0535790123963142,
      "learning_rate": 1.763412487303847e-05,
      "loss": 0.9538,
      "step": 1283
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9881613671865868,
      "learning_rate": 1.7630098817575578e-05,
      "loss": 0.9293,
      "step": 1284
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0598227094826485,
      "learning_rate": 1.762606979984561e-05,
      "loss": 1.0062,
      "step": 1285
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9495093820292638,
      "learning_rate": 1.7622037821412775e-05,
      "loss": 0.9605,
      "step": 1286
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.979526280169225,
      "learning_rate": 1.7618002883842426e-05,
      "loss": 0.9337,
      "step": 1287
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0319316784756183,
      "learning_rate": 1.7613964988701057e-05,
      "loss": 0.978,
      "step": 1288
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.018085827859465,
      "learning_rate": 1.7609924137556326e-05,
      "loss": 0.9761,
      "step": 1289
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1779991476199019,
      "learning_rate": 1.7605880331977022e-05,
      "loss": 1.0519,
      "step": 1290
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.065108502920036,
      "learning_rate": 1.76018335735331e-05,
      "loss": 1.0025,
      "step": 1291
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8315025973305785,
      "learning_rate": 1.7597783863795644e-05,
      "loss": 0.7249,
      "step": 1292
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8558648767603517,
      "learning_rate": 1.7593731204336895e-05,
      "loss": 0.6552,
      "step": 1293
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0746390331166444,
      "learning_rate": 1.7589675596730233e-05,
      "loss": 0.9775,
      "step": 1294
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0667223124661473,
      "learning_rate": 1.758561704255018e-05,
      "loss": 1.0105,
      "step": 1295
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1768066985711696,
      "learning_rate": 1.7581555543372413e-05,
      "loss": 1.0029,
      "step": 1296
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.193230068155575,
      "learning_rate": 1.7577491100773744e-05,
      "loss": 1.0492,
      "step": 1297
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0492252258759913,
      "learning_rate": 1.7573423716332128e-05,
      "loss": 0.9847,
      "step": 1298
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0772701505492916,
      "learning_rate": 1.7569353391626665e-05,
      "loss": 0.945,
      "step": 1299
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9037987872353717,
      "learning_rate": 1.7565280128237595e-05,
      "loss": 0.9909,
      "step": 1300
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.075685534430029,
      "learning_rate": 1.75612039277463e-05,
      "loss": 0.979,
      "step": 1301
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0191787008780995,
      "learning_rate": 1.75571247917353e-05,
      "loss": 0.9556,
      "step": 1302
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9576506002184492,
      "learning_rate": 1.7553042721788255e-05,
      "loss": 0.6217,
      "step": 1303
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1003963859166601,
      "learning_rate": 1.754895771948997e-05,
      "loss": 0.9741,
      "step": 1304
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2305074188991014,
      "learning_rate": 1.754486978642637e-05,
      "loss": 1.0648,
      "step": 1305
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.063577867727944,
      "learning_rate": 1.7540778924184553e-05,
      "loss": 1.0455,
      "step": 1306
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0836734342945435,
      "learning_rate": 1.7536685134352717e-05,
      "loss": 0.9374,
      "step": 1307
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1796769457859686,
      "learning_rate": 1.7532588418520215e-05,
      "loss": 0.8922,
      "step": 1308
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.162002476014983,
      "learning_rate": 1.7528488778277535e-05,
      "loss": 1.0307,
      "step": 1309
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0324532763782255,
      "learning_rate": 1.75243862152163e-05,
      "loss": 0.9657,
      "step": 1310
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0216863939132261,
      "learning_rate": 1.752028073092926e-05,
      "loss": 0.9926,
      "step": 1311
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8238178053110518,
      "learning_rate": 1.7516172327010314e-05,
      "loss": 0.6782,
      "step": 1312
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.33917930764309,
      "learning_rate": 1.751206100505448e-05,
      "loss": 1.0658,
      "step": 1313
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9768939171138256,
      "learning_rate": 1.7507946766657914e-05,
      "loss": 0.9645,
      "step": 1314
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9963851204539365,
      "learning_rate": 1.7503829613417905e-05,
      "loss": 0.9284,
      "step": 1315
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2343781856378058,
      "learning_rate": 1.749970954693288e-05,
      "loss": 0.9757,
      "step": 1316
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3813436148426945,
      "learning_rate": 1.7495586568802384e-05,
      "loss": 1.0547,
      "step": 1317
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1135284289548806,
      "learning_rate": 1.7491460680627105e-05,
      "loss": 1.052,
      "step": 1318
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.149679954956521,
      "learning_rate": 1.7487331884008845e-05,
      "loss": 1.0372,
      "step": 1319
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1273152225834475,
      "learning_rate": 1.7483200180550554e-05,
      "loss": 0.9843,
      "step": 1320
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.117837108469692,
      "learning_rate": 1.74790655718563e-05,
      "loss": 1.0357,
      "step": 1321
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0293273661130897,
      "learning_rate": 1.747492805953128e-05,
      "loss": 0.9859,
      "step": 1322
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9293236998648661,
      "learning_rate": 1.7470787645181818e-05,
      "loss": 0.9992,
      "step": 1323
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0954646800509287,
      "learning_rate": 1.7466644330415362e-05,
      "loss": 1.0184,
      "step": 1324
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.029481154482043,
      "learning_rate": 1.7462498116840496e-05,
      "loss": 0.9809,
      "step": 1325
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.03961014701294,
      "learning_rate": 1.745834900606692e-05,
      "loss": 0.9384,
      "step": 1326
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9587878202553475,
      "learning_rate": 1.7454196999705458e-05,
      "loss": 0.9265,
      "step": 1327
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9100551178107802,
      "learning_rate": 1.7450042099368066e-05,
      "loss": 0.9703,
      "step": 1328
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9388690223640207,
      "learning_rate": 1.7445884306667823e-05,
      "loss": 0.9315,
      "step": 1329
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0389849467573193,
      "learning_rate": 1.7441723623218917e-05,
      "loss": 0.9676,
      "step": 1330
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.985901439668714,
      "learning_rate": 1.7437560050636678e-05,
      "loss": 0.9406,
      "step": 1331
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9157275741329556,
      "learning_rate": 1.7433393590537543e-05,
      "loss": 0.6944,
      "step": 1332
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7849841878256706,
      "learning_rate": 1.7429224244539077e-05,
      "loss": 0.6111,
      "step": 1333
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0230644896299181,
      "learning_rate": 1.7425052014259965e-05,
      "loss": 0.9965,
      "step": 1334
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0654206902747325,
      "learning_rate": 1.7420876901320006e-05,
      "loss": 0.9406,
      "step": 1335
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.825857445897112,
      "learning_rate": 1.7416698907340128e-05,
      "loss": 0.6018,
      "step": 1336
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9448868336494014,
      "learning_rate": 1.741251803394237e-05,
      "loss": 0.9709,
      "step": 1337
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8981514468864719,
      "learning_rate": 1.740833428274989e-05,
      "loss": 0.9662,
      "step": 1338
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0557969550694952,
      "learning_rate": 1.7404147655386966e-05,
      "loss": 0.9195,
      "step": 1339
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0580867646636571,
      "learning_rate": 1.739995815347899e-05,
      "loss": 1.0647,
      "step": 1340
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9903189607238635,
      "learning_rate": 1.739576577865247e-05,
      "loss": 0.9831,
      "step": 1341
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1558108748412586,
      "learning_rate": 1.739157053253503e-05,
      "loss": 1.0112,
      "step": 1342
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9731412260427201,
      "learning_rate": 1.738737241675541e-05,
      "loss": 0.9559,
      "step": 1343
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8864549805651406,
      "learning_rate": 1.7383171432943466e-05,
      "loss": 0.9484,
      "step": 1344
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8874961621241799,
      "learning_rate": 1.737896758273016e-05,
      "loss": 0.6759,
      "step": 1345
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9735689668373884,
      "learning_rate": 1.7374760867747574e-05,
      "loss": 0.9682,
      "step": 1346
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.971681574000152,
      "learning_rate": 1.7370551289628895e-05,
      "loss": 0.9137,
      "step": 1347
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0374367245802747,
      "learning_rate": 1.7366338850008432e-05,
      "loss": 0.9821,
      "step": 1348
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0450313629963195,
      "learning_rate": 1.73621235505216e-05,
      "loss": 0.9091,
      "step": 1349
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9887192529681557,
      "learning_rate": 1.7357905392804918e-05,
      "loss": 0.8765,
      "step": 1350
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0345880969058132,
      "learning_rate": 1.735368437849602e-05,
      "loss": 0.9552,
      "step": 1351
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0448407041359962,
      "learning_rate": 1.7349460509233654e-05,
      "loss": 0.9783,
      "step": 1352
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0633121232862741,
      "learning_rate": 1.734523378665767e-05,
      "loss": 1.0496,
      "step": 1353
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.135345490770955,
      "learning_rate": 1.7341004212409026e-05,
      "loss": 0.9818,
      "step": 1354
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0260519539254631,
      "learning_rate": 1.7336771788129785e-05,
      "loss": 0.9907,
      "step": 1355
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.76469146992917,
      "learning_rate": 1.7332536515463126e-05,
      "loss": 0.6556,
      "step": 1356
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9532228776487741,
      "learning_rate": 1.7328298396053324e-05,
      "loss": 0.9892,
      "step": 1357
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1287217317330063,
      "learning_rate": 1.7324057431545768e-05,
      "loss": 0.9335,
      "step": 1358
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.039131657236765,
      "learning_rate": 1.7319813623586935e-05,
      "loss": 0.9457,
      "step": 1359
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1372898283775024,
      "learning_rate": 1.7315566973824433e-05,
      "loss": 1.0052,
      "step": 1360
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0788542010882285,
      "learning_rate": 1.7311317483906946e-05,
      "loss": 0.9955,
      "step": 1361
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1127489055297344,
      "learning_rate": 1.730706515548427e-05,
      "loss": 0.9943,
      "step": 1362
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.093665215918668,
      "learning_rate": 1.730280999020732e-05,
      "loss": 1.0103,
      "step": 1363
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1016588120341986,
      "learning_rate": 1.729855198972808e-05,
      "loss": 1.027,
      "step": 1364
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9893222271959606,
      "learning_rate": 1.729429115569967e-05,
      "loss": 1.0116,
      "step": 1365
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.835480403485345,
      "learning_rate": 1.729002748977628e-05,
      "loss": 0.6682,
      "step": 1366
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0896745002578045,
      "learning_rate": 1.7285760993613215e-05,
      "loss": 0.9768,
      "step": 1367
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1623832007377142,
      "learning_rate": 1.7281491668866874e-05,
      "loss": 0.9499,
      "step": 1368
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9992385210425526,
      "learning_rate": 1.727721951719476e-05,
      "loss": 0.9832,
      "step": 1369
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9555358465178306,
      "learning_rate": 1.7272944540255468e-05,
      "loss": 0.9751,
      "step": 1370
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1319873427135303,
      "learning_rate": 1.726866673970869e-05,
      "loss": 1.0183,
      "step": 1371
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9564394915586485,
      "learning_rate": 1.7264386117215216e-05,
      "loss": 0.9545,
      "step": 1372
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.062277640669776,
      "learning_rate": 1.7260102674436933e-05,
      "loss": 0.9683,
      "step": 1373
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9544473517331433,
      "learning_rate": 1.7255816413036818e-05,
      "loss": 0.8811,
      "step": 1374
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0452300958519456,
      "learning_rate": 1.7251527334678946e-05,
      "loss": 1.0013,
      "step": 1375
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9790712823300398,
      "learning_rate": 1.7247235441028486e-05,
      "loss": 0.6164,
      "step": 1376
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9614233290106797,
      "learning_rate": 1.7242940733751696e-05,
      "loss": 0.9437,
      "step": 1377
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.953737324944904,
      "learning_rate": 1.7238643214515934e-05,
      "loss": 1.0,
      "step": 1378
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0250180321363342,
      "learning_rate": 1.7234342884989642e-05,
      "loss": 0.8934,
      "step": 1379
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1714327883834004,
      "learning_rate": 1.7230039746842352e-05,
      "loss": 1.0153,
      "step": 1380
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0615524711998794,
      "learning_rate": 1.7225733801744698e-05,
      "loss": 0.9883,
      "step": 1381
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0801718869047265,
      "learning_rate": 1.7221425051368394e-05,
      "loss": 0.9892,
      "step": 1382
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0081331919666443,
      "learning_rate": 1.7217113497386245e-05,
      "loss": 1.0204,
      "step": 1383
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1632203343555259,
      "learning_rate": 1.721279914147214e-05,
      "loss": 0.9162,
      "step": 1384
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9831981795970529,
      "learning_rate": 1.7208481985301065e-05,
      "loss": 0.9803,
      "step": 1385
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0283457924058066,
      "learning_rate": 1.7204162030549093e-05,
      "loss": 1.0196,
      "step": 1386
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0680616360976654,
      "learning_rate": 1.7199839278893368e-05,
      "loss": 0.6777,
      "step": 1387
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0596972426843658,
      "learning_rate": 1.719551373201214e-05,
      "loss": 0.9838,
      "step": 1388
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9223079934924691,
      "learning_rate": 1.7191185391584736e-05,
      "loss": 0.7174,
      "step": 1389
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0281395039995576,
      "learning_rate": 1.7186854259291558e-05,
      "loss": 1.0174,
      "step": 1390
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1721908929360223,
      "learning_rate": 1.7182520336814105e-05,
      "loss": 1.0132,
      "step": 1391
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9029748843195381,
      "learning_rate": 1.717818362583496e-05,
      "loss": 0.9931,
      "step": 1392
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9975315677101948,
      "learning_rate": 1.7173844128037777e-05,
      "loss": 0.9529,
      "step": 1393
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0309402768799993,
      "learning_rate": 1.71695018451073e-05,
      "loss": 0.682,
      "step": 1394
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.914697683794776,
      "learning_rate": 1.7165156778729355e-05,
      "loss": 0.9309,
      "step": 1395
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8874596399902822,
      "learning_rate": 1.7160808930590845e-05,
      "loss": 0.715,
      "step": 1396
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.06878140650685,
      "learning_rate": 1.7156458302379753e-05,
      "loss": 1.0426,
      "step": 1397
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0262137759103247,
      "learning_rate": 1.7152104895785147e-05,
      "loss": 1.0139,
      "step": 1398
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2256146830345098,
      "learning_rate": 1.7147748712497162e-05,
      "loss": 0.9745,
      "step": 1399
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0047027562816269,
      "learning_rate": 1.7143389754207026e-05,
      "loss": 0.9558,
      "step": 1400
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9453577073179947,
      "learning_rate": 1.713902802260703e-05,
      "loss": 0.6352,
      "step": 1401
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8334043834492945,
      "learning_rate": 1.7134663519390557e-05,
      "loss": 0.6383,
      "step": 1402
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.098334989446273,
      "learning_rate": 1.7130296246252048e-05,
      "loss": 0.9681,
      "step": 1403
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1315638998598954,
      "learning_rate": 1.7125926204887034e-05,
      "loss": 1.0048,
      "step": 1404
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0865706667568829,
      "learning_rate": 1.712155339699211e-05,
      "loss": 0.9961,
      "step": 1405
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7871120693323956,
      "learning_rate": 1.7117177824264962e-05,
      "loss": 0.6325,
      "step": 1406
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2216913156390536,
      "learning_rate": 1.7112799488404327e-05,
      "loss": 0.986,
      "step": 1407
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0190877109497039,
      "learning_rate": 1.7108418391110033e-05,
      "loss": 0.9759,
      "step": 1408
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2804901272837867,
      "learning_rate": 1.7104034534082968e-05,
      "loss": 0.9267,
      "step": 1409
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8746606033378297,
      "learning_rate": 1.7099647919025096e-05,
      "loss": 0.645,
      "step": 1410
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.020352788494129,
      "learning_rate": 1.7095258547639456e-05,
      "loss": 1.038,
      "step": 1411
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9481367888445391,
      "learning_rate": 1.709086642163015e-05,
      "loss": 0.9615,
      "step": 1412
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.058334959428144,
      "learning_rate": 1.7086471542702355e-05,
      "loss": 0.9581,
      "step": 1413
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0312521931776655,
      "learning_rate": 1.708207391256231e-05,
      "loss": 0.9675,
      "step": 1414
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1449773747140255,
      "learning_rate": 1.707767353291733e-05,
      "loss": 0.9935,
      "step": 1415
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0090102809920913,
      "learning_rate": 1.7073270405475796e-05,
      "loss": 0.9842,
      "step": 1416
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.099178291749185,
      "learning_rate": 1.7068864531947147e-05,
      "loss": 0.9978,
      "step": 1417
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1341972907838358,
      "learning_rate": 1.70644559140419e-05,
      "loss": 0.9752,
      "step": 1418
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.085417312019273,
      "learning_rate": 1.706004455347163e-05,
      "loss": 0.9986,
      "step": 1419
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9880738217372029,
      "learning_rate": 1.705563045194898e-05,
      "loss": 0.9769,
      "step": 1420
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9215072321553329,
      "learning_rate": 1.7051213611187657e-05,
      "loss": 0.6403,
      "step": 1421
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8900576149344396,
      "learning_rate": 1.704679403290243e-05,
      "loss": 0.683,
      "step": 1422
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9388489140830658,
      "learning_rate": 1.7042371718809132e-05,
      "loss": 0.9277,
      "step": 1423
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.908082309757416,
      "learning_rate": 1.7037946670624652e-05,
      "loss": 0.6714,
      "step": 1424
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1681686333192338,
      "learning_rate": 1.7033518890066956e-05,
      "loss": 0.9266,
      "step": 1425
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9902902705877681,
      "learning_rate": 1.7029088378855055e-05,
      "loss": 0.9811,
      "step": 1426
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0004445331023182,
      "learning_rate": 1.7024655138709025e-05,
      "loss": 0.9523,
      "step": 1427
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0139027204728048,
      "learning_rate": 1.7020219171350004e-05,
      "loss": 0.9852,
      "step": 1428
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0715549773808875,
      "learning_rate": 1.7015780478500187e-05,
      "loss": 0.9927,
      "step": 1429
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0581533861686643,
      "learning_rate": 1.701133906188283e-05,
      "loss": 0.9657,
      "step": 1430
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0076996027103244,
      "learning_rate": 1.700689492322224e-05,
      "loss": 0.9242,
      "step": 1431
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0972109329102926,
      "learning_rate": 1.700244806424379e-05,
      "loss": 1.0011,
      "step": 1432
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.203398083682969,
      "learning_rate": 1.6997998486673893e-05,
      "loss": 1.0261,
      "step": 1433
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9977541795737497,
      "learning_rate": 1.699354619224004e-05,
      "loss": 1.0092,
      "step": 1434
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1521719055829869,
      "learning_rate": 1.698909118267076e-05,
      "loss": 0.992,
      "step": 1435
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9959259695978188,
      "learning_rate": 1.6984633459695646e-05,
      "loss": 0.676,
      "step": 1436
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.088840460021598,
      "learning_rate": 1.6980173025045328e-05,
      "loss": 1.0015,
      "step": 1437
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0279302851363352,
      "learning_rate": 1.697570988045151e-05,
      "loss": 0.936,
      "step": 1438
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9898653817146535,
      "learning_rate": 1.6971244027646937e-05,
      "loss": 0.9396,
      "step": 1439
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.04659879378168,
      "learning_rate": 1.69667754683654e-05,
      "loss": 0.9738,
      "step": 1440
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9808231338803901,
      "learning_rate": 1.6962304204341758e-05,
      "loss": 0.9952,
      "step": 1441
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2167037682173028,
      "learning_rate": 1.6957830237311904e-05,
      "loss": 0.9667,
      "step": 1442
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0652911339135152,
      "learning_rate": 1.6953353569012784e-05,
      "loss": 0.9794,
      "step": 1443
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9891938694183696,
      "learning_rate": 1.6948874201182402e-05,
      "loss": 1.0036,
      "step": 1444
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0419294575048452,
      "learning_rate": 1.6944392135559798e-05,
      "loss": 0.9517,
      "step": 1445
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.081691287658249,
      "learning_rate": 1.6939907373885062e-05,
      "loss": 0.9532,
      "step": 1446
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.979638505914835,
      "learning_rate": 1.6935419917899335e-05,
      "loss": 0.9244,
      "step": 1447
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8435911690263926,
      "learning_rate": 1.6930929769344807e-05,
      "loss": 0.6862,
      "step": 1448
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.035084439881812,
      "learning_rate": 1.69264369299647e-05,
      "loss": 0.9473,
      "step": 1449
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9020152013775594,
      "learning_rate": 1.692194140150329e-05,
      "loss": 0.694,
      "step": 1450
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0732609360613188,
      "learning_rate": 1.69174431857059e-05,
      "loss": 1.0001,
      "step": 1451
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1525316372140746,
      "learning_rate": 1.6912942284318898e-05,
      "loss": 1.0976,
      "step": 1452
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0898551555775058,
      "learning_rate": 1.6908438699089674e-05,
      "loss": 0.9984,
      "step": 1453
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7567439226389934,
      "learning_rate": 1.690393243176668e-05,
      "loss": 0.3518,
      "step": 1454
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7897266393798166,
      "learning_rate": 1.6899423484099413e-05,
      "loss": 0.6535,
      "step": 1455
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0451731665985264,
      "learning_rate": 1.6894911857838394e-05,
      "loss": 0.9403,
      "step": 1456
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0797030584943634,
      "learning_rate": 1.689039755473519e-05,
      "loss": 0.9835,
      "step": 1457
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.209227581397346,
      "learning_rate": 1.6885880576542417e-05,
      "loss": 0.9736,
      "step": 1458
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.972635327826534,
      "learning_rate": 1.6881360925013712e-05,
      "loss": 0.9698,
      "step": 1459
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9590501689871723,
      "learning_rate": 1.6876838601903765e-05,
      "loss": 0.9647,
      "step": 1460
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8226786829522152,
      "learning_rate": 1.6872313608968296e-05,
      "loss": 0.659,
      "step": 1461
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9006106644091924,
      "learning_rate": 1.6867785947964065e-05,
      "loss": 0.6933,
      "step": 1462
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9851663073178379,
      "learning_rate": 1.6863255620648866e-05,
      "loss": 0.9389,
      "step": 1463
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9779833925649258,
      "learning_rate": 1.685872262878152e-05,
      "loss": 0.9644,
      "step": 1464
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1324429178388196,
      "learning_rate": 1.6854186974121903e-05,
      "loss": 0.9968,
      "step": 1465
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1057375018446076,
      "learning_rate": 1.68496486584309e-05,
      "loss": 0.9449,
      "step": 1466
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9499174985601271,
      "learning_rate": 1.6845107683470453e-05,
      "loss": 0.9292,
      "step": 1467
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7937554354380654,
      "learning_rate": 1.6840564051003517e-05,
      "loss": 0.6644,
      "step": 1468
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9802738177345695,
      "learning_rate": 1.6836017762794087e-05,
      "loss": 0.9523,
      "step": 1469
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2641449117349142,
      "learning_rate": 1.6831468820607192e-05,
      "loss": 0.651,
      "step": 1470
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9773946671159375,
      "learning_rate": 1.6826917226208886e-05,
      "loss": 0.9445,
      "step": 1471
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2168480897155014,
      "learning_rate": 1.6822362981366257e-05,
      "loss": 1.0403,
      "step": 1472
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0632392019193635,
      "learning_rate": 1.6817806087847417e-05,
      "loss": 1.0189,
      "step": 1473
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.000599242829687,
      "learning_rate": 1.681324654742151e-05,
      "loss": 0.9367,
      "step": 1474
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1262467900431812,
      "learning_rate": 1.6808684361858706e-05,
      "loss": 0.9461,
      "step": 1475
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9471705477477169,
      "learning_rate": 1.6804119532930202e-05,
      "loss": 1.0098,
      "step": 1476
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7365872767546413,
      "learning_rate": 1.6799552062408225e-05,
      "loss": 0.56,
      "step": 1477
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0036509920830063,
      "learning_rate": 1.6794981952066018e-05,
      "loss": 0.9691,
      "step": 1478
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1456330414669045,
      "learning_rate": 1.6790409203677863e-05,
      "loss": 0.964,
      "step": 1479
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.056197632388694,
      "learning_rate": 1.6785833819019052e-05,
      "loss": 0.986,
      "step": 1480
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0048103898413465,
      "learning_rate": 1.678125579986591e-05,
      "loss": 0.9857,
      "step": 1481
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0620035543895272,
      "learning_rate": 1.677667514799578e-05,
      "loss": 0.9547,
      "step": 1482
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.025395683548575,
      "learning_rate": 1.6772091865187032e-05,
      "loss": 0.9863,
      "step": 1483
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9863728763323665,
      "learning_rate": 1.676750595321905e-05,
      "loss": 1.0201,
      "step": 1484
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7380166873541875,
      "learning_rate": 1.6762917413872246e-05,
      "loss": 0.6248,
      "step": 1485
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.018141132009077,
      "learning_rate": 1.675832624892805e-05,
      "loss": 0.9708,
      "step": 1486
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.008820048592125,
      "learning_rate": 1.6753732460168907e-05,
      "loss": 0.9756,
      "step": 1487
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9974922353227484,
      "learning_rate": 1.674913604937828e-05,
      "loss": 1.011,
      "step": 1488
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9544609703575772,
      "learning_rate": 1.6744537018340662e-05,
      "loss": 0.9757,
      "step": 1489
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0096528757703034,
      "learning_rate": 1.6739935368841555e-05,
      "loss": 1.0553,
      "step": 1490
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0814161809742369,
      "learning_rate": 1.6735331102667475e-05,
      "loss": 0.9721,
      "step": 1491
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8796077699550896,
      "learning_rate": 1.6730724221605955e-05,
      "loss": 0.6944,
      "step": 1492
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0485909525025803,
      "learning_rate": 1.6726114727445547e-05,
      "loss": 0.9447,
      "step": 1493
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.104161153457556,
      "learning_rate": 1.6721502621975813e-05,
      "loss": 0.9828,
      "step": 1494
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9463482561188868,
      "learning_rate": 1.6716887906987332e-05,
      "loss": 1.0127,
      "step": 1495
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9847954626000465,
      "learning_rate": 1.6712270584271703e-05,
      "loss": 0.9508,
      "step": 1496
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1326762787536926,
      "learning_rate": 1.670765065562152e-05,
      "loss": 1.0008,
      "step": 1497
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0097401360259077,
      "learning_rate": 1.67030281228304e-05,
      "loss": 0.9764,
      "step": 1498
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0253708060937818,
      "learning_rate": 1.6698402987692968e-05,
      "loss": 0.9344,
      "step": 1499
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8941412894574384,
      "learning_rate": 1.6693775252004866e-05,
      "loss": 0.9141,
      "step": 1500
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0188077977204082,
      "learning_rate": 1.668914491756274e-05,
      "loss": 1.0006,
      "step": 1501
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1050850763503446,
      "learning_rate": 1.668451198616424e-05,
      "loss": 0.9753,
      "step": 1502
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0761705072293875,
      "learning_rate": 1.6679876459608033e-05,
      "loss": 0.9706,
      "step": 1503
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0366508265798373,
      "learning_rate": 1.667523833969379e-05,
      "loss": 0.9754,
      "step": 1504
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1640406162889965,
      "learning_rate": 1.667059762822219e-05,
      "loss": 0.9554,
      "step": 1505
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.990443865047959,
      "learning_rate": 1.666595432699491e-05,
      "loss": 0.9507,
      "step": 1506
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2558830679472457,
      "learning_rate": 1.6661308437814652e-05,
      "loss": 1.017,
      "step": 1507
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0300197876651558,
      "learning_rate": 1.6656659962485097e-05,
      "loss": 0.9604,
      "step": 1508
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0666694115540396,
      "learning_rate": 1.6652008902810952e-05,
      "loss": 0.9913,
      "step": 1509
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0172948368588186,
      "learning_rate": 1.6647355260597915e-05,
      "loss": 0.9903,
      "step": 1510
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9969537540291366,
      "learning_rate": 1.664269903765269e-05,
      "loss": 0.9133,
      "step": 1511
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9388480115017003,
      "learning_rate": 1.6638040235782983e-05,
      "loss": 0.6657,
      "step": 1512
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0356456542110157,
      "learning_rate": 1.6633378856797505e-05,
      "loss": 0.9815,
      "step": 1513
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.081141721854011,
      "learning_rate": 1.662871490250596e-05,
      "loss": 0.9514,
      "step": 1514
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0918907166143064,
      "learning_rate": 1.662404837471905e-05,
      "loss": 1.0389,
      "step": 1515
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0504885092026945,
      "learning_rate": 1.66193792752485e-05,
      "loss": 0.9993,
      "step": 1516
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1118076491230908,
      "learning_rate": 1.6614707605906995e-05,
      "loss": 0.9081,
      "step": 1517
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8705480066389394,
      "learning_rate": 1.661003336850825e-05,
      "loss": 0.9749,
      "step": 1518
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8841502339462933,
      "learning_rate": 1.660535656486696e-05,
      "loss": 0.9339,
      "step": 1519
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8049959081008878,
      "learning_rate": 1.660067719679882e-05,
      "loss": 0.6178,
      "step": 1520
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0615993911694148,
      "learning_rate": 1.6595995266120528e-05,
      "loss": 0.9502,
      "step": 1521
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9782285159646611,
      "learning_rate": 1.6591310774649766e-05,
      "loss": 0.9668,
      "step": 1522
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8123865250348419,
      "learning_rate": 1.6586623724205216e-05,
      "loss": 0.7037,
      "step": 1523
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9481182519524296,
      "learning_rate": 1.6581934116606554e-05,
      "loss": 0.9797,
      "step": 1524
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0242108099303775,
      "learning_rate": 1.657724195367444e-05,
      "loss": 1.0263,
      "step": 1525
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4377723444282269,
      "learning_rate": 1.657254723723054e-05,
      "loss": 0.9251,
      "step": 1526
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0046117978308196,
      "learning_rate": 1.6567849969097505e-05,
      "loss": 0.9951,
      "step": 1527
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1809126719985075,
      "learning_rate": 1.6563150151098973e-05,
      "loss": 0.9842,
      "step": 1528
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1003879389748716,
      "learning_rate": 1.6558447785059577e-05,
      "loss": 0.9715,
      "step": 1529
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.215961219864573,
      "learning_rate": 1.655374287280494e-05,
      "loss": 0.9966,
      "step": 1530
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0706433682995855,
      "learning_rate": 1.6549035416161662e-05,
      "loss": 0.9916,
      "step": 1531
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0346227393406626,
      "learning_rate": 1.654432541695735e-05,
      "loss": 0.9637,
      "step": 1532
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0884706753505342,
      "learning_rate": 1.653961287702058e-05,
      "loss": 0.9632,
      "step": 1533
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8724467665360683,
      "learning_rate": 1.653489779818093e-05,
      "loss": 0.9482,
      "step": 1534
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1298443726716074,
      "learning_rate": 1.6530180182268946e-05,
      "loss": 1.0186,
      "step": 1535
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1104676170056178,
      "learning_rate": 1.652546003111618e-05,
      "loss": 1.0129,
      "step": 1536
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1200278066910978,
      "learning_rate": 1.652073734655515e-05,
      "loss": 0.9301,
      "step": 1537
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0685139043935956,
      "learning_rate": 1.6516012130419366e-05,
      "loss": 0.9913,
      "step": 1538
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1669538000808062,
      "learning_rate": 1.6511284384543317e-05,
      "loss": 1.0507,
      "step": 1539
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9082275211738964,
      "learning_rate": 1.6506554110762483e-05,
      "loss": 0.9177,
      "step": 1540
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9798155674615285,
      "learning_rate": 1.650182131091332e-05,
      "loss": 0.9125,
      "step": 1541
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1739314584425797,
      "learning_rate": 1.6497085986833252e-05,
      "loss": 1.0017,
      "step": 1542
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9434866466420797,
      "learning_rate": 1.6492348140360704e-05,
      "loss": 0.939,
      "step": 1543
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0436416383684852,
      "learning_rate": 1.6487607773335074e-05,
      "loss": 0.9535,
      "step": 1544
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9845004733807848,
      "learning_rate": 1.648286488759673e-05,
      "loss": 0.9523,
      "step": 1545
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1035060554064142,
      "learning_rate": 1.6478119484987026e-05,
      "loss": 0.9341,
      "step": 1546
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9186569517990653,
      "learning_rate": 1.6473371567348287e-05,
      "loss": 0.6544,
      "step": 1547
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.059100381077552,
      "learning_rate": 1.6468621136523823e-05,
      "loss": 0.9716,
      "step": 1548
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.112989622953563,
      "learning_rate": 1.646386819435791e-05,
      "loss": 0.9628,
      "step": 1549
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1396470776756562,
      "learning_rate": 1.6459112742695807e-05,
      "loss": 0.925,
      "step": 1550
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9819859372665444,
      "learning_rate": 1.6454354783383748e-05,
      "loss": 0.9431,
      "step": 1551
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8757456009327508,
      "learning_rate": 1.644959431826893e-05,
      "loss": 0.6249,
      "step": 1552
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9054180911095577,
      "learning_rate": 1.6444831349199528e-05,
      "loss": 0.9402,
      "step": 1553
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1804005905312087,
      "learning_rate": 1.6440065878024697e-05,
      "loss": 0.979,
      "step": 1554
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0487633432950658,
      "learning_rate": 1.6435297906594553e-05,
      "loss": 0.9817,
      "step": 1555
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1055263465950445,
      "learning_rate": 1.643052743676019e-05,
      "loss": 0.9528,
      "step": 1556
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.192588133751209,
      "learning_rate": 1.6425754470373667e-05,
      "loss": 0.9573,
      "step": 1557
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9452053341336425,
      "learning_rate": 1.642097900928801e-05,
      "loss": 0.6399,
      "step": 1558
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8131707095617441,
      "learning_rate": 1.6416201055357225e-05,
      "loss": 0.6439,
      "step": 1559
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.820798543147497,
      "learning_rate": 1.641142061043627e-05,
      "loss": 0.6681,
      "step": 1560
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8098323513556751,
      "learning_rate": 1.640663767638108e-05,
      "loss": 0.646,
      "step": 1561
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0514945713121442,
      "learning_rate": 1.6401852255048564e-05,
      "loss": 0.9719,
      "step": 1562
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7803286400414017,
      "learning_rate": 1.6397064348296578e-05,
      "loss": 0.6537,
      "step": 1563
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9424461018602247,
      "learning_rate": 1.6392273957983955e-05,
      "loss": 0.6893,
      "step": 1564
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2093310587037978,
      "learning_rate": 1.638748108597049e-05,
      "loss": 0.9858,
      "step": 1565
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0406045765320653,
      "learning_rate": 1.6382685734116934e-05,
      "loss": 0.9384,
      "step": 1566
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0238636641063177,
      "learning_rate": 1.6377887904285018e-05,
      "loss": 0.9992,
      "step": 1567
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0396009810911067,
      "learning_rate": 1.637308759833742e-05,
      "loss": 0.9512,
      "step": 1568
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9618954953226768,
      "learning_rate": 1.6368284818137787e-05,
      "loss": 0.9786,
      "step": 1569
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9456096160817027,
      "learning_rate": 1.636347956555072e-05,
      "loss": 0.9519,
      "step": 1570
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8469437882464992,
      "learning_rate": 1.635867184244178e-05,
      "loss": 0.6697,
      "step": 1571
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1265514228480569,
      "learning_rate": 1.63538616506775e-05,
      "loss": 1.0106,
      "step": 1572
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0310184045308866,
      "learning_rate": 1.6349048992125358e-05,
      "loss": 0.9909,
      "step": 1573
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.171856026992591,
      "learning_rate": 1.634423386865379e-05,
      "loss": 0.9732,
      "step": 1574
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0546635742034485,
      "learning_rate": 1.6339416282132196e-05,
      "loss": 0.9924,
      "step": 1575
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0032979900441807,
      "learning_rate": 1.633459623443093e-05,
      "loss": 0.983,
      "step": 1576
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8614123202172655,
      "learning_rate": 1.6329773727421297e-05,
      "loss": 0.6973,
      "step": 1577
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9584357282243517,
      "learning_rate": 1.6324948762975567e-05,
      "loss": 0.9728,
      "step": 1578
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0674528755967925,
      "learning_rate": 1.632012134296695e-05,
      "loss": 0.9607,
      "step": 1579
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3851936259649,
      "learning_rate": 1.6315291469269617e-05,
      "loss": 1.0111,
      "step": 1580
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9695861840611761,
      "learning_rate": 1.63104591437587e-05,
      "loss": 0.9495,
      "step": 1581
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0411696505710653,
      "learning_rate": 1.6305624368310265e-05,
      "loss": 0.9205,
      "step": 1582
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0747486695278798,
      "learning_rate": 1.630078714480134e-05,
      "loss": 0.9778,
      "step": 1583
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0619727208661556,
      "learning_rate": 1.6295947475109904e-05,
      "loss": 0.9883,
      "step": 1584
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0221987112963165,
      "learning_rate": 1.629110536111488e-05,
      "loss": 0.9757,
      "step": 1585
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9750582079818525,
      "learning_rate": 1.628626080469615e-05,
      "loss": 0.9753,
      "step": 1586
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9753080918428795,
      "learning_rate": 1.628141380773453e-05,
      "loss": 0.9549,
      "step": 1587
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9637614931885939,
      "learning_rate": 1.6276564372111797e-05,
      "loss": 0.968,
      "step": 1588
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0887582946862666,
      "learning_rate": 1.6271712499710663e-05,
      "loss": 1.0367,
      "step": 1589
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9735672914589272,
      "learning_rate": 1.62668581924148e-05,
      "loss": 0.99,
      "step": 1590
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9756810420688973,
      "learning_rate": 1.6262001452108807e-05,
      "loss": 0.9254,
      "step": 1591
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0258140595669938,
      "learning_rate": 1.6257142280678247e-05,
      "loss": 0.942,
      "step": 1592
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1153754242823601,
      "learning_rate": 1.6252280680009613e-05,
      "loss": 0.9446,
      "step": 1593
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0027429299784651,
      "learning_rate": 1.6247416651990343e-05,
      "loss": 0.9676,
      "step": 1594
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9888450359432278,
      "learning_rate": 1.624255019850883e-05,
      "loss": 0.9173,
      "step": 1595
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9854885139414403,
      "learning_rate": 1.6237681321454387e-05,
      "loss": 1.0016,
      "step": 1596
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9903574562326202,
      "learning_rate": 1.623281002271729e-05,
      "loss": 0.942,
      "step": 1597
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0890635629803356,
      "learning_rate": 1.6227936304188738e-05,
      "loss": 0.9312,
      "step": 1598
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1136584484700254,
      "learning_rate": 1.622306016776088e-05,
      "loss": 0.9748,
      "step": 1599
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9046618863479882,
      "learning_rate": 1.6218181615326795e-05,
      "loss": 0.9615,
      "step": 1600
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9903231994538662,
      "learning_rate": 1.6213300648780515e-05,
      "loss": 0.9738,
      "step": 1601
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9276117710614492,
      "learning_rate": 1.620841727001699e-05,
      "loss": 0.9056,
      "step": 1602
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0470053243384005,
      "learning_rate": 1.6203531480932114e-05,
      "loss": 0.9261,
      "step": 1603
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9393356017014453,
      "learning_rate": 1.619864328342273e-05,
      "loss": 0.9546,
      "step": 1604
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9096152834370488,
      "learning_rate": 1.6193752679386593e-05,
      "loss": 0.9365,
      "step": 1605
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9837700161691014,
      "learning_rate": 1.6188859670722414e-05,
      "loss": 0.9181,
      "step": 1606
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1607738436076744,
      "learning_rate": 1.6183964259329817e-05,
      "loss": 0.9323,
      "step": 1607
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.054120539007942,
      "learning_rate": 1.6179066447109376e-05,
      "loss": 0.9548,
      "step": 1608
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9785129407647906,
      "learning_rate": 1.6174166235962588e-05,
      "loss": 0.8844,
      "step": 1609
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0736712831489177,
      "learning_rate": 1.6169263627791886e-05,
      "loss": 0.9727,
      "step": 1610
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2542712918389805,
      "learning_rate": 1.616435862450063e-05,
      "loss": 1.0079,
      "step": 1611
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8044620846029171,
      "learning_rate": 1.615945122799311e-05,
      "loss": 0.6785,
      "step": 1612
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.043414965755875,
      "learning_rate": 1.6154541440174547e-05,
      "loss": 0.9744,
      "step": 1613
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8413351729196806,
      "learning_rate": 1.614962926295109e-05,
      "loss": 0.6549,
      "step": 1614
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6105094178155354,
      "learning_rate": 1.6144714698229814e-05,
      "loss": 0.3551,
      "step": 1615
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9860738192735853,
      "learning_rate": 1.6139797747918725e-05,
      "loss": 0.9149,
      "step": 1616
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0656690368020276,
      "learning_rate": 1.613487841392675e-05,
      "loss": 0.9511,
      "step": 1617
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9241815099303592,
      "learning_rate": 1.612995669816375e-05,
      "loss": 0.9548,
      "step": 1618
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9468282803432361,
      "learning_rate": 1.6125032602540492e-05,
      "loss": 0.9281,
      "step": 1619
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.139295015720058,
      "learning_rate": 1.6120106128968686e-05,
      "loss": 0.8625,
      "step": 1620
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9744717570319779,
      "learning_rate": 1.6115177279360965e-05,
      "loss": 0.9761,
      "step": 1621
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0193156702071897,
      "learning_rate": 1.611024605563087e-05,
      "loss": 0.9587,
      "step": 1622
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0955978052245452,
      "learning_rate": 1.610531245969287e-05,
      "loss": 0.8973,
      "step": 1623
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0251166002515137,
      "learning_rate": 1.6100376493462368e-05,
      "loss": 0.9722,
      "step": 1624
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.125202203003455,
      "learning_rate": 1.6095438158855668e-05,
      "loss": 1.0045,
      "step": 1625
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9793997962997281,
      "learning_rate": 1.609049745779e-05,
      "loss": 0.9442,
      "step": 1626
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.122577803548193,
      "learning_rate": 1.6085554392183517e-05,
      "loss": 0.9517,
      "step": 1627
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.157561591196132,
      "learning_rate": 1.608060896395529e-05,
      "loss": 0.9362,
      "step": 1628
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0604944254575668,
      "learning_rate": 1.60756611750253e-05,
      "loss": 0.9283,
      "step": 1629
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0690063933334732,
      "learning_rate": 1.6070711027314446e-05,
      "loss": 1.0033,
      "step": 1630
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.008261405277575,
      "learning_rate": 1.606575852274456e-05,
      "loss": 0.9719,
      "step": 1631
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9751079554723052,
      "learning_rate": 1.6060803663238357e-05,
      "loss": 0.985,
      "step": 1632
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9481231585773454,
      "learning_rate": 1.6055846450719498e-05,
      "loss": 0.9367,
      "step": 1633
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0205274110034661,
      "learning_rate": 1.6050886887112535e-05,
      "loss": 0.9396,
      "step": 1634
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2391602050873836,
      "learning_rate": 1.6045924974342945e-05,
      "loss": 0.991,
      "step": 1635
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.099285620336005,
      "learning_rate": 1.604096071433711e-05,
      "loss": 0.9207,
      "step": 1636
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9828354401575848,
      "learning_rate": 1.6035994109022333e-05,
      "loss": 0.9886,
      "step": 1637
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.914241867164748,
      "learning_rate": 1.6031025160326814e-05,
      "loss": 0.9235,
      "step": 1638
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0971317209937324,
      "learning_rate": 1.6026053870179678e-05,
      "loss": 1.0057,
      "step": 1639
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9174335231971825,
      "learning_rate": 1.6021080240510943e-05,
      "loss": 0.9186,
      "step": 1640
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8314210241751996,
      "learning_rate": 1.601610427325155e-05,
      "loss": 0.6695,
      "step": 1641
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.21315739181217,
      "learning_rate": 1.6011125970333333e-05,
      "loss": 0.958,
      "step": 1642
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.808215092659811,
      "learning_rate": 1.600614533368905e-05,
      "loss": 0.6256,
      "step": 1643
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0028875328396145,
      "learning_rate": 1.6001162365252348e-05,
      "loss": 0.9778,
      "step": 1644
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7693307477413324,
      "learning_rate": 1.5996177066957787e-05,
      "loss": 0.6015,
      "step": 1645
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0240201441578909,
      "learning_rate": 1.5991189440740838e-05,
      "loss": 0.9088,
      "step": 1646
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9970094034494368,
      "learning_rate": 1.5986199488537867e-05,
      "loss": 0.9532,
      "step": 1647
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1823062966499918,
      "learning_rate": 1.598120721228614e-05,
      "loss": 0.9774,
      "step": 1648
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.000919079758282,
      "learning_rate": 1.5976212613923836e-05,
      "loss": 0.9491,
      "step": 1649
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1549051062202889,
      "learning_rate": 1.5971215695390026e-05,
      "loss": 0.9858,
      "step": 1650
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.928801205256441,
      "learning_rate": 1.5966216458624692e-05,
      "loss": 0.9681,
      "step": 1651
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0131381948480045,
      "learning_rate": 1.5961214905568705e-05,
      "loss": 0.9198,
      "step": 1652
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2473742104307408,
      "learning_rate": 1.595621103816384e-05,
      "loss": 1.0015,
      "step": 1653
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3012316734074547,
      "learning_rate": 1.5951204858352772e-05,
      "loss": 0.9096,
      "step": 1654
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9062547416398541,
      "learning_rate": 1.594619636807907e-05,
      "loss": 0.8649,
      "step": 1655
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.262805643022486,
      "learning_rate": 1.5941185569287206e-05,
      "loss": 1.0034,
      "step": 1656
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1925470396656486,
      "learning_rate": 1.5936172463922542e-05,
      "loss": 1.0667,
      "step": 1657
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.085333837759487,
      "learning_rate": 1.593115705393134e-05,
      "loss": 0.9358,
      "step": 1658
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8381788776655699,
      "learning_rate": 1.5926139341260755e-05,
      "loss": 0.6847,
      "step": 1659
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9214486749213248,
      "learning_rate": 1.5921119327858835e-05,
      "loss": 0.9358,
      "step": 1660
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9662589240645184,
      "learning_rate": 1.5916097015674518e-05,
      "loss": 1.0057,
      "step": 1661
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9742105617404693,
      "learning_rate": 1.5911072406657646e-05,
      "loss": 0.9157,
      "step": 1662
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0812958320891406,
      "learning_rate": 1.5906045502758943e-05,
      "loss": 0.9879,
      "step": 1663
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9686828852485496,
      "learning_rate": 1.590101630593002e-05,
      "loss": 0.975,
      "step": 1664
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2329895107301407,
      "learning_rate": 1.5895984818123392e-05,
      "loss": 0.9824,
      "step": 1665
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7792348805664642,
      "learning_rate": 1.5890951041292453e-05,
      "loss": 0.6235,
      "step": 1666
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9671793146347475,
      "learning_rate": 1.588591497739149e-05,
      "loss": 0.9877,
      "step": 1667
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0574311537929135,
      "learning_rate": 1.5880876628375668e-05,
      "loss": 0.9468,
      "step": 1668
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0497506719931537,
      "learning_rate": 1.587583599620106e-05,
      "loss": 0.9665,
      "step": 1669
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0195454935836423,
      "learning_rate": 1.5870793082824604e-05,
      "loss": 0.8524,
      "step": 1670
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1321275321453925,
      "learning_rate": 1.5865747890204138e-05,
      "loss": 0.9897,
      "step": 1671
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1203254927562785,
      "learning_rate": 1.5860700420298377e-05,
      "loss": 1.0213,
      "step": 1672
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1139599017798356,
      "learning_rate": 1.5855650675066924e-05,
      "loss": 0.9845,
      "step": 1673
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0499746107486791,
      "learning_rate": 1.5850598656470265e-05,
      "loss": 0.9013,
      "step": 1674
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9600958112140636,
      "learning_rate": 1.584554436646976e-05,
      "loss": 0.9586,
      "step": 1675
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.994099877656896,
      "learning_rate": 1.5840487807027665e-05,
      "loss": 0.9858,
      "step": 1676
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9284776473332909,
      "learning_rate": 1.5835428980107113e-05,
      "loss": 0.9252,
      "step": 1677
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9777408690699838,
      "learning_rate": 1.583036788767211e-05,
      "loss": 0.973,
      "step": 1678
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9296106268652073,
      "learning_rate": 1.5825304531687548e-05,
      "loss": 0.6003,
      "step": 1679
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.150272860028633,
      "learning_rate": 1.5820238914119195e-05,
      "loss": 0.9627,
      "step": 1680
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0648423557226885,
      "learning_rate": 1.5815171036933697e-05,
      "loss": 1.0024,
      "step": 1681
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9324707235797833,
      "learning_rate": 1.5810100902098582e-05,
      "loss": 0.9303,
      "step": 1682
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1209628511812504,
      "learning_rate": 1.580502851158225e-05,
      "loss": 1.0232,
      "step": 1683
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8085330615557997,
      "learning_rate": 1.5799953867353975e-05,
      "loss": 0.6417,
      "step": 1684
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9942749732227157,
      "learning_rate": 1.579487697138391e-05,
      "loss": 0.9361,
      "step": 1685
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0882488306064526,
      "learning_rate": 1.5789797825643086e-05,
      "loss": 1.0329,
      "step": 1686
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9010023371075441,
      "learning_rate": 1.5784716432103394e-05,
      "loss": 0.6342,
      "step": 1687
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2485662207694215,
      "learning_rate": 1.5779632792737608e-05,
      "loss": 0.6657,
      "step": 1688
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1195336547557535,
      "learning_rate": 1.5774546909519376e-05,
      "loss": 0.9326,
      "step": 1689
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0672804837068304,
      "learning_rate": 1.5769458784423206e-05,
      "loss": 0.9678,
      "step": 1690
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0012091287401086,
      "learning_rate": 1.5764368419424488e-05,
      "loss": 0.9339,
      "step": 1691
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2317484390682547,
      "learning_rate": 1.575927581649948e-05,
      "loss": 1.0508,
      "step": 1692
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.283729782270879,
      "learning_rate": 1.5754180977625303e-05,
      "loss": 0.9919,
      "step": 1693
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0846487078008804,
      "learning_rate": 1.574908390477995e-05,
      "loss": 0.9865,
      "step": 1694
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9935988090771469,
      "learning_rate": 1.5743984599942273e-05,
      "loss": 0.9704,
      "step": 1695
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.996703675474974,
      "learning_rate": 1.5738883065092005e-05,
      "loss": 0.9314,
      "step": 1696
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9768741295222063,
      "learning_rate": 1.5733779302209735e-05,
      "loss": 0.9507,
      "step": 1697
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0632438737885392,
      "learning_rate": 1.572867331327692e-05,
      "loss": 0.9974,
      "step": 1698
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.079238345250093,
      "learning_rate": 1.5723565100275884e-05,
      "loss": 0.9275,
      "step": 1699
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0111074519011938,
      "learning_rate": 1.5718454665189806e-05,
      "loss": 0.9772,
      "step": 1700
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9630232810513198,
      "learning_rate": 1.5713342010002733e-05,
      "loss": 0.9059,
      "step": 1701
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9469153656279207,
      "learning_rate": 1.5708227136699578e-05,
      "loss": 1.0044,
      "step": 1702
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.027388307698419,
      "learning_rate": 1.5703110047266105e-05,
      "loss": 0.9854,
      "step": 1703
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.949900672975249,
      "learning_rate": 1.569799074368895e-05,
      "loss": 0.8678,
      "step": 1704
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9586768853095338,
      "learning_rate": 1.5692869227955603e-05,
      "loss": 0.9103,
      "step": 1705
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0223321004615145,
      "learning_rate": 1.5687745502054407e-05,
      "loss": 1.0472,
      "step": 1706
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0844258865549208,
      "learning_rate": 1.5682619567974575e-05,
      "loss": 0.9813,
      "step": 1707
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9270839746550911,
      "learning_rate": 1.567749142770617e-05,
      "loss": 0.6561,
      "step": 1708
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1185259087791826,
      "learning_rate": 1.5672361083240106e-05,
      "loss": 0.8857,
      "step": 1709
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9854335937436944,
      "learning_rate": 1.5667228536568167e-05,
      "loss": 0.9165,
      "step": 1710
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9259854371452818,
      "learning_rate": 1.566209378968298e-05,
      "loss": 0.9407,
      "step": 1711
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0899601291187921,
      "learning_rate": 1.565695684457803e-05,
      "loss": 0.9576,
      "step": 1712
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2025322979143098,
      "learning_rate": 1.5651817703247666e-05,
      "loss": 0.9475,
      "step": 1713
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9354633968994015,
      "learning_rate": 1.5646676367687067e-05,
      "loss": 0.9806,
      "step": 1714
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.071895839162586,
      "learning_rate": 1.564153283989228e-05,
      "loss": 1.0001,
      "step": 1715
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9337359852757239,
      "learning_rate": 1.5636387121860207e-05,
      "loss": 0.9917,
      "step": 1716
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9580917738413104,
      "learning_rate": 1.5631239215588578e-05,
      "loss": 0.9582,
      "step": 1717
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9394187652060223,
      "learning_rate": 1.5626089123076004e-05,
      "loss": 0.9579,
      "step": 1718
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0403543293318909,
      "learning_rate": 1.5620936846321917e-05,
      "loss": 0.9688,
      "step": 1719
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.097679955847574,
      "learning_rate": 1.561578238732661e-05,
      "loss": 0.9435,
      "step": 1720
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.113743249080542,
      "learning_rate": 1.561062574809123e-05,
      "loss": 0.9421,
      "step": 1721
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8840974433058667,
      "learning_rate": 1.5605466930617747e-05,
      "loss": 0.6566,
      "step": 1722
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.094021869189686,
      "learning_rate": 1.5600305936909005e-05,
      "loss": 0.9229,
      "step": 1723
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2477788503051939,
      "learning_rate": 1.559514276896867e-05,
      "loss": 0.992,
      "step": 1724
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.085359068973814,
      "learning_rate": 1.558997742880127e-05,
      "loss": 0.9857,
      "step": 1725
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1256230492905246,
      "learning_rate": 1.5584809918412158e-05,
      "loss": 0.715,
      "step": 1726
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1999891672881005,
      "learning_rate": 1.557964023980755e-05,
      "loss": 1.0152,
      "step": 1727
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0549346061484763,
      "learning_rate": 1.5574468394994486e-05,
      "loss": 1.0072,
      "step": 1728
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9885854263569103,
      "learning_rate": 1.5569294385980856e-05,
      "loss": 0.9176,
      "step": 1729
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9974039439738627,
      "learning_rate": 1.556411821477539e-05,
      "loss": 1.0072,
      "step": 1730
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0894339093787269,
      "learning_rate": 1.5558939883387657e-05,
      "loss": 0.9362,
      "step": 1731
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0721936495957802,
      "learning_rate": 1.5553759393828058e-05,
      "loss": 1.0273,
      "step": 1732
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9723123964515811,
      "learning_rate": 1.554857674810784e-05,
      "loss": 1.0214,
      "step": 1733
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1582336606701205,
      "learning_rate": 1.554339194823909e-05,
      "loss": 0.9852,
      "step": 1734
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0956178921427755,
      "learning_rate": 1.553820499623472e-05,
      "loss": 0.9399,
      "step": 1735
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0969443467889417,
      "learning_rate": 1.553301589410848e-05,
      "loss": 1.0004,
      "step": 1736
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9647513314678468,
      "learning_rate": 1.5527824643874968e-05,
      "loss": 0.9748,
      "step": 1737
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0760836318438924,
      "learning_rate": 1.5522631247549598e-05,
      "loss": 0.992,
      "step": 1738
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0591663628863262,
      "learning_rate": 1.5517435707148628e-05,
      "loss": 0.9373,
      "step": 1739
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.113354045596051,
      "learning_rate": 1.5512238024689144e-05,
      "loss": 1.0158,
      "step": 1740
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0573769361018843,
      "learning_rate": 1.550703820218907e-05,
      "loss": 0.987,
      "step": 1741
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.065118547999642,
      "learning_rate": 1.550183624166715e-05,
      "loss": 0.969,
      "step": 1742
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1089363208832046,
      "learning_rate": 1.549663214514297e-05,
      "loss": 0.9658,
      "step": 1743
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1138466335260095,
      "learning_rate": 1.5491425914636934e-05,
      "loss": 1.0186,
      "step": 1744
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0348237802679774,
      "learning_rate": 1.5486217552170283e-05,
      "loss": 1.0416,
      "step": 1745
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1167617666911114,
      "learning_rate": 1.548100705976508e-05,
      "loss": 1.0152,
      "step": 1746
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0531692214996655,
      "learning_rate": 1.5475794439444226e-05,
      "loss": 0.9596,
      "step": 1747
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.07545361590721,
      "learning_rate": 1.5470579693231432e-05,
      "loss": 1.0093,
      "step": 1748
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9726810331279993,
      "learning_rate": 1.5465362823151245e-05,
      "loss": 0.993,
      "step": 1749
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8855971076389677,
      "learning_rate": 1.5460143831229026e-05,
      "loss": 0.6096,
      "step": 1750
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0734766418995176,
      "learning_rate": 1.545492271949098e-05,
      "loss": 1.0246,
      "step": 1751
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1629758880064447,
      "learning_rate": 1.544969948996411e-05,
      "loss": 0.9803,
      "step": 1752
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8044233810247862,
      "learning_rate": 1.544447414467626e-05,
      "loss": 0.6101,
      "step": 1753
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0075579222833428,
      "learning_rate": 1.5439246685656093e-05,
      "loss": 0.935,
      "step": 1754
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0922338406944916,
      "learning_rate": 1.5434017114933082e-05,
      "loss": 1.007,
      "step": 1755
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9942553910419768,
      "learning_rate": 1.5428785434537527e-05,
      "loss": 0.9729,
      "step": 1756
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9384406022854351,
      "learning_rate": 1.542355164650055e-05,
      "loss": 0.9442,
      "step": 1757
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0178931731525136,
      "learning_rate": 1.541831575285408e-05,
      "loss": 0.9174,
      "step": 1758
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1712887231242033,
      "learning_rate": 1.541307775563088e-05,
      "loss": 1.0105,
      "step": 1759
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9469526716546641,
      "learning_rate": 1.540783765686452e-05,
      "loss": 0.9644,
      "step": 1760
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0583056442958274,
      "learning_rate": 1.540259545858938e-05,
      "loss": 0.9826,
      "step": 1761
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8254161465865409,
      "learning_rate": 1.539735116284067e-05,
      "loss": 0.6381,
      "step": 1762
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.067469599442253,
      "learning_rate": 1.53921047716544e-05,
      "loss": 0.9677,
      "step": 1763
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9416887699798777,
      "learning_rate": 1.53868562870674e-05,
      "loss": 0.998,
      "step": 1764
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0761785850990726,
      "learning_rate": 1.5381605711117318e-05,
      "loss": 0.9355,
      "step": 1765
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9667363671672796,
      "learning_rate": 1.5376353045842604e-05,
      "loss": 0.935,
      "step": 1766
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0129565249926256,
      "learning_rate": 1.5371098293282526e-05,
      "loss": 0.9334,
      "step": 1767
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0578247163460188,
      "learning_rate": 1.5365841455477158e-05,
      "loss": 1.0309,
      "step": 1768
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9120924775010357,
      "learning_rate": 1.5360582534467382e-05,
      "loss": 1.0004,
      "step": 1769
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0347870516961732,
      "learning_rate": 1.5355321532294897e-05,
      "loss": 0.9492,
      "step": 1770
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0827760792465404,
      "learning_rate": 1.5350058451002204e-05,
      "loss": 0.9198,
      "step": 1771
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9893439308143626,
      "learning_rate": 1.5344793292632614e-05,
      "loss": 0.9511,
      "step": 1772
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8706745676816455,
      "learning_rate": 1.533952605923024e-05,
      "loss": 0.9294,
      "step": 1773
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8897672847564347,
      "learning_rate": 1.5334256752840007e-05,
      "loss": 0.9248,
      "step": 1774
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.009424219900404,
      "learning_rate": 1.532898537550764e-05,
      "loss": 0.9646,
      "step": 1775
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9440705997641395,
      "learning_rate": 1.532371192927966e-05,
      "loss": 0.9395,
      "step": 1776
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1461272061066754,
      "learning_rate": 1.5318436416203412e-05,
      "loss": 1.042,
      "step": 1777
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9594958136338632,
      "learning_rate": 1.531315883832703e-05,
      "loss": 0.8883,
      "step": 1778
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1758729096579654,
      "learning_rate": 1.530787919769945e-05,
      "loss": 0.9832,
      "step": 1779
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9440536335940265,
      "learning_rate": 1.5302597496370408e-05,
      "loss": 0.9249,
      "step": 1780
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0932821265083967,
      "learning_rate": 1.5297313736390447e-05,
      "loss": 0.995,
      "step": 1781
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.062228996587738,
      "learning_rate": 1.5292027919810898e-05,
      "loss": 0.9673,
      "step": 1782
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0949909182822666,
      "learning_rate": 1.52867400486839e-05,
      "loss": 1.0292,
      "step": 1783
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9687009110866927,
      "learning_rate": 1.528145012506239e-05,
      "loss": 0.9579,
      "step": 1784
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1316200145836843,
      "learning_rate": 1.5276158151000096e-05,
      "loss": 0.9673,
      "step": 1785
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2133907451858,
      "learning_rate": 1.5270864128551542e-05,
      "loss": 0.9775,
      "step": 1786
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9101539272065873,
      "learning_rate": 1.5265568059772053e-05,
      "loss": 0.9566,
      "step": 1787
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9633378367169824,
      "learning_rate": 1.5260269946717746e-05,
      "loss": 0.9491,
      "step": 1788
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9378227435824172,
      "learning_rate": 1.5254969791445526e-05,
      "loss": 0.9261,
      "step": 1789
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0014225267548937,
      "learning_rate": 1.5249667596013102e-05,
      "loss": 1.0419,
      "step": 1790
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0032957565292664,
      "learning_rate": 1.5244363362478967e-05,
      "loss": 0.9302,
      "step": 1791
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9573794893034485,
      "learning_rate": 1.5239057092902404e-05,
      "loss": 0.981,
      "step": 1792
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0915648321711229,
      "learning_rate": 1.523374878934349e-05,
      "loss": 0.9557,
      "step": 1793
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8085197098668706,
      "learning_rate": 1.5228438453863095e-05,
      "loss": 0.655,
      "step": 1794
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.037692177338067,
      "learning_rate": 1.522312608852287e-05,
      "loss": 1.0202,
      "step": 1795
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0854132382911295,
      "learning_rate": 1.5217811695385263e-05,
      "loss": 0.9615,
      "step": 1796
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0993582443414935,
      "learning_rate": 1.52124952765135e-05,
      "loss": 1.0028,
      "step": 1797
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.011633948828017,
      "learning_rate": 1.5207176833971598e-05,
      "loss": 1.0142,
      "step": 1798
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.073721178733671,
      "learning_rate": 1.520185636982436e-05,
      "loss": 0.9949,
      "step": 1799
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1325141096515754,
      "learning_rate": 1.5196533886137376e-05,
      "loss": 0.9931,
      "step": 1800
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9402573769063322,
      "learning_rate": 1.5191209384977014e-05,
      "loss": 0.9602,
      "step": 1801
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8633702378452169,
      "learning_rate": 1.5185882868410431e-05,
      "loss": 0.6396,
      "step": 1802
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7542796461038154,
      "learning_rate": 1.5180554338505564e-05,
      "loss": 0.6602,
      "step": 1803
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8018273228774319,
      "learning_rate": 1.517522379733113e-05,
      "loss": 0.6843,
      "step": 1804
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9977225745604883,
      "learning_rate": 1.5169891246956629e-05,
      "loss": 0.918,
      "step": 1805
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0716078846942454,
      "learning_rate": 1.5164556689452346e-05,
      "loss": 0.944,
      "step": 1806
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9487390968128715,
      "learning_rate": 1.5159220126889329e-05,
      "loss": 0.9385,
      "step": 1807
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8392429161225662,
      "learning_rate": 1.5153881561339426e-05,
      "loss": 0.6795,
      "step": 1808
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.039634430889681,
      "learning_rate": 1.5148540994875242e-05,
      "loss": 0.9685,
      "step": 1809
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9574502263818773,
      "learning_rate": 1.5143198429570181e-05,
      "loss": 0.9008,
      "step": 1810
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1618801413772315,
      "learning_rate": 1.5137853867498403e-05,
      "loss": 0.9561,
      "step": 1811
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.082112241827022,
      "learning_rate": 1.5132507310734847e-05,
      "loss": 1.0203,
      "step": 1812
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9927766276555399,
      "learning_rate": 1.5127158761355241e-05,
      "loss": 0.9376,
      "step": 1813
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0180022227457428,
      "learning_rate": 1.512180822143607e-05,
      "loss": 0.9932,
      "step": 1814
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1209086197176081,
      "learning_rate": 1.5116455693054594e-05,
      "loss": 0.9941,
      "step": 1815
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2101457124147248,
      "learning_rate": 1.5111101178288858e-05,
      "loss": 0.9877,
      "step": 1816
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9557826502152701,
      "learning_rate": 1.510574467921766e-05,
      "loss": 1.0214,
      "step": 1817
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0645378753844585,
      "learning_rate": 1.5100386197920585e-05,
      "loss": 0.9239,
      "step": 1818
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9703841220123243,
      "learning_rate": 1.5095025736477977e-05,
      "loss": 0.9316,
      "step": 1819
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1541803566624145,
      "learning_rate": 1.5089663296970952e-05,
      "loss": 0.95,
      "step": 1820
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1581272719368867,
      "learning_rate": 1.5084298881481388e-05,
      "loss": 1.0285,
      "step": 1821
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0194802578646829,
      "learning_rate": 1.5078932492091942e-05,
      "loss": 0.9778,
      "step": 1822
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9998802933206844,
      "learning_rate": 1.5073564130886032e-05,
      "loss": 0.9166,
      "step": 1823
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9781841870307209,
      "learning_rate": 1.506819379994784e-05,
      "loss": 0.9216,
      "step": 1824
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0138020016179599,
      "learning_rate": 1.5062821501362308e-05,
      "loss": 0.9743,
      "step": 1825
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0699101698197886,
      "learning_rate": 1.5057447237215152e-05,
      "loss": 1.0326,
      "step": 1826
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1014485809673935,
      "learning_rate": 1.5052071009592846e-05,
      "loss": 0.9803,
      "step": 1827
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0334429385975334,
      "learning_rate": 1.5046692820582625e-05,
      "loss": 1.0086,
      "step": 1828
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.105349100258766,
      "learning_rate": 1.504131267227249e-05,
      "loss": 0.9217,
      "step": 1829
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0109436877815245,
      "learning_rate": 1.5035930566751198e-05,
      "loss": 0.9438,
      "step": 1830
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0061016226225135,
      "learning_rate": 1.5030546506108268e-05,
      "loss": 0.9399,
      "step": 1831
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0161035086763548,
      "learning_rate": 1.5025160492433976e-05,
      "loss": 0.9481,
      "step": 1832
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.124456748620959,
      "learning_rate": 1.501977252781936e-05,
      "loss": 0.9909,
      "step": 1833
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9479724207373277,
      "learning_rate": 1.5014382614356213e-05,
      "loss": 0.964,
      "step": 1834
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.075754326959277,
      "learning_rate": 1.5008990754137088e-05,
      "loss": 0.9695,
      "step": 1835
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9793208008918234,
      "learning_rate": 1.5003596949255284e-05,
      "loss": 0.8964,
      "step": 1836
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.046407997197139,
      "learning_rate": 1.4998201201804867e-05,
      "loss": 0.9647,
      "step": 1837
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1302297203029819,
      "learning_rate": 1.499280351388065e-05,
      "loss": 1.0222,
      "step": 1838
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.011136755429944,
      "learning_rate": 1.49874038875782e-05,
      "loss": 0.9947,
      "step": 1839
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.826692613492878,
      "learning_rate": 1.498200232499384e-05,
      "loss": 0.6273,
      "step": 1840
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0148411444798677,
      "learning_rate": 1.4976598828224643e-05,
      "loss": 0.9343,
      "step": 1841
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7042320939969292,
      "learning_rate": 1.497119339936843e-05,
      "loss": 0.362,
      "step": 1842
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9196992672383488,
      "learning_rate": 1.4965786040523779e-05,
      "loss": 0.9445,
      "step": 1843
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8245338102184926,
      "learning_rate": 1.496037675379001e-05,
      "loss": 0.6697,
      "step": 1844
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1414790209085004,
      "learning_rate": 1.4954965541267192e-05,
      "loss": 1.0122,
      "step": 1845
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9017256414571174,
      "learning_rate": 1.494955240505615e-05,
      "loss": 0.98,
      "step": 1846
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9830657873200433,
      "learning_rate": 1.494413734725844e-05,
      "loss": 0.9369,
      "step": 1847
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9496601978218093,
      "learning_rate": 1.4938720369976385e-05,
      "loss": 0.9618,
      "step": 1848
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0275530615496309,
      "learning_rate": 1.4933301475313036e-05,
      "loss": 0.9677,
      "step": 1849
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1947931395535203,
      "learning_rate": 1.4927880665372197e-05,
      "loss": 1.0008,
      "step": 1850
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9659234238751012,
      "learning_rate": 1.4922457942258411e-05,
      "loss": 0.9915,
      "step": 1851
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9959787543151098,
      "learning_rate": 1.4917033308076967e-05,
      "loss": 0.9249,
      "step": 1852
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.258828584429396,
      "learning_rate": 1.4911606764933892e-05,
      "loss": 0.9803,
      "step": 1853
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.91805688303064,
      "learning_rate": 1.490617831493596e-05,
      "loss": 0.6383,
      "step": 1854
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0709883916478253,
      "learning_rate": 1.4900747960190682e-05,
      "loss": 0.9679,
      "step": 1855
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9136331198751051,
      "learning_rate": 1.489531570280631e-05,
      "loss": 0.8936,
      "step": 1856
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0910289572547522,
      "learning_rate": 1.488988154489183e-05,
      "loss": 0.9932,
      "step": 1857
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.981798579562163,
      "learning_rate": 1.4884445488556972e-05,
      "loss": 0.9604,
      "step": 1858
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.912706110103883,
      "learning_rate": 1.4879007535912198e-05,
      "loss": 0.9459,
      "step": 1859
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0149422852027776,
      "learning_rate": 1.4873567689068708e-05,
      "loss": 0.8708,
      "step": 1860
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0129136707208675,
      "learning_rate": 1.4868125950138442e-05,
      "loss": 0.9216,
      "step": 1861
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.052984122233298,
      "learning_rate": 1.4862682321234064e-05,
      "loss": 0.9037,
      "step": 1862
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9047228614938874,
      "learning_rate": 1.4857236804468983e-05,
      "loss": 0.6529,
      "step": 1863
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.157148941672085,
      "learning_rate": 1.4851789401957338e-05,
      "loss": 1.0236,
      "step": 1864
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1748257192788656,
      "learning_rate": 1.4846340115813993e-05,
      "loss": 0.9715,
      "step": 1865
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9915914073098743,
      "learning_rate": 1.484088894815455e-05,
      "loss": 1.0341,
      "step": 1866
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.801074188574921,
      "learning_rate": 1.4835435901095341e-05,
      "loss": 0.6296,
      "step": 1867
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8801662708764562,
      "learning_rate": 1.4829980976753426e-05,
      "loss": 0.7203,
      "step": 1868
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8197718418116551,
      "learning_rate": 1.4824524177246597e-05,
      "loss": 0.6631,
      "step": 1869
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.897444204329071,
      "learning_rate": 1.4819065504693365e-05,
      "loss": 0.9495,
      "step": 1870
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0343350187095361,
      "learning_rate": 1.4813604961212984e-05,
      "loss": 0.9446,
      "step": 1871
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.015083841659557,
      "learning_rate": 1.4808142548925417e-05,
      "loss": 1.0045,
      "step": 1872
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.012168840312588,
      "learning_rate": 1.4802678269951365e-05,
      "loss": 0.8904,
      "step": 1873
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8944866340298975,
      "learning_rate": 1.4797212126412243e-05,
      "loss": 0.8799,
      "step": 1874
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9340858610197325,
      "learning_rate": 1.4791744120430202e-05,
      "loss": 0.9512,
      "step": 1875
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0642158501953385,
      "learning_rate": 1.4786274254128112e-05,
      "loss": 0.9591,
      "step": 1876
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.149583248644542,
      "learning_rate": 1.4780802529629559e-05,
      "loss": 0.9488,
      "step": 1877
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9772522812440924,
      "learning_rate": 1.4775328949058856e-05,
      "loss": 0.9426,
      "step": 1878
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8859680542594723,
      "learning_rate": 1.4769853514541037e-05,
      "loss": 0.6876,
      "step": 1879
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0654676755781416,
      "learning_rate": 1.4764376228201848e-05,
      "loss": 0.9542,
      "step": 1880
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.063051519103078,
      "learning_rate": 1.475889709216777e-05,
      "loss": 0.9602,
      "step": 1881
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0155630656761738,
      "learning_rate": 1.4753416108565985e-05,
      "loss": 0.9396,
      "step": 1882
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9629631628130588,
      "learning_rate": 1.47479332795244e-05,
      "loss": 0.9314,
      "step": 1883
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0633637812478798,
      "learning_rate": 1.4742448607171644e-05,
      "loss": 0.6043,
      "step": 1884
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1021636154906582,
      "learning_rate": 1.473696209363705e-05,
      "loss": 0.9841,
      "step": 1885
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.098988728939536,
      "learning_rate": 1.4731473741050673e-05,
      "loss": 0.9756,
      "step": 1886
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1511051081516723,
      "learning_rate": 1.4725983551543279e-05,
      "loss": 0.9692,
      "step": 1887
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9935973151010361,
      "learning_rate": 1.472049152724635e-05,
      "loss": 1.0019,
      "step": 1888
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.059307237127531,
      "learning_rate": 1.471499767029208e-05,
      "loss": 0.9707,
      "step": 1889
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9908376947021609,
      "learning_rate": 1.470950198281337e-05,
      "loss": 0.9376,
      "step": 1890
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9726649574455914,
      "learning_rate": 1.470400446694384e-05,
      "loss": 0.9079,
      "step": 1891
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0145763724089603,
      "learning_rate": 1.4698505124817811e-05,
      "loss": 0.9636,
      "step": 1892
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8872086735417855,
      "learning_rate": 1.4693003958570318e-05,
      "loss": 0.6206,
      "step": 1893
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9793911420155104,
      "learning_rate": 1.4687500970337103e-05,
      "loss": 0.9401,
      "step": 1894
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9374903554221604,
      "learning_rate": 1.4681996162254618e-05,
      "loss": 0.9461,
      "step": 1895
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0312076520623616,
      "learning_rate": 1.4676489536460015e-05,
      "loss": 0.949,
      "step": 1896
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0862503227251514,
      "learning_rate": 1.467098109509116e-05,
      "loss": 0.9621,
      "step": 1897
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0147094249295103,
      "learning_rate": 1.4665470840286614e-05,
      "loss": 0.9853,
      "step": 1898
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.016246820677226,
      "learning_rate": 1.4659958774185654e-05,
      "loss": 0.9332,
      "step": 1899
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0117788800943537,
      "learning_rate": 1.4654444898928249e-05,
      "loss": 0.9597,
      "step": 1900
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9728962358508069,
      "learning_rate": 1.4648929216655077e-05,
      "loss": 0.9169,
      "step": 1901
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0622181036343046,
      "learning_rate": 1.4643411729507517e-05,
      "loss": 0.9238,
      "step": 1902
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9936321108864654,
      "learning_rate": 1.4637892439627644e-05,
      "loss": 0.9664,
      "step": 1903
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.018157816289098,
      "learning_rate": 1.4632371349158241e-05,
      "loss": 0.9399,
      "step": 1904
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0199373339033362,
      "learning_rate": 1.4626848460242782e-05,
      "loss": 0.9217,
      "step": 1905
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9700581988572852,
      "learning_rate": 1.4621323775025444e-05,
      "loss": 0.9758,
      "step": 1906
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7827230476561043,
      "learning_rate": 1.4615797295651099e-05,
      "loss": 0.6884,
      "step": 1907
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0052694125212556,
      "learning_rate": 1.4610269024265317e-05,
      "loss": 0.9398,
      "step": 1908
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0725087882874793,
      "learning_rate": 1.4604738963014365e-05,
      "loss": 0.9425,
      "step": 1909
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8033828016253509,
      "learning_rate": 1.4599207114045202e-05,
      "loss": 0.6054,
      "step": 1910
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1237099948533598,
      "learning_rate": 1.4593673479505482e-05,
      "loss": 0.9998,
      "step": 1911
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0296495277312614,
      "learning_rate": 1.4588138061543551e-05,
      "loss": 0.9721,
      "step": 1912
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2187705739360204,
      "learning_rate": 1.458260086230845e-05,
      "loss": 0.9652,
      "step": 1913
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2773143381413157,
      "learning_rate": 1.4577061883949912e-05,
      "loss": 0.9656,
      "step": 1914
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0285778812153967,
      "learning_rate": 1.4571521128618358e-05,
      "loss": 0.9283,
      "step": 1915
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1268923641225783,
      "learning_rate": 1.4565978598464895e-05,
      "loss": 0.9474,
      "step": 1916
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0433180430397355,
      "learning_rate": 1.4560434295641338e-05,
      "loss": 0.996,
      "step": 1917
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7107150578372431,
      "learning_rate": 1.455488822230016e-05,
      "loss": 0.3785,
      "step": 1918
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0850208866354072,
      "learning_rate": 1.4549340380594545e-05,
      "loss": 0.9534,
      "step": 1919
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9803674475040338,
      "learning_rate": 1.454379077267836e-05,
      "loss": 0.9503,
      "step": 1920
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8433754363304108,
      "learning_rate": 1.4538239400706147e-05,
      "loss": 0.6228,
      "step": 1921
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1170448004929483,
      "learning_rate": 1.4532686266833143e-05,
      "loss": 0.9362,
      "step": 1922
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1179901487540427,
      "learning_rate": 1.4527131373215265e-05,
      "loss": 0.8917,
      "step": 1923
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0403276235697805,
      "learning_rate": 1.4521574722009115e-05,
      "loss": 0.9162,
      "step": 1924
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0330689215530757,
      "learning_rate": 1.4516016315371974e-05,
      "loss": 0.9634,
      "step": 1925
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0944480292817507,
      "learning_rate": 1.4510456155461807e-05,
      "loss": 0.9613,
      "step": 1926
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9558257076186464,
      "learning_rate": 1.4504894244437264e-05,
      "loss": 0.9699,
      "step": 1927
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.010125238909889,
      "learning_rate": 1.4499330584457667e-05,
      "loss": 0.9861,
      "step": 1928
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0315826432221056,
      "learning_rate": 1.4493765177683017e-05,
      "loss": 0.9743,
      "step": 1929
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1267581154060322,
      "learning_rate": 1.4488198026274007e-05,
      "loss": 0.9364,
      "step": 1930
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0613966243609414,
      "learning_rate": 1.4482629132391985e-05,
      "loss": 0.9402,
      "step": 1931
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9914221207439596,
      "learning_rate": 1.4477058498198993e-05,
      "loss": 0.9612,
      "step": 1932
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0150380343497498,
      "learning_rate": 1.4471486125857743e-05,
      "loss": 0.9253,
      "step": 1933
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.012100708691394,
      "learning_rate": 1.446591201753162e-05,
      "loss": 0.9699,
      "step": 1934
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0709698146174493,
      "learning_rate": 1.4460336175384688e-05,
      "loss": 1.0075,
      "step": 1935
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0806900018155348,
      "learning_rate": 1.4454758601581675e-05,
      "loss": 0.9838,
      "step": 1936
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0215320336710008,
      "learning_rate": 1.4449179298287999e-05,
      "loss": 0.9629,
      "step": 1937
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9895160186035197,
      "learning_rate": 1.4443598267669723e-05,
      "loss": 0.9165,
      "step": 1938
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0291459541362726,
      "learning_rate": 1.4438015511893602e-05,
      "loss": 0.9704,
      "step": 1939
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.020655546720019,
      "learning_rate": 1.4432431033127056e-05,
      "loss": 0.9757,
      "step": 1940
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9900302736579326,
      "learning_rate": 1.442684483353817e-05,
      "loss": 1.0093,
      "step": 1941
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8938000123371197,
      "learning_rate": 1.4421256915295697e-05,
      "loss": 0.9443,
      "step": 1942
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0083929717103997,
      "learning_rate": 1.4415667280569064e-05,
      "loss": 0.9822,
      "step": 1943
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0419235533121607,
      "learning_rate": 1.4410075931528356e-05,
      "loss": 0.9597,
      "step": 1944
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0936114572586764,
      "learning_rate": 1.4404482870344322e-05,
      "loss": 0.9049,
      "step": 1945
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7857003281068364,
      "learning_rate": 1.4398888099188396e-05,
      "loss": 0.6178,
      "step": 1946
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9557042794815845,
      "learning_rate": 1.4393291620232646e-05,
      "loss": 0.9309,
      "step": 1947
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9120561212262487,
      "learning_rate": 1.4387693435649826e-05,
      "loss": 0.993,
      "step": 1948
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0094733721466789,
      "learning_rate": 1.4382093547613338e-05,
      "loss": 0.9803,
      "step": 1949
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9792280393633183,
      "learning_rate": 1.4376491958297263e-05,
      "loss": 0.9664,
      "step": 1950
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8959991160089729,
      "learning_rate": 1.4370888669876317e-05,
      "loss": 0.641,
      "step": 1951
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0474545147869792,
      "learning_rate": 1.4365283684525895e-05,
      "loss": 0.9612,
      "step": 1952
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0480617010739939,
      "learning_rate": 1.4359677004422045e-05,
      "loss": 0.9975,
      "step": 1953
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.894591598057618,
      "learning_rate": 1.4354068631741476e-05,
      "loss": 0.9342,
      "step": 1954
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9856174415570392,
      "learning_rate": 1.4348458568661548e-05,
      "loss": 0.9148,
      "step": 1955
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9267322479727704,
      "learning_rate": 1.434284681736028e-05,
      "loss": 0.939,
      "step": 1956
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8853535678909399,
      "learning_rate": 1.4337233380016354e-05,
      "loss": 0.9435,
      "step": 1957
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.866182972188878,
      "learning_rate": 1.433161825880909e-05,
      "loss": 0.6331,
      "step": 1958
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.000759074030009,
      "learning_rate": 1.432600145591848e-05,
      "loss": 0.9371,
      "step": 1959
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8987475122085256,
      "learning_rate": 1.4320382973525151e-05,
      "loss": 0.9462,
      "step": 1960
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.149185946814549,
      "learning_rate": 1.43147628138104e-05,
      "loss": 0.977,
      "step": 1961
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0680062497804823,
      "learning_rate": 1.4309140978956161e-05,
      "loss": 0.9501,
      "step": 1962
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9048494966134598,
      "learning_rate": 1.430351747114503e-05,
      "loss": 0.9873,
      "step": 1963
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9893719141578641,
      "learning_rate": 1.429789229256024e-05,
      "loss": 0.9328,
      "step": 1964
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0533662290519177,
      "learning_rate": 1.429226544538568e-05,
      "loss": 0.9595,
      "step": 1965
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0103872710358641,
      "learning_rate": 1.4286636931805887e-05,
      "loss": 0.9423,
      "step": 1966
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.108516347176009,
      "learning_rate": 1.4281006754006045e-05,
      "loss": 0.9404,
      "step": 1967
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9598886431357941,
      "learning_rate": 1.427537491417198e-05,
      "loss": 0.9195,
      "step": 1968
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0035196679732867,
      "learning_rate": 1.426974141449017e-05,
      "loss": 0.9371,
      "step": 1969
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0716436832825407,
      "learning_rate": 1.4264106257147732e-05,
      "loss": 0.8935,
      "step": 1970
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0059700543888521,
      "learning_rate": 1.4258469444332423e-05,
      "loss": 0.9748,
      "step": 1971
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9849991470639715,
      "learning_rate": 1.4252830978232658e-05,
      "loss": 0.9796,
      "step": 1972
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9519771169931014,
      "learning_rate": 1.4247190861037474e-05,
      "loss": 0.8952,
      "step": 1973
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1082888332657015,
      "learning_rate": 1.4241549094936567e-05,
      "loss": 0.9497,
      "step": 1974
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0174478815456425,
      "learning_rate": 1.4235905682120255e-05,
      "loss": 0.913,
      "step": 1975
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0222887891204635,
      "learning_rate": 1.4230260624779512e-05,
      "loss": 0.9665,
      "step": 1976
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0986287661583878,
      "learning_rate": 1.4224613925105947e-05,
      "loss": 0.9605,
      "step": 1977
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.021993336602727,
      "learning_rate": 1.4218965585291792e-05,
      "loss": 0.9827,
      "step": 1978
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1358006733721169,
      "learning_rate": 1.4213315607529939e-05,
      "loss": 1.0249,
      "step": 1979
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0608424991805148,
      "learning_rate": 1.4207663994013896e-05,
      "loss": 0.9716,
      "step": 1980
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0776464831726986,
      "learning_rate": 1.4202010746937815e-05,
      "loss": 0.764,
      "step": 1981
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0787128729106152,
      "learning_rate": 1.4196355868496485e-05,
      "loss": 0.9556,
      "step": 1982
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9627791589506264,
      "learning_rate": 1.4190699360885323e-05,
      "loss": 0.9349,
      "step": 1983
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7785307953797261,
      "learning_rate": 1.4185041226300376e-05,
      "loss": 0.6189,
      "step": 1984
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1494831187620036,
      "learning_rate": 1.4179381466938332e-05,
      "loss": 0.9373,
      "step": 1985
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9927466936438021,
      "learning_rate": 1.4173720084996501e-05,
      "loss": 0.9437,
      "step": 1986
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9647258468599712,
      "learning_rate": 1.4168057082672828e-05,
      "loss": 0.9399,
      "step": 1987
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8625221923162019,
      "learning_rate": 1.4162392462165884e-05,
      "loss": 0.6754,
      "step": 1988
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.02781976581513,
      "learning_rate": 1.4156726225674874e-05,
      "loss": 0.9677,
      "step": 1989
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0328167198875124,
      "learning_rate": 1.415105837539962e-05,
      "loss": 0.9723,
      "step": 1990
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0115128045417245,
      "learning_rate": 1.414538891354058e-05,
      "loss": 0.9634,
      "step": 1991
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0650400293605717,
      "learning_rate": 1.4139717842298835e-05,
      "loss": 1.001,
      "step": 1992
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8751747967105946,
      "learning_rate": 1.4134045163876086e-05,
      "loss": 0.9236,
      "step": 1993
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9537534076090312,
      "learning_rate": 1.4128370880474667e-05,
      "loss": 0.9687,
      "step": 1994
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9879876096809339,
      "learning_rate": 1.412269499429753e-05,
      "loss": 0.951,
      "step": 1995
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1287028142168425,
      "learning_rate": 1.4117017507548244e-05,
      "loss": 0.9357,
      "step": 1996
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0884453049219922,
      "learning_rate": 1.4111338422431013e-05,
      "loss": 0.9791,
      "step": 1997
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8552363024910556,
      "learning_rate": 1.4105657741150648e-05,
      "loss": 0.6232,
      "step": 1998
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9332518130950831,
      "learning_rate": 1.4099975465912584e-05,
      "loss": 0.6347,
      "step": 1999
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9819414365162156,
      "learning_rate": 1.4094291598922877e-05,
      "loss": 0.9239,
      "step": 2000
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.983829947814466,
      "learning_rate": 1.40886061423882e-05,
      "loss": 0.9268,
      "step": 2001
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9017418974210837,
      "learning_rate": 1.4082919098515846e-05,
      "loss": 0.9073,
      "step": 2002
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1176319497022726,
      "learning_rate": 1.407723046951372e-05,
      "loss": 0.9081,
      "step": 2003
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0726727965906169,
      "learning_rate": 1.4071540257590341e-05,
      "loss": 0.9527,
      "step": 2004
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0123808523373212,
      "learning_rate": 1.4065848464954848e-05,
      "loss": 0.9706,
      "step": 2005
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8424208881368768,
      "learning_rate": 1.4060155093816988e-05,
      "loss": 0.6512,
      "step": 2006
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8776306672374167,
      "learning_rate": 1.4054460146387124e-05,
      "loss": 0.6599,
      "step": 2007
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.117994693980558,
      "learning_rate": 1.4048763624876233e-05,
      "loss": 0.9657,
      "step": 2008
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9285332409969402,
      "learning_rate": 1.4043065531495904e-05,
      "loss": 0.9178,
      "step": 2009
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.104305237809932,
      "learning_rate": 1.4037365868458325e-05,
      "loss": 0.9438,
      "step": 2010
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0251845971390707,
      "learning_rate": 1.4031664637976305e-05,
      "loss": 0.9645,
      "step": 2011
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0430608496202318,
      "learning_rate": 1.402596184226326e-05,
      "loss": 0.9251,
      "step": 2012
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0800815939200847,
      "learning_rate": 1.4020257483533208e-05,
      "loss": 0.9228,
      "step": 2013
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9370502637436386,
      "learning_rate": 1.401455156400078e-05,
      "loss": 0.9178,
      "step": 2014
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9794222125942598,
      "learning_rate": 1.400884408588121e-05,
      "loss": 0.9553,
      "step": 2015
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8432493979277902,
      "learning_rate": 1.400313505139034e-05,
      "loss": 0.6448,
      "step": 2016
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8948005516605132,
      "learning_rate": 1.3997424462744607e-05,
      "loss": 0.6239,
      "step": 2017
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1139344777255415,
      "learning_rate": 1.3991712322161065e-05,
      "loss": 1.0062,
      "step": 2018
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9073185088803565,
      "learning_rate": 1.3985998631857359e-05,
      "loss": 0.9619,
      "step": 2019
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1280885495775423,
      "learning_rate": 1.398028339405174e-05,
      "loss": 0.992,
      "step": 2020
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9003878487897925,
      "learning_rate": 1.3974566610963068e-05,
      "loss": 0.8878,
      "step": 2021
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0586794342575965,
      "learning_rate": 1.3968848284810785e-05,
      "loss": 0.9097,
      "step": 2022
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8803411347490628,
      "learning_rate": 1.3963128417814951e-05,
      "loss": 0.9447,
      "step": 2023
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0666161390006017,
      "learning_rate": 1.3957407012196204e-05,
      "loss": 0.976,
      "step": 2024
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.192743243401694,
      "learning_rate": 1.3951684070175802e-05,
      "loss": 1.018,
      "step": 2025
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.822748927885213,
      "learning_rate": 1.3945959593975582e-05,
      "loss": 0.6505,
      "step": 2026
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9547028169831402,
      "learning_rate": 1.3940233585817984e-05,
      "loss": 0.9017,
      "step": 2027
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0384376595942493,
      "learning_rate": 1.3934506047926042e-05,
      "loss": 0.9229,
      "step": 2028
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9942889540865519,
      "learning_rate": 1.3928776982523384e-05,
      "loss": 0.9502,
      "step": 2029
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3278842273016482,
      "learning_rate": 1.3923046391834229e-05,
      "loss": 1.012,
      "step": 2030
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9591188999244696,
      "learning_rate": 1.3917314278083391e-05,
      "loss": 0.9466,
      "step": 2031
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8258507987613136,
      "learning_rate": 1.3911580643496272e-05,
      "loss": 0.6008,
      "step": 2032
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9390977930519564,
      "learning_rate": 1.3905845490298867e-05,
      "loss": 0.9004,
      "step": 2033
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.84999554849105,
      "learning_rate": 1.390010882071776e-05,
      "loss": 0.6539,
      "step": 2034
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.988945580103983,
      "learning_rate": 1.3894370636980128e-05,
      "loss": 0.9209,
      "step": 2035
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2004152599656408,
      "learning_rate": 1.3888630941313728e-05,
      "loss": 0.9666,
      "step": 2036
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.986729484398346,
      "learning_rate": 1.3882889735946901e-05,
      "loss": 0.942,
      "step": 2037
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9922242514494508,
      "learning_rate": 1.3877147023108592e-05,
      "loss": 0.9248,
      "step": 2038
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.133377352013382,
      "learning_rate": 1.3871402805028314e-05,
      "loss": 0.986,
      "step": 2039
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0615906101931727,
      "learning_rate": 1.3865657083936167e-05,
      "loss": 1.0063,
      "step": 2040
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9034474332390209,
      "learning_rate": 1.3859909862062844e-05,
      "loss": 0.6695,
      "step": 2041
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0394357935271599,
      "learning_rate": 1.385416114163961e-05,
      "loss": 0.9592,
      "step": 2042
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9237348008349145,
      "learning_rate": 1.3848410924898321e-05,
      "loss": 0.6498,
      "step": 2043
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9552550583461423,
      "learning_rate": 1.3842659214071406e-05,
      "loss": 0.9735,
      "step": 2044
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9509469434342989,
      "learning_rate": 1.3836906011391878e-05,
      "loss": 0.9995,
      "step": 2045
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9157862114186139,
      "learning_rate": 1.3831151319093323e-05,
      "loss": 0.9343,
      "step": 2046
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.06069709386187,
      "learning_rate": 1.382539513940992e-05,
      "loss": 0.9802,
      "step": 2047
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0408908487854944,
      "learning_rate": 1.3819637474576411e-05,
      "loss": 0.9548,
      "step": 2048
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0200581858274023,
      "learning_rate": 1.381387832682812e-05,
      "loss": 0.9436,
      "step": 2049
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9081618471404518,
      "learning_rate": 1.380811769840095e-05,
      "loss": 0.9782,
      "step": 2050
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.028706048282443,
      "learning_rate": 1.3802355591531366e-05,
      "loss": 0.9593,
      "step": 2051
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9446670788442857,
      "learning_rate": 1.3796592008456427e-05,
      "loss": 0.9641,
      "step": 2052
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9717987388973163,
      "learning_rate": 1.3790826951413747e-05,
      "loss": 0.9414,
      "step": 2053
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8168475403469463,
      "learning_rate": 1.3785060422641526e-05,
      "loss": 0.6612,
      "step": 2054
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9974605389109276,
      "learning_rate": 1.3779292424378521e-05,
      "loss": 0.9538,
      "step": 2055
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9113755505865522,
      "learning_rate": 1.3773522958864076e-05,
      "loss": 0.9124,
      "step": 2056
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0388127778437888,
      "learning_rate": 1.3767752028338091e-05,
      "loss": 0.9449,
      "step": 2057
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1331710531626706,
      "learning_rate": 1.376197963504104e-05,
      "loss": 0.9905,
      "step": 2058
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9810223377281708,
      "learning_rate": 1.3756205781213965e-05,
      "loss": 0.9972,
      "step": 2059
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0427643983377362,
      "learning_rate": 1.375043046909848e-05,
      "loss": 0.9247,
      "step": 2060
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0803738237998362,
      "learning_rate": 1.3744653700936752e-05,
      "loss": 0.9936,
      "step": 2061
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0455229832852488,
      "learning_rate": 1.3738875478971526e-05,
      "loss": 0.9866,
      "step": 2062
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.822507126668907,
      "learning_rate": 1.3733095805446107e-05,
      "loss": 0.6147,
      "step": 2063
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0673816378570404,
      "learning_rate": 1.372731468260436e-05,
      "loss": 0.9409,
      "step": 2064
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0407994758945533,
      "learning_rate": 1.372153211269072e-05,
      "loss": 0.9472,
      "step": 2065
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.019752754364769,
      "learning_rate": 1.3715748097950176e-05,
      "loss": 0.9237,
      "step": 2066
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.041051120582331,
      "learning_rate": 1.3709962640628284e-05,
      "loss": 0.9806,
      "step": 2067
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.07836184640602,
      "learning_rate": 1.3704175742971158e-05,
      "loss": 0.9385,
      "step": 2068
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9734579580221504,
      "learning_rate": 1.369838740722547e-05,
      "loss": 0.9841,
      "step": 2069
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9871000463171719,
      "learning_rate": 1.3692597635638452e-05,
      "loss": 0.925,
      "step": 2070
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2365118098567127,
      "learning_rate": 1.368680643045789e-05,
      "loss": 0.964,
      "step": 2071
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9610837022214752,
      "learning_rate": 1.3681013793932132e-05,
      "loss": 0.9613,
      "step": 2072
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0372368990986072,
      "learning_rate": 1.3675219728310076e-05,
      "loss": 0.9589,
      "step": 2073
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1379000383371405,
      "learning_rate": 1.3669424235841185e-05,
      "loss": 0.9905,
      "step": 2074
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0607306927472742,
      "learning_rate": 1.3663627318775459e-05,
      "loss": 0.9371,
      "step": 2075
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.843511865431606,
      "learning_rate": 1.3657828979363468e-05,
      "loss": 0.6813,
      "step": 2076
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.145563848641989,
      "learning_rate": 1.3652029219856324e-05,
      "loss": 1.0024,
      "step": 2077
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9241512245952934,
      "learning_rate": 1.3646228042505694e-05,
      "loss": 0.9062,
      "step": 2078
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0832861646030476,
      "learning_rate": 1.3640425449563793e-05,
      "loss": 0.9822,
      "step": 2079
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9933057457041327,
      "learning_rate": 1.3634621443283389e-05,
      "loss": 0.9319,
      "step": 2080
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8462228334449436,
      "learning_rate": 1.36288160259178e-05,
      "loss": 0.6528,
      "step": 2081
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.017236140741003,
      "learning_rate": 1.3623009199720882e-05,
      "loss": 0.9206,
      "step": 2082
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9556086052848908,
      "learning_rate": 1.3617200966947053e-05,
      "loss": 0.9409,
      "step": 2083
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9975076193022928,
      "learning_rate": 1.3611391329851262e-05,
      "loss": 0.9758,
      "step": 2084
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1881009858617761,
      "learning_rate": 1.3605580290689013e-05,
      "loss": 0.9931,
      "step": 2085
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9796145866369724,
      "learning_rate": 1.3599767851716353e-05,
      "loss": 0.9358,
      "step": 2086
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0368462301012227,
      "learning_rate": 1.3593954015189867e-05,
      "loss": 1.0007,
      "step": 2087
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9979223443797169,
      "learning_rate": 1.3588138783366692e-05,
      "loss": 1.0088,
      "step": 2088
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9836360426292967,
      "learning_rate": 1.3582322158504495e-05,
      "loss": 0.897,
      "step": 2089
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0922624127778893,
      "learning_rate": 1.3576504142861496e-05,
      "loss": 1.0014,
      "step": 2090
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0180087869105303,
      "learning_rate": 1.3570684738696444e-05,
      "loss": 0.9803,
      "step": 2091
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0823582665473719,
      "learning_rate": 1.3564863948268631e-05,
      "loss": 0.9379,
      "step": 2092
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.88379249782843,
      "learning_rate": 1.3559041773837898e-05,
      "loss": 0.6454,
      "step": 2093
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7985794474846934,
      "learning_rate": 1.3553218217664603e-05,
      "loss": 0.6796,
      "step": 2094
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0446621699108682,
      "learning_rate": 1.3547393282009656e-05,
      "loss": 1.0013,
      "step": 2095
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9883400415189996,
      "learning_rate": 1.3541566969134496e-05,
      "loss": 0.9578,
      "step": 2096
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9815242932457442,
      "learning_rate": 1.3535739281301102e-05,
      "loss": 0.9545,
      "step": 2097
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0944809546663967,
      "learning_rate": 1.3529910220771975e-05,
      "loss": 1.0049,
      "step": 2098
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9815803878505215,
      "learning_rate": 1.3524079789810163e-05,
      "loss": 0.9549,
      "step": 2099
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.012875526656265,
      "learning_rate": 1.3518247990679241e-05,
      "loss": 0.9996,
      "step": 2100
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.02524830534412,
      "learning_rate": 1.3512414825643312e-05,
      "loss": 0.9984,
      "step": 2101
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.197516341196219,
      "learning_rate": 1.3506580296967011e-05,
      "loss": 1.0123,
      "step": 2102
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9152311579665654,
      "learning_rate": 1.3500744406915505e-05,
      "loss": 0.9144,
      "step": 2103
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8702044273348516,
      "learning_rate": 1.3494907157754485e-05,
      "loss": 0.6156,
      "step": 2104
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.127478543591371,
      "learning_rate": 1.348906855175017e-05,
      "loss": 0.8948,
      "step": 2105
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9314641416731849,
      "learning_rate": 1.3483228591169315e-05,
      "loss": 0.913,
      "step": 2106
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9351438222045788,
      "learning_rate": 1.347738727827919e-05,
      "loss": 0.9237,
      "step": 2107
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7850161213578077,
      "learning_rate": 1.3471544615347591e-05,
      "loss": 0.6592,
      "step": 2108
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.991776461717964,
      "learning_rate": 1.3465700604642847e-05,
      "loss": 0.9407,
      "step": 2109
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0574819873786017,
      "learning_rate": 1.34598552484338e-05,
      "loss": 0.9638,
      "step": 2110
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9646972286065839,
      "learning_rate": 1.3454008548989816e-05,
      "loss": 0.949,
      "step": 2111
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.003000300790026,
      "learning_rate": 1.3448160508580789e-05,
      "loss": 0.9675,
      "step": 2112
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0016560860133032,
      "learning_rate": 1.3442311129477133e-05,
      "loss": 0.945,
      "step": 2113
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0135993755880055,
      "learning_rate": 1.343646041394977e-05,
      "loss": 0.9515,
      "step": 2114
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0661098438150565,
      "learning_rate": 1.3430608364270156e-05,
      "loss": 0.9723,
      "step": 2115
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9339617733033485,
      "learning_rate": 1.3424754982710256e-05,
      "loss": 0.6625,
      "step": 2116
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.987225634072919,
      "learning_rate": 1.3418900271542552e-05,
      "loss": 0.9866,
      "step": 2117
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1630687031722189,
      "learning_rate": 1.3413044233040045e-05,
      "loss": 0.9531,
      "step": 2118
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0168468586387343,
      "learning_rate": 1.3407186869476253e-05,
      "loss": 0.9756,
      "step": 2119
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.081124794719034,
      "learning_rate": 1.3401328183125208e-05,
      "loss": 1.0044,
      "step": 2120
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1703331014439498,
      "learning_rate": 1.339546817626145e-05,
      "loss": 0.9542,
      "step": 2121
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9488306422119703,
      "learning_rate": 1.3389606851160037e-05,
      "loss": 0.9372,
      "step": 2122
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9581195091029099,
      "learning_rate": 1.3383744210096537e-05,
      "loss": 0.9084,
      "step": 2123
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9759379017871361,
      "learning_rate": 1.3377880255347026e-05,
      "loss": 0.9091,
      "step": 2124
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9344873195926506,
      "learning_rate": 1.3372014989188098e-05,
      "loss": 0.9577,
      "step": 2125
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0545356633466965,
      "learning_rate": 1.3366148413896851e-05,
      "loss": 0.9702,
      "step": 2126
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9524102286370763,
      "learning_rate": 1.3360280531750886e-05,
      "loss": 0.9136,
      "step": 2127
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9895709122225407,
      "learning_rate": 1.3354411345028324e-05,
      "loss": 0.9594,
      "step": 2128
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.016628209721258,
      "learning_rate": 1.3348540856007782e-05,
      "loss": 1.0019,
      "step": 2129
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9987739404559424,
      "learning_rate": 1.3342669066968385e-05,
      "loss": 0.9832,
      "step": 2130
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9453874993719854,
      "learning_rate": 1.3336795980189763e-05,
      "loss": 0.898,
      "step": 2131
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0085029565616992,
      "learning_rate": 1.3330921597952056e-05,
      "loss": 0.9464,
      "step": 2132
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1784813018421432,
      "learning_rate": 1.3325045922535896e-05,
      "loss": 0.9622,
      "step": 2133
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1203645958419497,
      "learning_rate": 1.3319168956222423e-05,
      "loss": 0.9419,
      "step": 2134
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0153243066787951,
      "learning_rate": 1.331329070129328e-05,
      "loss": 0.9371,
      "step": 2135
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9713068969425089,
      "learning_rate": 1.3307411160030608e-05,
      "loss": 0.9531,
      "step": 2136
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0771504652448645,
      "learning_rate": 1.3301530334717046e-05,
      "loss": 0.9606,
      "step": 2137
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0776229867462133,
      "learning_rate": 1.3295648227635729e-05,
      "loss": 0.9608,
      "step": 2138
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.100128462081169,
      "learning_rate": 1.32897648410703e-05,
      "loss": 0.9825,
      "step": 2139
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0819321242263171,
      "learning_rate": 1.328388017730489e-05,
      "loss": 0.9434,
      "step": 2140
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9577843933431014,
      "learning_rate": 1.327799423862413e-05,
      "loss": 0.9804,
      "step": 2141
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.077561906301385,
      "learning_rate": 1.3272107027313142e-05,
      "loss": 0.9504,
      "step": 2142
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9803052477132098,
      "learning_rate": 1.3266218545657541e-05,
      "loss": 0.9067,
      "step": 2143
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1510225314837592,
      "learning_rate": 1.326032879594344e-05,
      "loss": 0.98,
      "step": 2144
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8495000208437248,
      "learning_rate": 1.3254437780457448e-05,
      "loss": 0.6463,
      "step": 2145
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0849781497470723,
      "learning_rate": 1.3248545501486654e-05,
      "loss": 0.9385,
      "step": 2146
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9282914979711657,
      "learning_rate": 1.3242651961318646e-05,
      "loss": 0.9482,
      "step": 2147
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.145787256515085,
      "learning_rate": 1.32367571622415e-05,
      "loss": 0.9323,
      "step": 2148
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0427675650315693,
      "learning_rate": 1.3230861106543777e-05,
      "loss": 0.9391,
      "step": 2149
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0157447785575253,
      "learning_rate": 1.3224963796514532e-05,
      "loss": 0.9506,
      "step": 2150
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0099869382720819,
      "learning_rate": 1.32190652344433e-05,
      "loss": 0.9341,
      "step": 2151
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9725732461526724,
      "learning_rate": 1.3213165422620111e-05,
      "loss": 0.9204,
      "step": 2152
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0672808456229135,
      "learning_rate": 1.3207264363335472e-05,
      "loss": 0.9614,
      "step": 2153
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.826308005725167,
      "learning_rate": 1.3201362058880375e-05,
      "loss": 0.6477,
      "step": 2154
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.998941199398773,
      "learning_rate": 1.3195458511546307e-05,
      "loss": 0.9101,
      "step": 2155
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9946912107865085,
      "learning_rate": 1.3189553723625217e-05,
      "loss": 0.9777,
      "step": 2156
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0046099371294444,
      "learning_rate": 1.318364769740955e-05,
      "loss": 0.9643,
      "step": 2157
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.152024895896298,
      "learning_rate": 1.3177740435192235e-05,
      "loss": 0.9141,
      "step": 2158
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0842823920535687,
      "learning_rate": 1.3171831939266668e-05,
      "loss": 0.9211,
      "step": 2159
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1446913177604652,
      "learning_rate": 1.3165922211926734e-05,
      "loss": 1.031,
      "step": 2160
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.030229234425153,
      "learning_rate": 1.3160011255466791e-05,
      "loss": 0.9305,
      "step": 2161
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0008949842007464,
      "learning_rate": 1.3154099072181677e-05,
      "loss": 0.9637,
      "step": 2162
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8145945279608241,
      "learning_rate": 1.3148185664366704e-05,
      "loss": 0.6079,
      "step": 2163
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1205418784771446,
      "learning_rate": 1.314227103431766e-05,
      "loss": 0.9326,
      "step": 2164
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0896756695657157,
      "learning_rate": 1.3136355184330809e-05,
      "loss": 0.9574,
      "step": 2165
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9752768240665992,
      "learning_rate": 1.3130438116702888e-05,
      "loss": 0.949,
      "step": 2166
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9402725520721517,
      "learning_rate": 1.3124519833731106e-05,
      "loss": 0.8961,
      "step": 2167
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1280954777881413,
      "learning_rate": 1.3118600337713146e-05,
      "loss": 0.9319,
      "step": 2168
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9086766259714759,
      "learning_rate": 1.3112679630947156e-05,
      "loss": 0.9275,
      "step": 2169
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.056994602094244,
      "learning_rate": 1.310675771573176e-05,
      "loss": 0.9818,
      "step": 2170
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1042078750121256,
      "learning_rate": 1.310083459436605e-05,
      "loss": 1.0013,
      "step": 2171
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9321723658590099,
      "learning_rate": 1.3094910269149587e-05,
      "loss": 0.9673,
      "step": 2172
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.845854409531281,
      "learning_rate": 1.3088984742382395e-05,
      "loss": 0.6526,
      "step": 2173
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0118075250372616,
      "learning_rate": 1.3083058016364972e-05,
      "loss": 0.9109,
      "step": 2174
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1450855252376275,
      "learning_rate": 1.3077130093398274e-05,
      "loss": 0.9691,
      "step": 2175
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9902339891568397,
      "learning_rate": 1.3071200975783725e-05,
      "loss": 0.8772,
      "step": 2176
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0517726473660847,
      "learning_rate": 1.3065270665823206e-05,
      "loss": 0.934,
      "step": 2177
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7210304852958046,
      "learning_rate": 1.3059339165819082e-05,
      "loss": 0.5958,
      "step": 2178
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1060486707131345,
      "learning_rate": 1.3053406478074155e-05,
      "loss": 0.9862,
      "step": 2179
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0865574900481738,
      "learning_rate": 1.3047472604891701e-05,
      "loss": 0.97,
      "step": 2180
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0614518511690123,
      "learning_rate": 1.3041537548575455e-05,
      "loss": 0.972,
      "step": 2181
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.009527029278252,
      "learning_rate": 1.303560131142961e-05,
      "loss": 0.9646,
      "step": 2182
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9817420890337346,
      "learning_rate": 1.3029663895758814e-05,
      "loss": 0.9723,
      "step": 2183
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0751943900080994,
      "learning_rate": 1.3023725303868183e-05,
      "loss": 1.0017,
      "step": 2184
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0796817726814536,
      "learning_rate": 1.3017785538063277e-05,
      "loss": 0.9528,
      "step": 2185
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9132206200395585,
      "learning_rate": 1.3011844600650121e-05,
      "loss": 0.9307,
      "step": 2186
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9739291066444119,
      "learning_rate": 1.300590249393519e-05,
      "loss": 0.951,
      "step": 2187
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.208944149380384,
      "learning_rate": 1.2999959220225416e-05,
      "loss": 0.9545,
      "step": 2188
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9513512849506524,
      "learning_rate": 1.299401478182818e-05,
      "loss": 0.9331,
      "step": 2189
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0484396074617812,
      "learning_rate": 1.2988069181051314e-05,
      "loss": 0.9892,
      "step": 2190
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.077981091825546,
      "learning_rate": 1.2982122420203114e-05,
      "loss": 0.8733,
      "step": 2191
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1413887860073766,
      "learning_rate": 1.2976174501592313e-05,
      "loss": 0.9779,
      "step": 2192
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9097974808395103,
      "learning_rate": 1.2970225427528098e-05,
      "loss": 0.6554,
      "step": 2193
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0727642568437044,
      "learning_rate": 1.2964275200320104e-05,
      "loss": 0.9604,
      "step": 2194
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8130364890597642,
      "learning_rate": 1.2958323822278413e-05,
      "loss": 0.671,
      "step": 2195
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.175812771424554,
      "learning_rate": 1.2952371295713558e-05,
      "loss": 0.9309,
      "step": 2196
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1714528915029463,
      "learning_rate": 1.2946417622936512e-05,
      "loss": 0.9943,
      "step": 2197
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2249900159380296,
      "learning_rate": 1.2940462806258696e-05,
      "loss": 1.0019,
      "step": 2198
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.103569063859156,
      "learning_rate": 1.2934506847991976e-05,
      "loss": 0.9767,
      "step": 2199
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0461205503532722,
      "learning_rate": 1.2928549750448661e-05,
      "loss": 0.9394,
      "step": 2200
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0946023659607296,
      "learning_rate": 1.2922591515941498e-05,
      "loss": 0.8652,
      "step": 2201
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0971532344465382,
      "learning_rate": 1.2916632146783683e-05,
      "loss": 0.8786,
      "step": 2202
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.98020302973232,
      "learning_rate": 1.2910671645288841e-05,
      "loss": 0.9401,
      "step": 2203
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9815412161358561,
      "learning_rate": 1.2904710013771054e-05,
      "loss": 0.8606,
      "step": 2204
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0003526728600183,
      "learning_rate": 1.2898747254544826e-05,
      "loss": 0.9977,
      "step": 2205
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9961077151535038,
      "learning_rate": 1.2892783369925105e-05,
      "loss": 0.9005,
      "step": 2206
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9004513918888465,
      "learning_rate": 1.2886818362227283e-05,
      "loss": 0.9458,
      "step": 2207
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9500315045713743,
      "learning_rate": 1.2880852233767174e-05,
      "loss": 0.9628,
      "step": 2208
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8609064684135026,
      "learning_rate": 1.2874884986861038e-05,
      "loss": 0.6259,
      "step": 2209
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0299210876964462,
      "learning_rate": 1.2868916623825561e-05,
      "loss": 0.972,
      "step": 2210
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0599012579889413,
      "learning_rate": 1.2862947146977876e-05,
      "loss": 0.9553,
      "step": 2211
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0647719443866537,
      "learning_rate": 1.2856976558635532e-05,
      "loss": 0.9842,
      "step": 2212
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3970077305910338,
      "learning_rate": 1.2851004861116519e-05,
      "loss": 0.9751,
      "step": 2213
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.910179125425195,
      "learning_rate": 1.2845032056739257e-05,
      "loss": 0.9756,
      "step": 2214
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9610733461016528,
      "learning_rate": 1.2839058147822595e-05,
      "loss": 0.9592,
      "step": 2215
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9574451133117142,
      "learning_rate": 1.2833083136685803e-05,
      "loss": 0.9608,
      "step": 2216
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.04190274250636,
      "learning_rate": 1.2827107025648595e-05,
      "loss": 0.9514,
      "step": 2217
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1869079520165688,
      "learning_rate": 1.2821129817031099e-05,
      "loss": 0.9399,
      "step": 2218
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9737498423252342,
      "learning_rate": 1.2815151513153874e-05,
      "loss": 0.9951,
      "step": 2219
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9853847424116365,
      "learning_rate": 1.2809172116337903e-05,
      "loss": 0.9804,
      "step": 2220
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9108024244238638,
      "learning_rate": 1.2803191628904594e-05,
      "loss": 0.9666,
      "step": 2221
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9633320501197988,
      "learning_rate": 1.2797210053175779e-05,
      "loss": 0.9132,
      "step": 2222
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9560499326326626,
      "learning_rate": 1.2791227391473706e-05,
      "loss": 0.9727,
      "step": 2223
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1418829063184939,
      "learning_rate": 1.2785243646121059e-05,
      "loss": 0.9386,
      "step": 2224
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9977744340908431,
      "learning_rate": 1.277925881944093e-05,
      "loss": 0.9758,
      "step": 2225
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0653349631497595,
      "learning_rate": 1.2773272913756833e-05,
      "loss": 0.9591,
      "step": 2226
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9928140363378918,
      "learning_rate": 1.2767285931392705e-05,
      "loss": 0.9364,
      "step": 2227
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8995749471280335,
      "learning_rate": 1.27612978746729e-05,
      "loss": 0.6692,
      "step": 2228
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8121412464087874,
      "learning_rate": 1.2755308745922182e-05,
      "loss": 0.6593,
      "step": 2229
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0043160708552188,
      "learning_rate": 1.2749318547465742e-05,
      "loss": 0.9706,
      "step": 2230
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.033728860150438,
      "learning_rate": 1.2743327281629181e-05,
      "loss": 0.9455,
      "step": 2231
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0506333877271492,
      "learning_rate": 1.2737334950738512e-05,
      "loss": 0.9369,
      "step": 2232
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0175727260175464,
      "learning_rate": 1.273134155712017e-05,
      "loss": 0.9083,
      "step": 2233
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0609644507712634,
      "learning_rate": 1.272534710310099e-05,
      "loss": 1.019,
      "step": 2234
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9574592837395307,
      "learning_rate": 1.2719351591008228e-05,
      "loss": 0.9104,
      "step": 2235
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8667670599484406,
      "learning_rate": 1.2713355023169547e-05,
      "loss": 0.6313,
      "step": 2236
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0634244432279745,
      "learning_rate": 1.2707357401913022e-05,
      "loss": 0.9899,
      "step": 2237
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0269102484638657,
      "learning_rate": 1.270135872956714e-05,
      "loss": 0.938,
      "step": 2238
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0973426532724606,
      "learning_rate": 1.2695359008460785e-05,
      "loss": 0.9564,
      "step": 2239
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9982789728777495,
      "learning_rate": 1.2689358240923264e-05,
      "loss": 0.9912,
      "step": 2240
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0330749907299281,
      "learning_rate": 1.2683356429284273e-05,
      "loss": 0.9986,
      "step": 2241
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1536280179549765,
      "learning_rate": 1.2677353575873926e-05,
      "loss": 0.9565,
      "step": 2242
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0583422844269206,
      "learning_rate": 1.2671349683022736e-05,
      "loss": 1.0015,
      "step": 2243
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9023174528732267,
      "learning_rate": 1.2665344753061622e-05,
      "loss": 0.6471,
      "step": 2244
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1738998092480137,
      "learning_rate": 1.2659338788321904e-05,
      "loss": 0.9978,
      "step": 2245
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9751173813973308,
      "learning_rate": 1.2653331791135308e-05,
      "loss": 0.9116,
      "step": 2246
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1050132988061654,
      "learning_rate": 1.2647323763833952e-05,
      "loss": 1.0086,
      "step": 2247
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0415367089107628,
      "learning_rate": 1.264131470875036e-05,
      "loss": 0.9392,
      "step": 2248
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9813642430391322,
      "learning_rate": 1.2635304628217452e-05,
      "loss": 0.9031,
      "step": 2249
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0388435467177932,
      "learning_rate": 1.2629293524568555e-05,
      "loss": 1.0084,
      "step": 2250
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0125320729721237,
      "learning_rate": 1.2623281400137383e-05,
      "loss": 0.92,
      "step": 2251
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9363636335712715,
      "learning_rate": 1.2617268257258051e-05,
      "loss": 0.9708,
      "step": 2252
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9018286822765528,
      "learning_rate": 1.2611254098265063e-05,
      "loss": 0.879,
      "step": 2253
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8154384532617212,
      "learning_rate": 1.2605238925493326e-05,
      "loss": 0.6086,
      "step": 2254
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9758959282242075,
      "learning_rate": 1.2599222741278136e-05,
      "loss": 0.9814,
      "step": 2255
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9758250320128492,
      "learning_rate": 1.2593205547955185e-05,
      "loss": 0.9533,
      "step": 2256
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.011642968486025,
      "learning_rate": 1.2587187347860554e-05,
      "loss": 0.8878,
      "step": 2257
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1212335260163395,
      "learning_rate": 1.2581168143330716e-05,
      "loss": 0.8895,
      "step": 2258
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7931693176073368,
      "learning_rate": 1.2575147936702531e-05,
      "loss": 0.6367,
      "step": 2259
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8982983258507655,
      "learning_rate": 1.2569126730313255e-05,
      "loss": 0.6535,
      "step": 2260
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9854814230711264,
      "learning_rate": 1.2563104526500523e-05,
      "loss": 0.9678,
      "step": 2261
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8831658372290063,
      "learning_rate": 1.2557081327602361e-05,
      "loss": 0.6486,
      "step": 2262
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.040727600337553,
      "learning_rate": 1.2551057135957187e-05,
      "loss": 0.9689,
      "step": 2263
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9091637898867311,
      "learning_rate": 1.2545031953903796e-05,
      "loss": 0.9064,
      "step": 2264
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9808660061948969,
      "learning_rate": 1.2539005783781374e-05,
      "loss": 0.9309,
      "step": 2265
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9282406008057931,
      "learning_rate": 1.2532978627929486e-05,
      "loss": 1.0173,
      "step": 2266
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0837865029194975,
      "learning_rate": 1.2526950488688083e-05,
      "loss": 1.0104,
      "step": 2267
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.020371304969042,
      "learning_rate": 1.2520921368397492e-05,
      "loss": 0.9767,
      "step": 2268
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9511364732211104,
      "learning_rate": 1.2514891269398429e-05,
      "loss": 0.932,
      "step": 2269
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0670925630929475,
      "learning_rate": 1.2508860194031986e-05,
      "loss": 0.9873,
      "step": 2270
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1044178125229902,
      "learning_rate": 1.2502828144639629e-05,
      "loss": 0.8898,
      "step": 2271
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9759675543012291,
      "learning_rate": 1.2496795123563218e-05,
      "loss": 0.9522,
      "step": 2272
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0920106232027962,
      "learning_rate": 1.249076113314497e-05,
      "loss": 0.9535,
      "step": 2273
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1481995729427934,
      "learning_rate": 1.248472617572749e-05,
      "loss": 0.9182,
      "step": 2274
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0225722225950606,
      "learning_rate": 1.2478690253653756e-05,
      "loss": 0.9707,
      "step": 2275
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0077413937708475,
      "learning_rate": 1.2472653369267122e-05,
      "loss": 0.9766,
      "step": 2276
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0497710544166565,
      "learning_rate": 1.2466615524911316e-05,
      "loss": 0.9263,
      "step": 2277
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0720367226400496,
      "learning_rate": 1.2460576722930432e-05,
      "loss": 0.963,
      "step": 2278
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0685212672555209,
      "learning_rate": 1.2454536965668949e-05,
      "loss": 0.9431,
      "step": 2279
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9730914199985975,
      "learning_rate": 1.24484962554717e-05,
      "loss": 0.9403,
      "step": 2280
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0716707532881187,
      "learning_rate": 1.24424545946839e-05,
      "loss": 0.9541,
      "step": 2281
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9757090813964833,
      "learning_rate": 1.2436411985651131e-05,
      "loss": 0.9735,
      "step": 2282
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9445724814832464,
      "learning_rate": 1.2430368430719342e-05,
      "loss": 0.9179,
      "step": 2283
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0237005697365755,
      "learning_rate": 1.242432393223485e-05,
      "loss": 0.9856,
      "step": 2284
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7686204980559347,
      "learning_rate": 1.2418278492544328e-05,
      "loss": 0.6536,
      "step": 2285
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9941648274700139,
      "learning_rate": 1.2412232113994841e-05,
      "loss": 0.9375,
      "step": 2286
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9887446056568004,
      "learning_rate": 1.2406184798933786e-05,
      "loss": 0.9233,
      "step": 2287
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.025215386448336,
      "learning_rate": 1.2400136549708945e-05,
      "loss": 0.9354,
      "step": 2288
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9078044284128923,
      "learning_rate": 1.239408736866846e-05,
      "loss": 0.9416,
      "step": 2289
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8969308261171576,
      "learning_rate": 1.2388037258160823e-05,
      "loss": 0.6649,
      "step": 2290
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1125112896191378,
      "learning_rate": 1.23819862205349e-05,
      "loss": 0.9757,
      "step": 2291
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.027884132034291,
      "learning_rate": 1.2375934258139917e-05,
      "loss": 0.9681,
      "step": 2292
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1822571153141654,
      "learning_rate": 1.2369881373325448e-05,
      "loss": 0.9105,
      "step": 2293
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0581344834641775,
      "learning_rate": 1.236382756844143e-05,
      "loss": 0.94,
      "step": 2294
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8999828747862229,
      "learning_rate": 1.2357772845838159e-05,
      "loss": 0.6723,
      "step": 2295
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9662430343681048,
      "learning_rate": 1.2351717207866292e-05,
      "loss": 0.9581,
      "step": 2296
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0151983616706177,
      "learning_rate": 1.2345660656876832e-05,
      "loss": 1.0016,
      "step": 2297
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9784758493339845,
      "learning_rate": 1.233960319522114e-05,
      "loss": 0.9286,
      "step": 2298
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0851597071668786,
      "learning_rate": 1.2333544825250938e-05,
      "loss": 0.8937,
      "step": 2299
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0773160675183904,
      "learning_rate": 1.2327485549318285e-05,
      "loss": 0.9594,
      "step": 2300
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0915284012241124,
      "learning_rate": 1.2321425369775601e-05,
      "loss": 0.9673,
      "step": 2301
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9204212757218053,
      "learning_rate": 1.2315364288975665e-05,
      "loss": 0.7038,
      "step": 2302
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.001118328708697,
      "learning_rate": 1.2309302309271587e-05,
      "loss": 0.9192,
      "step": 2303
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9878586299671255,
      "learning_rate": 1.2303239433016842e-05,
      "loss": 0.9727,
      "step": 2304
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0078320531371405,
      "learning_rate": 1.2297175662565248e-05,
      "loss": 0.9558,
      "step": 2305
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.070228620904846,
      "learning_rate": 1.229111100027097e-05,
      "loss": 1.0114,
      "step": 2306
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9948259485247875,
      "learning_rate": 1.228504544848851e-05,
      "loss": 0.8807,
      "step": 2307
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9587107773586286,
      "learning_rate": 1.2278979009572736e-05,
      "loss": 0.9628,
      "step": 2308
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1356109743632878,
      "learning_rate": 1.2272911685878841e-05,
      "loss": 0.9818,
      "step": 2309
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1038476341135244,
      "learning_rate": 1.2266843479762372e-05,
      "loss": 0.974,
      "step": 2310
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9627661682617823,
      "learning_rate": 1.2260774393579209e-05,
      "loss": 0.9827,
      "step": 2311
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.256734930846636,
      "learning_rate": 1.2254704429685593e-05,
      "loss": 0.9374,
      "step": 2312
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0797876775671948,
      "learning_rate": 1.2248633590438084e-05,
      "loss": 1.0012,
      "step": 2313
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0168148860406019,
      "learning_rate": 1.2242561878193589e-05,
      "loss": 0.9327,
      "step": 2314
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9989915408731932,
      "learning_rate": 1.2236489295309362e-05,
      "loss": 0.906,
      "step": 2315
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.29366604112452,
      "learning_rate": 1.2230415844142984e-05,
      "loss": 1.0228,
      "step": 2316
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9789225261473751,
      "learning_rate": 1.2224341527052378e-05,
      "loss": 0.9433,
      "step": 2317
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.177264534209734,
      "learning_rate": 1.2218266346395811e-05,
      "loss": 0.9314,
      "step": 2318
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8591130429063089,
      "learning_rate": 1.221219030453187e-05,
      "loss": 0.9042,
      "step": 2319
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0069725731776518,
      "learning_rate": 1.220611340381948e-05,
      "loss": 0.9506,
      "step": 2320
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.986016056529033,
      "learning_rate": 1.2200035646617912e-05,
      "loss": 0.9377,
      "step": 2321
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8693704160021872,
      "learning_rate": 1.2193957035286757e-05,
      "loss": 0.6886,
      "step": 2322
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.028331470791232,
      "learning_rate": 1.2187877572185937e-05,
      "loss": 0.9295,
      "step": 2323
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1791873074358425,
      "learning_rate": 1.2181797259675713e-05,
      "loss": 0.9109,
      "step": 2324
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8243538774834306,
      "learning_rate": 1.2175716100116677e-05,
      "loss": 0.6489,
      "step": 2325
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0273829361337274,
      "learning_rate": 1.2169634095869736e-05,
      "loss": 0.9458,
      "step": 2326
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9909874018058621,
      "learning_rate": 1.2163551249296132e-05,
      "loss": 0.9245,
      "step": 2327
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9807225099395254,
      "learning_rate": 1.2157467562757443e-05,
      "loss": 0.9708,
      "step": 2328
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0038360861553786,
      "learning_rate": 1.2151383038615563e-05,
      "loss": 0.9534,
      "step": 2329
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.120218390263878,
      "learning_rate": 1.214529767923271e-05,
      "loss": 0.9341,
      "step": 2330
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9319572812779247,
      "learning_rate": 1.2139211486971436e-05,
      "loss": 0.95,
      "step": 2331
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9721608925661976,
      "learning_rate": 1.213312446419461e-05,
      "loss": 0.9391,
      "step": 2332
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0367864370438657,
      "learning_rate": 1.2127036613265418e-05,
      "loss": 0.9396,
      "step": 2333
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9957500633861931,
      "learning_rate": 1.2120947936547375e-05,
      "loss": 0.9088,
      "step": 2334
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.047198468470042,
      "learning_rate": 1.2114858436404322e-05,
      "loss": 0.9462,
      "step": 2335
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0020358656113395,
      "learning_rate": 1.2108768115200405e-05,
      "loss": 0.9585,
      "step": 2336
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7920169306110849,
      "learning_rate": 1.2102676975300095e-05,
      "loss": 0.5877,
      "step": 2337
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0846586392129267,
      "learning_rate": 1.209658501906819e-05,
      "loss": 0.9489,
      "step": 2338
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.251781184013554,
      "learning_rate": 1.2090492248869795e-05,
      "loss": 0.9915,
      "step": 2339
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8206897991883351,
      "learning_rate": 1.2084398667070325e-05,
      "loss": 0.6838,
      "step": 2340
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0025415473870407,
      "learning_rate": 1.2078304276035527e-05,
      "loss": 0.9046,
      "step": 2341
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1779097700071737,
      "learning_rate": 1.2072209078131451e-05,
      "loss": 1.0195,
      "step": 2342
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9503621798267227,
      "learning_rate": 1.2066113075724461e-05,
      "loss": 0.9043,
      "step": 2343
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0028340339083432,
      "learning_rate": 1.206001627118124e-05,
      "loss": 0.9238,
      "step": 2344
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9931627958565346,
      "learning_rate": 1.2053918666868776e-05,
      "loss": 0.9077,
      "step": 2345
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.230369938561389,
      "learning_rate": 1.2047820265154362e-05,
      "loss": 0.9709,
      "step": 2346
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9856413333885923,
      "learning_rate": 1.2041721068405614e-05,
      "loss": 0.9369,
      "step": 2347
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9975726276713848,
      "learning_rate": 1.203562107899045e-05,
      "loss": 0.9543,
      "step": 2348
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.176979373417783,
      "learning_rate": 1.2029520299277095e-05,
      "loss": 1.0597,
      "step": 2349
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9590354245946089,
      "learning_rate": 1.2023418731634078e-05,
      "loss": 0.8888,
      "step": 2350
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9971237683187649,
      "learning_rate": 1.2017316378430244e-05,
      "loss": 0.9432,
      "step": 2351
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9723636473614496,
      "learning_rate": 1.2011213242034733e-05,
      "loss": 0.9032,
      "step": 2352
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0584288256169099,
      "learning_rate": 1.2005109324816992e-05,
      "loss": 0.9059,
      "step": 2353
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.883505630093833,
      "learning_rate": 1.1999004629146775e-05,
      "loss": 0.8984,
      "step": 2354
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9737905333440277,
      "learning_rate": 1.1992899157394133e-05,
      "loss": 0.9465,
      "step": 2355
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8784027254670288,
      "learning_rate": 1.1986792911929418e-05,
      "loss": 0.6617,
      "step": 2356
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0525470960295973,
      "learning_rate": 1.198068589512329e-05,
      "loss": 0.9675,
      "step": 2357
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0917708533705635,
      "learning_rate": 1.1974578109346702e-05,
      "loss": 0.9258,
      "step": 2358
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8692465878303917,
      "learning_rate": 1.1968469556970905e-05,
      "loss": 0.6548,
      "step": 2359
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1089127609759333,
      "learning_rate": 1.1962360240367445e-05,
      "loss": 0.9363,
      "step": 2360
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.918381840762901,
      "learning_rate": 1.1956250161908179e-05,
      "loss": 0.6549,
      "step": 2361
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9975388905880246,
      "learning_rate": 1.195013932396524e-05,
      "loss": 0.9767,
      "step": 2362
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.062879757035255,
      "learning_rate": 1.1944027728911072e-05,
      "loss": 0.9984,
      "step": 2363
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9893166847800783,
      "learning_rate": 1.1937915379118406e-05,
      "loss": 0.9301,
      "step": 2364
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8333197123810087,
      "learning_rate": 1.1931802276960265e-05,
      "loss": 0.6817,
      "step": 2365
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0536538712743266,
      "learning_rate": 1.1925688424809965e-05,
      "loss": 0.9799,
      "step": 2366
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9716124929504492,
      "learning_rate": 1.1919573825041115e-05,
      "loss": 0.9225,
      "step": 2367
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0367629379341454,
      "learning_rate": 1.1913458480027614e-05,
      "loss": 0.9458,
      "step": 2368
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9949887834407497,
      "learning_rate": 1.1907342392143646e-05,
      "loss": 0.9935,
      "step": 2369
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.163951476054191,
      "learning_rate": 1.1901225563763694e-05,
      "loss": 0.9955,
      "step": 2370
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9598918580214013,
      "learning_rate": 1.1895107997262516e-05,
      "loss": 0.9714,
      "step": 2371
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9131257478174771,
      "learning_rate": 1.1888989695015166e-05,
      "loss": 0.9159,
      "step": 2372
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9299791325309493,
      "learning_rate": 1.1882870659396968e-05,
      "loss": 0.9185,
      "step": 2373
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0616914628008611,
      "learning_rate": 1.1876750892783558e-05,
      "loss": 0.9967,
      "step": 2374
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1344342071967375,
      "learning_rate": 1.1870630397550831e-05,
      "loss": 0.9049,
      "step": 2375
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0086299929157085,
      "learning_rate": 1.1864509176074974e-05,
      "loss": 1.002,
      "step": 2376
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2295192146663598,
      "learning_rate": 1.185838723073246e-05,
      "loss": 0.9327,
      "step": 2377
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0162901336063341,
      "learning_rate": 1.1852264563900038e-05,
      "loss": 0.8845,
      "step": 2378
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9856213378974285,
      "learning_rate": 1.1846141177954733e-05,
      "loss": 0.9718,
      "step": 2379
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9817678697237351,
      "learning_rate": 1.1840017075273861e-05,
      "loss": 0.9719,
      "step": 2380
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0470108495203398,
      "learning_rate": 1.1833892258235008e-05,
      "loss": 0.9359,
      "step": 2381
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1348862729551419,
      "learning_rate": 1.1827766729216035e-05,
      "loss": 0.9929,
      "step": 2382
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0590903021110039,
      "learning_rate": 1.1821640490595086e-05,
      "loss": 0.9565,
      "step": 2383
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.007335741593861,
      "learning_rate": 1.181551354475058e-05,
      "loss": 0.926,
      "step": 2384
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9590201713519912,
      "learning_rate": 1.1809385894061206e-05,
      "loss": 0.9084,
      "step": 2385
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8203510791925543,
      "learning_rate": 1.1803257540905926e-05,
      "loss": 0.6593,
      "step": 2386
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1218183220390496,
      "learning_rate": 1.1797128487663982e-05,
      "loss": 0.9858,
      "step": 2387
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7948635063633411,
      "learning_rate": 1.1790998736714882e-05,
      "loss": 0.669,
      "step": 2388
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8969412239020462,
      "learning_rate": 1.1784868290438404e-05,
      "loss": 0.8726,
      "step": 2389
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1126618957783088,
      "learning_rate": 1.1778737151214606e-05,
      "loss": 0.9661,
      "step": 2390
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1253450929410804,
      "learning_rate": 1.17726053214238e-05,
      "loss": 0.9833,
      "step": 2391
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.823262754592919,
      "learning_rate": 1.1766472803446577e-05,
      "loss": 0.7167,
      "step": 2392
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0785455418034529,
      "learning_rate": 1.1760339599663788e-05,
      "loss": 0.9933,
      "step": 2393
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.031042329376684,
      "learning_rate": 1.1754205712456556e-05,
      "loss": 0.9745,
      "step": 2394
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9417524219440387,
      "learning_rate": 1.1748071144206266e-05,
      "loss": 0.9213,
      "step": 2395
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9834234836802013,
      "learning_rate": 1.1741935897294572e-05,
      "loss": 0.9179,
      "step": 2396
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9654389627533191,
      "learning_rate": 1.1735799974103388e-05,
      "loss": 0.9391,
      "step": 2397
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9462570193862145,
      "learning_rate": 1.1729663377014888e-05,
      "loss": 0.9202,
      "step": 2398
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9646477066953678,
      "learning_rate": 1.172352610841151e-05,
      "loss": 0.9432,
      "step": 2399
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0886927847758474,
      "learning_rate": 1.1717388170675954e-05,
      "loss": 0.9936,
      "step": 2400
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.002142152188504,
      "learning_rate": 1.1711249566191179e-05,
      "loss": 0.9342,
      "step": 2401
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0858365042379772,
      "learning_rate": 1.17051102973404e-05,
      "loss": 0.958,
      "step": 2402
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9720837467024854,
      "learning_rate": 1.1698970366507096e-05,
      "loss": 0.9771,
      "step": 2403
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0464817298586608,
      "learning_rate": 1.1692829776074999e-05,
      "loss": 0.8843,
      "step": 2404
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.948726977991268,
      "learning_rate": 1.1686688528428099e-05,
      "loss": 0.9112,
      "step": 2405
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0108018325400838,
      "learning_rate": 1.1680546625950635e-05,
      "loss": 0.9952,
      "step": 2406
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0719295937751427,
      "learning_rate": 1.167440407102711e-05,
      "loss": 0.9205,
      "step": 2407
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.037067830291738,
      "learning_rate": 1.1668260866042271e-05,
      "loss": 0.9284,
      "step": 2408
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2782183177348898,
      "learning_rate": 1.1662117013381126e-05,
      "loss": 0.9405,
      "step": 2409
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9761696992300742,
      "learning_rate": 1.1655972515428928e-05,
      "loss": 0.9378,
      "step": 2410
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.072372354561641,
      "learning_rate": 1.1649827374571182e-05,
      "loss": 0.9934,
      "step": 2411
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1078035823274108,
      "learning_rate": 1.1643681593193642e-05,
      "loss": 0.9563,
      "step": 2412
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0459197790842876,
      "learning_rate": 1.1637535173682318e-05,
      "loss": 0.9312,
      "step": 2413
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0414966728533346,
      "learning_rate": 1.1631388118423457e-05,
      "loss": 0.9336,
      "step": 2414
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.059581677534091,
      "learning_rate": 1.1625240429803553e-05,
      "loss": 0.9828,
      "step": 2415
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.024287852430201,
      "learning_rate": 1.1619092110209361e-05,
      "loss": 0.9046,
      "step": 2416
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9111530929303626,
      "learning_rate": 1.1612943162027863e-05,
      "loss": 0.669,
      "step": 2417
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0445738354290741,
      "learning_rate": 1.1606793587646295e-05,
      "loss": 0.9474,
      "step": 2418
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9894698674524531,
      "learning_rate": 1.160064338945213e-05,
      "loss": 0.9123,
      "step": 2419
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1523161749450006,
      "learning_rate": 1.1594492569833093e-05,
      "loss": 0.9702,
      "step": 2420
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0996766232869057,
      "learning_rate": 1.1588341131177137e-05,
      "loss": 0.9488,
      "step": 2421
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1025155312431913,
      "learning_rate": 1.1582189075872467e-05,
      "loss": 0.9454,
      "step": 2422
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9258064464518406,
      "learning_rate": 1.1576036406307523e-05,
      "loss": 0.9078,
      "step": 2423
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.050304059090224,
      "learning_rate": 1.156988312487098e-05,
      "loss": 0.901,
      "step": 2424
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0491660411097363,
      "learning_rate": 1.1563729233951757e-05,
      "loss": 0.9765,
      "step": 2425
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0673760141677955,
      "learning_rate": 1.1557574735939003e-05,
      "loss": 0.9151,
      "step": 2426
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9657693774511197,
      "learning_rate": 1.1551419633222107e-05,
      "loss": 0.9138,
      "step": 2427
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0365170693008718,
      "learning_rate": 1.1545263928190692e-05,
      "loss": 0.9322,
      "step": 2428
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9570334029416762,
      "learning_rate": 1.1539107623234618e-05,
      "loss": 0.9262,
      "step": 2429
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9757321923975381,
      "learning_rate": 1.153295072074397e-05,
      "loss": 0.9146,
      "step": 2430
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9202636583190308,
      "learning_rate": 1.1526793223109072e-05,
      "loss": 0.8995,
      "step": 2431
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1214396199916323,
      "learning_rate": 1.1520635132720475e-05,
      "loss": 0.9769,
      "step": 2432
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1970366283083553,
      "learning_rate": 1.1514476451968961e-05,
      "loss": 0.8993,
      "step": 2433
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9372408198178176,
      "learning_rate": 1.1508317183245545e-05,
      "loss": 0.9336,
      "step": 2434
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0022926055220596,
      "learning_rate": 1.1502157328941466e-05,
      "loss": 0.9758,
      "step": 2435
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.019576469168554,
      "learning_rate": 1.149599689144819e-05,
      "loss": 0.9264,
      "step": 2436
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0440245157605132,
      "learning_rate": 1.1489835873157414e-05,
      "loss": 0.9624,
      "step": 2437
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9724601182572162,
      "learning_rate": 1.1483674276461053e-05,
      "loss": 0.9485,
      "step": 2438
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0225556488220575,
      "learning_rate": 1.1477512103751254e-05,
      "loss": 0.9303,
      "step": 2439
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9387318464146085,
      "learning_rate": 1.1471349357420384e-05,
      "loss": 0.966,
      "step": 2440
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9509446989757159,
      "learning_rate": 1.1465186039861033e-05,
      "loss": 0.9,
      "step": 2441
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9693288022152836,
      "learning_rate": 1.1459022153466016e-05,
      "loss": 0.9416,
      "step": 2442
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.884805679598823,
      "learning_rate": 1.1452857700628362e-05,
      "loss": 0.9364,
      "step": 2443
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0062444670772004,
      "learning_rate": 1.1446692683741326e-05,
      "loss": 0.9291,
      "step": 2444
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.030946497706631,
      "learning_rate": 1.1440527105198377e-05,
      "loss": 0.8944,
      "step": 2445
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0304280433990904,
      "learning_rate": 1.143436096739321e-05,
      "loss": 0.917,
      "step": 2446
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9764460435809176,
      "learning_rate": 1.1428194272719729e-05,
      "loss": 0.8611,
      "step": 2447
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0546198982190198,
      "learning_rate": 1.1422027023572052e-05,
      "loss": 0.9572,
      "step": 2448
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9468133095977206,
      "learning_rate": 1.1415859222344525e-05,
      "loss": 0.9046,
      "step": 2449
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9881838982160628,
      "learning_rate": 1.14096908714317e-05,
      "loss": 0.9174,
      "step": 2450
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9605414224369428,
      "learning_rate": 1.1403521973228342e-05,
      "loss": 0.933,
      "step": 2451
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.007075625323075,
      "learning_rate": 1.1397352530129428e-05,
      "loss": 0.9437,
      "step": 2452
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.173040957964704,
      "learning_rate": 1.139118254453015e-05,
      "loss": 0.9814,
      "step": 2453
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9977936860951364,
      "learning_rate": 1.1385012018825907e-05,
      "loss": 0.9444,
      "step": 2454
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1002435261560408,
      "learning_rate": 1.1378840955412313e-05,
      "loss": 0.9525,
      "step": 2455
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0143967923140398,
      "learning_rate": 1.1372669356685185e-05,
      "loss": 0.9544,
      "step": 2456
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0193530864263443,
      "learning_rate": 1.1366497225040549e-05,
      "loss": 0.9399,
      "step": 2457
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8806355383879843,
      "learning_rate": 1.1360324562874643e-05,
      "loss": 0.8685,
      "step": 2458
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8611037536246636,
      "learning_rate": 1.1354151372583901e-05,
      "loss": 0.6173,
      "step": 2459
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0255122562203627,
      "learning_rate": 1.1347977656564974e-05,
      "loss": 0.9273,
      "step": 2460
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.991847600499858,
      "learning_rate": 1.1341803417214705e-05,
      "loss": 0.9214,
      "step": 2461
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0515108502617847,
      "learning_rate": 1.1335628656930153e-05,
      "loss": 0.9731,
      "step": 2462
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9182180761035168,
      "learning_rate": 1.132945337810857e-05,
      "loss": 0.9223,
      "step": 2463
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7978148854672382,
      "learning_rate": 1.132327758314741e-05,
      "loss": 0.6681,
      "step": 2464
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8421116241045838,
      "learning_rate": 1.131710127444433e-05,
      "loss": 0.6708,
      "step": 2465
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0392606432474172,
      "learning_rate": 1.1310924454397187e-05,
      "loss": 0.9368,
      "step": 2466
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9836179460824802,
      "learning_rate": 1.1304747125404031e-05,
      "loss": 0.9254,
      "step": 2467
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.215221267026826,
      "learning_rate": 1.129856928986312e-05,
      "loss": 0.9252,
      "step": 2468
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1047001441804554,
      "learning_rate": 1.12923909501729e-05,
      "loss": 1.0208,
      "step": 2469
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9731117117152279,
      "learning_rate": 1.1286212108732015e-05,
      "loss": 0.9661,
      "step": 2470
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.084489935179509,
      "learning_rate": 1.1280032767939302e-05,
      "loss": 0.9186,
      "step": 2471
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.951911702314376,
      "learning_rate": 1.1273852930193798e-05,
      "loss": 0.9218,
      "step": 2472
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9888850982851494,
      "learning_rate": 1.1267672597894725e-05,
      "loss": 0.9761,
      "step": 2473
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8313559876619007,
      "learning_rate": 1.12614917734415e-05,
      "loss": 0.7022,
      "step": 2474
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2936287525105343,
      "learning_rate": 1.1255310459233737e-05,
      "loss": 0.9953,
      "step": 2475
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9090694907201724,
      "learning_rate": 1.1249128657671233e-05,
      "loss": 0.9098,
      "step": 2476
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.010658758746886,
      "learning_rate": 1.1242946371153974e-05,
      "loss": 0.9354,
      "step": 2477
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.16049562433025,
      "learning_rate": 1.1236763602082136e-05,
      "loss": 1.0004,
      "step": 2478
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9289845589760762,
      "learning_rate": 1.1230580352856088e-05,
      "loss": 0.9387,
      "step": 2479
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.073886018287321,
      "learning_rate": 1.1224396625876375e-05,
      "loss": 0.9302,
      "step": 2480
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0100136718324069,
      "learning_rate": 1.1218212423543734e-05,
      "loss": 0.9014,
      "step": 2481
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9567468640321242,
      "learning_rate": 1.1212027748259086e-05,
      "loss": 0.9614,
      "step": 2482
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0335552788604465,
      "learning_rate": 1.1205842602423537e-05,
      "loss": 0.9693,
      "step": 2483
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0011716885069792,
      "learning_rate": 1.1199656988438373e-05,
      "loss": 0.9579,
      "step": 2484
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9671120040925459,
      "learning_rate": 1.1193470908705055e-05,
      "loss": 0.947,
      "step": 2485
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2404315115978755,
      "learning_rate": 1.1187284365625241e-05,
      "loss": 0.9994,
      "step": 2486
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9976309498225975,
      "learning_rate": 1.1181097361600754e-05,
      "loss": 0.9718,
      "step": 2487
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9239599693830667,
      "learning_rate": 1.1174909899033608e-05,
      "loss": 0.9482,
      "step": 2488
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1194558966530666,
      "learning_rate": 1.1168721980325987e-05,
      "loss": 0.9492,
      "step": 2489
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8808396293854553,
      "learning_rate": 1.1162533607880251e-05,
      "loss": 0.9436,
      "step": 2490
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7788428134635162,
      "learning_rate": 1.1156344784098942e-05,
      "loss": 0.6834,
      "step": 2491
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0158148036864993,
      "learning_rate": 1.1150155511384772e-05,
      "loss": 0.9302,
      "step": 2492
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1501572804196327,
      "learning_rate": 1.1143965792140631e-05,
      "loss": 0.9504,
      "step": 2493
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9699953830045867,
      "learning_rate": 1.1137775628769584e-05,
      "loss": 0.8963,
      "step": 2494
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1273172846322639,
      "learning_rate": 1.1131585023674863e-05,
      "loss": 0.9882,
      "step": 2495
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0375946083024081,
      "learning_rate": 1.1125393979259874e-05,
      "loss": 0.9569,
      "step": 2496
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9264261298606921,
      "learning_rate": 1.1119202497928192e-05,
      "loss": 0.9353,
      "step": 2497
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0174467352959207,
      "learning_rate": 1.1113010582083568e-05,
      "loss": 0.9446,
      "step": 2498
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0529315762457785,
      "learning_rate": 1.1106818234129913e-05,
      "loss": 0.9482,
      "step": 2499
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.094845799963352,
      "learning_rate": 1.1100625456471307e-05,
      "loss": 0.9027,
      "step": 2500
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.09646478612693,
      "learning_rate": 1.1094432251512006e-05,
      "loss": 0.9474,
      "step": 2501
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0378735613039392,
      "learning_rate": 1.1088238621656422e-05,
      "loss": 0.9034,
      "step": 2502
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9240115928132637,
      "learning_rate": 1.1082044569309138e-05,
      "loss": 0.6506,
      "step": 2503
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9881017579894996,
      "learning_rate": 1.1075850096874894e-05,
      "loss": 0.9327,
      "step": 2504
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9699060066590458,
      "learning_rate": 1.1069655206758603e-05,
      "loss": 0.959,
      "step": 2505
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9230469577553575,
      "learning_rate": 1.1063459901365325e-05,
      "loss": 0.6431,
      "step": 2506
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.004718210958251,
      "learning_rate": 1.1057264183100303e-05,
      "loss": 0.92,
      "step": 2507
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0101990452100014,
      "learning_rate": 1.1051068054368921e-05,
      "loss": 0.9671,
      "step": 2508
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0328648870783101,
      "learning_rate": 1.104487151757673e-05,
      "loss": 0.9562,
      "step": 2509
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8905684603241096,
      "learning_rate": 1.1038674575129442e-05,
      "loss": 0.9819,
      "step": 2510
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.979071170203725,
      "learning_rate": 1.1032477229432921e-05,
      "loss": 0.9097,
      "step": 2511
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9748140534167112,
      "learning_rate": 1.1026279482893187e-05,
      "loss": 0.9627,
      "step": 2512
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1052131700692673,
      "learning_rate": 1.1020081337916425e-05,
      "loss": 1.0082,
      "step": 2513
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9620554776492087,
      "learning_rate": 1.1013882796908963e-05,
      "loss": 0.852,
      "step": 2514
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.247225456400385,
      "learning_rate": 1.1007683862277292e-05,
      "loss": 0.8867,
      "step": 2515
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8256063653594924,
      "learning_rate": 1.1001484536428052e-05,
      "loss": 0.6456,
      "step": 2516
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9086167488299296,
      "learning_rate": 1.0995284821768029e-05,
      "loss": 0.9403,
      "step": 2517
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1287692472167992,
      "learning_rate": 1.098908472070417e-05,
      "loss": 0.9813,
      "step": 2518
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9627068072832564,
      "learning_rate": 1.0982884235643567e-05,
      "loss": 0.946,
      "step": 2519
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9718383038993124,
      "learning_rate": 1.0976683368993464e-05,
      "loss": 0.9034,
      "step": 2520
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0941340093794,
      "learning_rate": 1.0970482123161249e-05,
      "loss": 0.9678,
      "step": 2521
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.08159625544285,
      "learning_rate": 1.0964280500554459e-05,
      "loss": 0.9824,
      "step": 2522
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9354390747224384,
      "learning_rate": 1.0958078503580776e-05,
      "loss": 0.9113,
      "step": 2523
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9873457216447856,
      "learning_rate": 1.0951876134648032e-05,
      "loss": 0.9258,
      "step": 2524
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.05156005702153,
      "learning_rate": 1.0945673396164198e-05,
      "loss": 0.9935,
      "step": 2525
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0597248325096844,
      "learning_rate": 1.0939470290537389e-05,
      "loss": 0.9387,
      "step": 2526
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.086638432918338,
      "learning_rate": 1.0933266820175868e-05,
      "loss": 1.0089,
      "step": 2527
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0729897918644016,
      "learning_rate": 1.0927062987488035e-05,
      "loss": 0.9413,
      "step": 2528
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0801375193415559,
      "learning_rate": 1.0920858794882429e-05,
      "loss": 0.9336,
      "step": 2529
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9680169999071585,
      "learning_rate": 1.0914654244767736e-05,
      "loss": 0.9346,
      "step": 2530
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1493753975269354,
      "learning_rate": 1.0908449339552769e-05,
      "loss": 0.9378,
      "step": 2531
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0721556200167073,
      "learning_rate": 1.0902244081646489e-05,
      "loss": 0.8809,
      "step": 2532
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0536649470107888,
      "learning_rate": 1.0896038473457993e-05,
      "loss": 0.9674,
      "step": 2533
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0804413108106679,
      "learning_rate": 1.0889832517396511e-05,
      "loss": 0.9252,
      "step": 2534
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9588834701960526,
      "learning_rate": 1.0883626215871408e-05,
      "loss": 0.9339,
      "step": 2535
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8041193022854753,
      "learning_rate": 1.0877419571292183e-05,
      "loss": 0.595,
      "step": 2536
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9581012874656286,
      "learning_rate": 1.0871212586068469e-05,
      "loss": 0.897,
      "step": 2537
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9227903678828809,
      "learning_rate": 1.0865005262610033e-05,
      "loss": 0.9404,
      "step": 2538
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0521057553093445,
      "learning_rate": 1.085879760332677e-05,
      "loss": 0.944,
      "step": 2539
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0915090076064182,
      "learning_rate": 1.085258961062871e-05,
      "loss": 0.9744,
      "step": 2540
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0988858593815214,
      "learning_rate": 1.0846381286926007e-05,
      "loss": 0.9493,
      "step": 2541
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1594841749206612,
      "learning_rate": 1.0840172634628948e-05,
      "loss": 0.8993,
      "step": 2542
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3291519496736632,
      "learning_rate": 1.0833963656147944e-05,
      "loss": 0.9477,
      "step": 2543
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2745077321644425,
      "learning_rate": 1.082775435389353e-05,
      "loss": 1.0121,
      "step": 2544
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0817133220387372,
      "learning_rate": 1.0821544730276379e-05,
      "loss": 0.9654,
      "step": 2545
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.025891805056605,
      "learning_rate": 1.0815334787707277e-05,
      "loss": 0.9409,
      "step": 2546
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9811945797778753,
      "learning_rate": 1.0809124528597138e-05,
      "loss": 0.9819,
      "step": 2547
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.929415384883154,
      "learning_rate": 1.0802913955356998e-05,
      "loss": 0.9257,
      "step": 2548
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9285465620485437,
      "learning_rate": 1.0796703070398016e-05,
      "loss": 0.8506,
      "step": 2549
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1948235909889422,
      "learning_rate": 1.079049187613147e-05,
      "loss": 0.9767,
      "step": 2550
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1574167089334113,
      "learning_rate": 1.0784280374968761e-05,
      "loss": 0.9327,
      "step": 2551
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.064611805016352,
      "learning_rate": 1.0778068569321403e-05,
      "loss": 0.9217,
      "step": 2552
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9828348013649558,
      "learning_rate": 1.077185646160104e-05,
      "loss": 0.9408,
      "step": 2553
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0594602945228775,
      "learning_rate": 1.0765644054219422e-05,
      "loss": 0.9033,
      "step": 2554
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0177896601468028,
      "learning_rate": 1.0759431349588421e-05,
      "loss": 0.9273,
      "step": 2555
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7903485908951121,
      "learning_rate": 1.0753218350120023e-05,
      "loss": 0.602,
      "step": 2556
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1366594860761805,
      "learning_rate": 1.0747005058226325e-05,
      "loss": 0.9223,
      "step": 2557
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1314271935759341,
      "learning_rate": 1.0740791476319543e-05,
      "loss": 1.0345,
      "step": 2558
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0609340955334818,
      "learning_rate": 1.0734577606812007e-05,
      "loss": 0.9565,
      "step": 2559
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9993793271050031,
      "learning_rate": 1.0728363452116149e-05,
      "loss": 0.9267,
      "step": 2560
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9591173470212444,
      "learning_rate": 1.0722149014644523e-05,
      "loss": 0.9459,
      "step": 2561
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9768505055599782,
      "learning_rate": 1.0715934296809782e-05,
      "loss": 0.8767,
      "step": 2562
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0481325909893995,
      "learning_rate": 1.0709719301024698e-05,
      "loss": 0.9363,
      "step": 2563
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0520395514337604,
      "learning_rate": 1.0703504029702148e-05,
      "loss": 0.9326,
      "step": 2564
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0993548247047609,
      "learning_rate": 1.0697288485255107e-05,
      "loss": 0.9282,
      "step": 2565
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0678530596905293,
      "learning_rate": 1.0691072670096669e-05,
      "loss": 0.9762,
      "step": 2566
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.960297519463882,
      "learning_rate": 1.0684856586640026e-05,
      "loss": 0.915,
      "step": 2567
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9891170225475254,
      "learning_rate": 1.0678640237298476e-05,
      "loss": 0.6684,
      "step": 2568
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0960578108518406,
      "learning_rate": 1.0672423624485423e-05,
      "loss": 0.9742,
      "step": 2569
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0793426934704844,
      "learning_rate": 1.0666206750614363e-05,
      "loss": 0.9518,
      "step": 2570
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.98782315905771,
      "learning_rate": 1.0659989618098904e-05,
      "loss": 0.9539,
      "step": 2571
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1633858592298647,
      "learning_rate": 1.065377222935275e-05,
      "loss": 0.9329,
      "step": 2572
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7816387300168104,
      "learning_rate": 1.0647554586789708e-05,
      "loss": 0.6409,
      "step": 2573
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0835799183294497,
      "learning_rate": 1.064133669282368e-05,
      "loss": 0.9081,
      "step": 2574
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0635737787602535,
      "learning_rate": 1.0635118549868668e-05,
      "loss": 0.939,
      "step": 2575
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1455746571842034,
      "learning_rate": 1.0628900160338764e-05,
      "loss": 0.9297,
      "step": 2576
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9707079231020816,
      "learning_rate": 1.0622681526648167e-05,
      "loss": 0.9415,
      "step": 2577
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0293994247158087,
      "learning_rate": 1.0616462651211156e-05,
      "loss": 0.878,
      "step": 2578
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.015594217631712,
      "learning_rate": 1.0610243536442125e-05,
      "loss": 0.9013,
      "step": 2579
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0743972302765108,
      "learning_rate": 1.0604024184755539e-05,
      "loss": 0.8933,
      "step": 2580
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9184006088565891,
      "learning_rate": 1.0597804598565969e-05,
      "loss": 0.9456,
      "step": 2581
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8493103472603298,
      "learning_rate": 1.0591584780288069e-05,
      "loss": 0.6281,
      "step": 2582
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9916606174320888,
      "learning_rate": 1.0585364732336587e-05,
      "loss": 0.9329,
      "step": 2583
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8969840810605728,
      "learning_rate": 1.0579144457126365e-05,
      "loss": 0.9237,
      "step": 2584
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.944028879648287,
      "learning_rate": 1.057292395707232e-05,
      "loss": 0.9066,
      "step": 2585
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0220841430349379,
      "learning_rate": 1.0566703234589471e-05,
      "loss": 0.9626,
      "step": 2586
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9001578341141783,
      "learning_rate": 1.0560482292092912e-05,
      "loss": 0.868,
      "step": 2587
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9834564470306032,
      "learning_rate": 1.0554261131997833e-05,
      "loss": 0.9456,
      "step": 2588
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8499781296815381,
      "learning_rate": 1.0548039756719497e-05,
      "loss": 0.9069,
      "step": 2589
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.969844607806391,
      "learning_rate": 1.054181816867326e-05,
      "loss": 0.911,
      "step": 2590
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.848787374051497,
      "learning_rate": 1.053559637027455e-05,
      "loss": 0.6577,
      "step": 2591
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.984893545523589,
      "learning_rate": 1.0529374363938888e-05,
      "loss": 0.9172,
      "step": 2592
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1505102112032433,
      "learning_rate": 1.0523152152081875e-05,
      "loss": 0.9507,
      "step": 2593
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9476686115738208,
      "learning_rate": 1.051692973711918e-05,
      "loss": 0.9059,
      "step": 2594
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.036550549996605,
      "learning_rate": 1.0510707121466568e-05,
      "loss": 0.9185,
      "step": 2595
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0851746802690807,
      "learning_rate": 1.0504484307539864e-05,
      "loss": 0.9798,
      "step": 2596
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9273643050440509,
      "learning_rate": 1.0498261297754984e-05,
      "loss": 0.9053,
      "step": 2597
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9540434150854656,
      "learning_rate": 1.0492038094527907e-05,
      "loss": 0.9123,
      "step": 2598
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7780697857918157,
      "learning_rate": 1.0485814700274706e-05,
      "loss": 0.6197,
      "step": 2599
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0045746560402782,
      "learning_rate": 1.047959111741151e-05,
      "loss": 0.8836,
      "step": 2600
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.971157786251059,
      "learning_rate": 1.0473367348354529e-05,
      "loss": 0.9464,
      "step": 2601
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0314717929157289,
      "learning_rate": 1.0467143395520044e-05,
      "loss": 0.9106,
      "step": 2602
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0039804647803812,
      "learning_rate": 1.046091926132441e-05,
      "loss": 0.9504,
      "step": 2603
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9788031915739958,
      "learning_rate": 1.0454694948184045e-05,
      "loss": 0.9466,
      "step": 2604
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7727653666716308,
      "learning_rate": 1.044847045851545e-05,
      "loss": 0.6418,
      "step": 2605
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.958527224540909,
      "learning_rate": 1.044224579473518e-05,
      "loss": 0.9392,
      "step": 2606
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0356114232491926,
      "learning_rate": 1.0436020959259862e-05,
      "loss": 0.9718,
      "step": 2607
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9818647484757145,
      "learning_rate": 1.0429795954506203e-05,
      "loss": 0.9428,
      "step": 2608
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8704771538610622,
      "learning_rate": 1.0423570782890951e-05,
      "loss": 0.6388,
      "step": 2609
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.013945431217143,
      "learning_rate": 1.0417345446830938e-05,
      "loss": 0.9371,
      "step": 2610
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.071999229716194,
      "learning_rate": 1.0411119948743052e-05,
      "loss": 0.9719,
      "step": 2611
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9234686273253915,
      "learning_rate": 1.0404894291044247e-05,
      "loss": 0.9553,
      "step": 2612
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2228313842772578,
      "learning_rate": 1.0398668476151538e-05,
      "loss": 1.0197,
      "step": 2613
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0784330072764652,
      "learning_rate": 1.0392442506482e-05,
      "loss": 0.9277,
      "step": 2614
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8143112012650012,
      "learning_rate": 1.038621638445277e-05,
      "loss": 0.6352,
      "step": 2615
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9834219630002446,
      "learning_rate": 1.037999011248104e-05,
      "loss": 0.9027,
      "step": 2616
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1089975083553676,
      "learning_rate": 1.0373763692984062e-05,
      "loss": 0.9866,
      "step": 2617
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0546229360386628,
      "learning_rate": 1.0367537128379154e-05,
      "loss": 0.9213,
      "step": 2618
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.735053181983094,
      "learning_rate": 1.0361310421083677e-05,
      "loss": 0.6289,
      "step": 2619
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9950735905730327,
      "learning_rate": 1.0355083573515052e-05,
      "loss": 0.8517,
      "step": 2620
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9037837571740506,
      "learning_rate": 1.0348856588090764e-05,
      "loss": 0.9152,
      "step": 2621
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9955787206090285,
      "learning_rate": 1.0342629467228331e-05,
      "loss": 0.9785,
      "step": 2622
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0183765204346755,
      "learning_rate": 1.0336402213345345e-05,
      "loss": 0.9665,
      "step": 2623
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8488877296860831,
      "learning_rate": 1.0330174828859434e-05,
      "loss": 0.6423,
      "step": 2624
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7167745071553957,
      "learning_rate": 1.0323947316188288e-05,
      "loss": 0.6126,
      "step": 2625
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0539255412003155,
      "learning_rate": 1.031771967774964e-05,
      "loss": 0.9464,
      "step": 2626
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8973465359883848,
      "learning_rate": 1.0311491915961271e-05,
      "loss": 0.918,
      "step": 2627
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0380253574515912,
      "learning_rate": 1.030526403324102e-05,
      "loss": 0.9709,
      "step": 2628
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9735866348436765,
      "learning_rate": 1.0299036032006759e-05,
      "loss": 0.9597,
      "step": 2629
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9563989053715906,
      "learning_rate": 1.0292807914676412e-05,
      "loss": 0.9258,
      "step": 2630
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1196775417519855,
      "learning_rate": 1.0286579683667952e-05,
      "loss": 0.9417,
      "step": 2631
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9417007452358923,
      "learning_rate": 1.0280351341399392e-05,
      "loss": 0.8931,
      "step": 2632
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1348305878945426,
      "learning_rate": 1.027412289028879e-05,
      "loss": 0.9597,
      "step": 2633
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0966123739035116,
      "learning_rate": 1.0267894332754243e-05,
      "loss": 0.9683,
      "step": 2634
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8679237034335694,
      "learning_rate": 1.0261665671213891e-05,
      "loss": 0.8977,
      "step": 2635
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.938342839473002,
      "learning_rate": 1.0255436908085919e-05,
      "loss": 0.8641,
      "step": 2636
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0729407756306912,
      "learning_rate": 1.024920804578854e-05,
      "loss": 0.9611,
      "step": 2637
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0533555326792763,
      "learning_rate": 1.0242979086740019e-05,
      "loss": 0.9438,
      "step": 2638
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9375472337526344,
      "learning_rate": 1.023675003335865e-05,
      "loss": 0.8503,
      "step": 2639
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0641189142395333,
      "learning_rate": 1.0230520888062765e-05,
      "loss": 0.9611,
      "step": 2640
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.07385935439193,
      "learning_rate": 1.0224291653270739e-05,
      "loss": 0.9156,
      "step": 2641
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2774947112598818,
      "learning_rate": 1.0218062331400969e-05,
      "loss": 0.9558,
      "step": 2642
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0421602000627894,
      "learning_rate": 1.0211832924871889e-05,
      "loss": 0.9427,
      "step": 2643
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0576678318532715,
      "learning_rate": 1.0205603436101978e-05,
      "loss": 0.9527,
      "step": 2644
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.064115717328362,
      "learning_rate": 1.0199373867509734e-05,
      "loss": 0.952,
      "step": 2645
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9438842810981111,
      "learning_rate": 1.019314422151369e-05,
      "loss": 0.9488,
      "step": 2646
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8567152462433155,
      "learning_rate": 1.0186914500532408e-05,
      "loss": 0.6515,
      "step": 2647
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9832753011554455,
      "learning_rate": 1.0180684706984483e-05,
      "loss": 0.9609,
      "step": 2648
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0494478456470113,
      "learning_rate": 1.0174454843288533e-05,
      "loss": 0.9631,
      "step": 2649
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9531214427686192,
      "learning_rate": 1.0168224911863205e-05,
      "loss": 0.9151,
      "step": 2650
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9539683649459976,
      "learning_rate": 1.0161994915127173e-05,
      "loss": 0.9142,
      "step": 2651
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9184886740851799,
      "learning_rate": 1.015576485549914e-05,
      "loss": 0.9317,
      "step": 2652
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8355692854384258,
      "learning_rate": 1.0149534735397823e-05,
      "loss": 0.6771,
      "step": 2653
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1050973062192269,
      "learning_rate": 1.0143304557241979e-05,
      "loss": 0.9449,
      "step": 2654
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.006857479810348,
      "learning_rate": 1.0137074323450372e-05,
      "loss": 0.9565,
      "step": 2655
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0302525332892856,
      "learning_rate": 1.0130844036441787e-05,
      "loss": 0.8788,
      "step": 2656
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2112319492164034,
      "learning_rate": 1.0124613698635043e-05,
      "loss": 0.9759,
      "step": 2657
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0874990956361617,
      "learning_rate": 1.0118383312448973e-05,
      "loss": 0.9793,
      "step": 2658
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9256788463231342,
      "learning_rate": 1.0112152880302426e-05,
      "loss": 0.91,
      "step": 2659
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9737439878862562,
      "learning_rate": 1.0105922404614265e-05,
      "loss": 0.9813,
      "step": 2660
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9439087882228423,
      "learning_rate": 1.0099691887803385e-05,
      "loss": 0.8993,
      "step": 2661
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0481145287490679,
      "learning_rate": 1.0093461332288678e-05,
      "loss": 0.9169,
      "step": 2662
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.024579789326812,
      "learning_rate": 1.0087230740489065e-05,
      "loss": 0.9759,
      "step": 2663
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8671670850346195,
      "learning_rate": 1.0081000114823473e-05,
      "loss": 0.6173,
      "step": 2664
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8648250652550228,
      "learning_rate": 1.007476945771085e-05,
      "loss": 0.654,
      "step": 2665
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0734515487054894,
      "learning_rate": 1.006853877157015e-05,
      "loss": 0.9523,
      "step": 2666
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9811867404150767,
      "learning_rate": 1.0062308058820337e-05,
      "loss": 0.9084,
      "step": 2667
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9865234453072526,
      "learning_rate": 1.0056077321880393e-05,
      "loss": 0.8976,
      "step": 2668
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.120700579078923,
      "learning_rate": 1.0049846563169297e-05,
      "loss": 0.8993,
      "step": 2669
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8395721183406989,
      "learning_rate": 1.0043615785106051e-05,
      "loss": 0.6763,
      "step": 2670
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.097496590529086,
      "learning_rate": 1.0037384990109658e-05,
      "loss": 0.9019,
      "step": 2671
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9708374140903564,
      "learning_rate": 1.0031154180599123e-05,
      "loss": 0.9013,
      "step": 2672
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9679739036665322,
      "learning_rate": 1.0024923358993458e-05,
      "loss": 0.9337,
      "step": 2673
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0195456296904226,
      "learning_rate": 1.0018692527711695e-05,
      "loss": 0.8914,
      "step": 2674
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9563610304040743,
      "learning_rate": 1.0012461689172846e-05,
      "loss": 0.9157,
      "step": 2675
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.024730665041066,
      "learning_rate": 1.0006230845795937e-05,
      "loss": 0.9705,
      "step": 2676
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1224103158086893,
      "learning_rate": 1e-05,
      "loss": 0.9937,
      "step": 2677
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1389612011056847,
      "learning_rate": 9.993769154204063e-06,
      "loss": 0.9541,
      "step": 2678
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0979194810169102,
      "learning_rate": 9.987538310827159e-06,
      "loss": 0.9047,
      "step": 2679
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9415084664165405,
      "learning_rate": 9.981307472288308e-06,
      "loss": 0.895,
      "step": 2680
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.007163841925523,
      "learning_rate": 9.975076641006542e-06,
      "loss": 0.9138,
      "step": 2681
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9203377911530394,
      "learning_rate": 9.968845819400883e-06,
      "loss": 0.9133,
      "step": 2682
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0421408789162314,
      "learning_rate": 9.962615009890346e-06,
      "loss": 0.9595,
      "step": 2683
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0132241237530926,
      "learning_rate": 9.956384214893949e-06,
      "loss": 0.9218,
      "step": 2684
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.026445471103826,
      "learning_rate": 9.950153436830707e-06,
      "loss": 0.8824,
      "step": 2685
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9874609815005302,
      "learning_rate": 9.94392267811961e-06,
      "loss": 0.9373,
      "step": 2686
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9822651508675732,
      "learning_rate": 9.937691941179665e-06,
      "loss": 0.921,
      "step": 2687
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0131983638743922,
      "learning_rate": 9.931461228429856e-06,
      "loss": 0.9172,
      "step": 2688
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0449450861278045,
      "learning_rate": 9.925230542289151e-06,
      "loss": 0.9266,
      "step": 2689
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0119094475598769,
      "learning_rate": 9.91899988517653e-06,
      "loss": 0.9761,
      "step": 2690
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9919782778400698,
      "learning_rate": 9.912769259510938e-06,
      "loss": 0.9302,
      "step": 2691
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9820590012649535,
      "learning_rate": 9.906538667711324e-06,
      "loss": 0.9206,
      "step": 2692
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9887744408944767,
      "learning_rate": 9.90030811219662e-06,
      "loss": 0.9678,
      "step": 2693
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0541814849628777,
      "learning_rate": 9.894077595385736e-06,
      "loss": 0.9356,
      "step": 2694
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7969878446395873,
      "learning_rate": 9.887847119697577e-06,
      "loss": 0.6315,
      "step": 2695
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.080626307877143,
      "learning_rate": 9.881616687551032e-06,
      "loss": 1.0015,
      "step": 2696
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.044019986141342,
      "learning_rate": 9.875386301364958e-06,
      "loss": 0.9784,
      "step": 2697
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9483687136424249,
      "learning_rate": 9.869155963558215e-06,
      "loss": 0.9191,
      "step": 2698
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1016887414277385,
      "learning_rate": 9.862925676549635e-06,
      "loss": 0.9318,
      "step": 2699
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0884740331381844,
      "learning_rate": 9.856695442758023e-06,
      "loss": 0.943,
      "step": 2700
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9702551436921738,
      "learning_rate": 9.850465264602175e-06,
      "loss": 0.8793,
      "step": 2701
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.011157357599742,
      "learning_rate": 9.844235144500865e-06,
      "loss": 0.9002,
      "step": 2702
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0073963128852002,
      "learning_rate": 9.83800508487283e-06,
      "loss": 0.9494,
      "step": 2703
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9419354441761406,
      "learning_rate": 9.831775088136797e-06,
      "loss": 0.6785,
      "step": 2704
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.845049215172552,
      "learning_rate": 9.82554515671147e-06,
      "loss": 0.6746,
      "step": 2705
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.206142290308344,
      "learning_rate": 9.819315293015519e-06,
      "loss": 0.9664,
      "step": 2706
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.964161029721775,
      "learning_rate": 9.813085499467594e-06,
      "loss": 0.9256,
      "step": 2707
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0105071770153489,
      "learning_rate": 9.806855778486314e-06,
      "loss": 0.9793,
      "step": 2708
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0009594315714443,
      "learning_rate": 9.800626132490268e-06,
      "loss": 0.93,
      "step": 2709
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.044551657329871,
      "learning_rate": 9.794396563898022e-06,
      "loss": 0.9725,
      "step": 2710
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9891708409695217,
      "learning_rate": 9.788167075128113e-06,
      "loss": 0.9336,
      "step": 2711
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0842638890423608,
      "learning_rate": 9.781937668599035e-06,
      "loss": 0.9499,
      "step": 2712
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0336979510839361,
      "learning_rate": 9.775708346729263e-06,
      "loss": 0.9886,
      "step": 2713
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9885771998797919,
      "learning_rate": 9.769479111937238e-06,
      "loss": 0.9479,
      "step": 2714
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9511184816923123,
      "learning_rate": 9.763249966641352e-06,
      "loss": 0.918,
      "step": 2715
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.035218372595174,
      "learning_rate": 9.757020913259986e-06,
      "loss": 0.9675,
      "step": 2716
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.040215503026875,
      "learning_rate": 9.750791954211464e-06,
      "loss": 0.8735,
      "step": 2717
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9130268391440043,
      "learning_rate": 9.744563091914085e-06,
      "loss": 0.9285,
      "step": 2718
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.777194295556237,
      "learning_rate": 9.738334328786114e-06,
      "loss": 0.6561,
      "step": 2719
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2740887842477,
      "learning_rate": 9.732105667245759e-06,
      "loss": 0.9075,
      "step": 2720
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8508483823178317,
      "learning_rate": 9.725877109711212e-06,
      "loss": 0.8949,
      "step": 2721
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0341197344086421,
      "learning_rate": 9.719648658600611e-06,
      "loss": 0.9195,
      "step": 2722
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.032972962900896,
      "learning_rate": 9.71342031633205e-06,
      "loss": 0.964,
      "step": 2723
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1241825858596102,
      "learning_rate": 9.70719208532359e-06,
      "loss": 0.9595,
      "step": 2724
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1948680631904574,
      "learning_rate": 9.700963967993246e-06,
      "loss": 0.9603,
      "step": 2725
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9180058876347497,
      "learning_rate": 9.694735966758982e-06,
      "loss": 0.9705,
      "step": 2726
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.019270029667276,
      "learning_rate": 9.688508084038729e-06,
      "loss": 0.9584,
      "step": 2727
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9010364081420753,
      "learning_rate": 9.682280322250365e-06,
      "loss": 0.8695,
      "step": 2728
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1117334649029638,
      "learning_rate": 9.676052683811715e-06,
      "loss": 0.9659,
      "step": 2729
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9467811889334272,
      "learning_rate": 9.669825171140568e-06,
      "loss": 0.9258,
      "step": 2730
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6309800945953122,
      "learning_rate": 9.66359778665466e-06,
      "loss": 0.3275,
      "step": 2731
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9650718956395554,
      "learning_rate": 9.657370532771672e-06,
      "loss": 0.9372,
      "step": 2732
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8973399799149576,
      "learning_rate": 9.651143411909241e-06,
      "loss": 0.7014,
      "step": 2733
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9826421767950607,
      "learning_rate": 9.64491642648495e-06,
      "loss": 0.9417,
      "step": 2734
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.078924153274007,
      "learning_rate": 9.638689578916326e-06,
      "loss": 0.997,
      "step": 2735
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9075763756066588,
      "learning_rate": 9.632462871620847e-06,
      "loss": 0.9522,
      "step": 2736
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.945319968038411,
      "learning_rate": 9.62623630701594e-06,
      "loss": 0.8899,
      "step": 2737
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8538259350227377,
      "learning_rate": 9.620009887518963e-06,
      "loss": 0.6279,
      "step": 2738
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7896088276264198,
      "learning_rate": 9.613783615547233e-06,
      "loss": 0.6484,
      "step": 2739
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.996780876804824,
      "learning_rate": 9.607557493518006e-06,
      "loss": 0.8986,
      "step": 2740
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1815221295567222,
      "learning_rate": 9.601331523848464e-06,
      "loss": 0.9586,
      "step": 2741
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.962065887571875,
      "learning_rate": 9.595105708955758e-06,
      "loss": 0.9107,
      "step": 2742
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0790258001253243,
      "learning_rate": 9.588880051256951e-06,
      "loss": 0.941,
      "step": 2743
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9267978688447919,
      "learning_rate": 9.582654553169064e-06,
      "loss": 0.9273,
      "step": 2744
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0949960444142963,
      "learning_rate": 9.576429217109054e-06,
      "loss": 0.9491,
      "step": 2745
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1138475650809456,
      "learning_rate": 9.5702040454938e-06,
      "loss": 0.9264,
      "step": 2746
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0087753996150006,
      "learning_rate": 9.563979040740138e-06,
      "loss": 0.9419,
      "step": 2747
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1045588649030817,
      "learning_rate": 9.557754205264826e-06,
      "loss": 0.9502,
      "step": 2748
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9158888011749534,
      "learning_rate": 9.551529541484554e-06,
      "loss": 0.7013,
      "step": 2749
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1530015630732322,
      "learning_rate": 9.545305051815957e-06,
      "loss": 0.9265,
      "step": 2750
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0383710256665137,
      "learning_rate": 9.539080738675597e-06,
      "loss": 0.9679,
      "step": 2751
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9300447904912774,
      "learning_rate": 9.53285660447996e-06,
      "loss": 0.9492,
      "step": 2752
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1227930922487186,
      "learning_rate": 9.526632651645476e-06,
      "loss": 0.9737,
      "step": 2753
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9956005704580597,
      "learning_rate": 9.520408882588497e-06,
      "loss": 0.94,
      "step": 2754
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9714708537540113,
      "learning_rate": 9.514185299725299e-06,
      "loss": 0.954,
      "step": 2755
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6550804115396537,
      "learning_rate": 9.507961905472093e-06,
      "loss": 0.3313,
      "step": 2756
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8618518395204513,
      "learning_rate": 9.501738702245023e-06,
      "loss": 0.8481,
      "step": 2757
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9360902846633095,
      "learning_rate": 9.495515692460138e-06,
      "loss": 0.6542,
      "step": 2758
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0692768363645544,
      "learning_rate": 9.489292878533436e-06,
      "loss": 0.9234,
      "step": 2759
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9687585424135029,
      "learning_rate": 9.483070262880823e-06,
      "loss": 0.9283,
      "step": 2760
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.991852244906415,
      "learning_rate": 9.476847847918126e-06,
      "loss": 0.8985,
      "step": 2761
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.088368910379489,
      "learning_rate": 9.47062563606111e-06,
      "loss": 0.983,
      "step": 2762
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.107985176439882,
      "learning_rate": 9.464403629725454e-06,
      "loss": 0.9754,
      "step": 2763
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8312224385233907,
      "learning_rate": 9.458181831326744e-06,
      "loss": 0.6904,
      "step": 2764
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9335491786276929,
      "learning_rate": 9.451960243280506e-06,
      "loss": 0.9025,
      "step": 2765
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2585221763057646,
      "learning_rate": 9.44573886800217e-06,
      "loss": 0.9631,
      "step": 2766
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0592826787957217,
      "learning_rate": 9.43951770790709e-06,
      "loss": 0.9485,
      "step": 2767
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.826368711532025,
      "learning_rate": 9.433296765410534e-06,
      "loss": 0.6645,
      "step": 2768
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9323183711907344,
      "learning_rate": 9.427076042927683e-06,
      "loss": 0.9484,
      "step": 2769
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.854322800332307,
      "learning_rate": 9.420855542873638e-06,
      "loss": 0.6035,
      "step": 2770
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1005970472842994,
      "learning_rate": 9.414635267663416e-06,
      "loss": 0.9477,
      "step": 2771
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.012163121370086,
      "learning_rate": 9.408415219711934e-06,
      "loss": 0.9129,
      "step": 2772
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0297725004302067,
      "learning_rate": 9.402195401434036e-06,
      "loss": 0.936,
      "step": 2773
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1282332556797963,
      "learning_rate": 9.395975815244468e-06,
      "loss": 0.992,
      "step": 2774
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0277368418771953,
      "learning_rate": 9.389756463557878e-06,
      "loss": 0.8997,
      "step": 2775
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.132929784051341,
      "learning_rate": 9.383537348788844e-06,
      "loss": 1.0076,
      "step": 2776
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9420392697463983,
      "learning_rate": 9.377318473351838e-06,
      "loss": 0.9438,
      "step": 2777
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.040470768425601,
      "learning_rate": 9.371099839661238e-06,
      "loss": 0.9375,
      "step": 2778
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0998023536713595,
      "learning_rate": 9.364881450131335e-06,
      "loss": 0.984,
      "step": 2779
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8056538777719106,
      "learning_rate": 9.358663307176323e-06,
      "loss": 0.8956,
      "step": 2780
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2643968814323783,
      "learning_rate": 9.352445413210294e-06,
      "loss": 0.9004,
      "step": 2781
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9658137126943969,
      "learning_rate": 9.346227770647251e-06,
      "loss": 0.9168,
      "step": 2782
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9232258392197525,
      "learning_rate": 9.3400103819011e-06,
      "loss": 0.93,
      "step": 2783
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1845957416421502,
      "learning_rate": 9.33379324938564e-06,
      "loss": 0.8602,
      "step": 2784
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2457833912583545,
      "learning_rate": 9.327576375514582e-06,
      "loss": 0.9442,
      "step": 2785
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0461113625746694,
      "learning_rate": 9.321359762701527e-06,
      "loss": 0.9073,
      "step": 2786
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1930994617620927,
      "learning_rate": 9.315143413359975e-06,
      "loss": 1.0055,
      "step": 2787
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0067617391816777,
      "learning_rate": 9.308927329903333e-06,
      "loss": 0.9399,
      "step": 2788
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9707113228227243,
      "learning_rate": 9.302711514744897e-06,
      "loss": 0.881,
      "step": 2789
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2230821593594399,
      "learning_rate": 9.296495970297855e-06,
      "loss": 1.0267,
      "step": 2790
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0228053497074867,
      "learning_rate": 9.290280698975307e-06,
      "loss": 0.9471,
      "step": 2791
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8250702682816274,
      "learning_rate": 9.284065703190221e-06,
      "loss": 0.6362,
      "step": 2792
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9414343988001175,
      "learning_rate": 9.27785098535548e-06,
      "loss": 0.9011,
      "step": 2793
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8139616137224812,
      "learning_rate": 9.271636547883856e-06,
      "loss": 0.6519,
      "step": 2794
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0025797194332444,
      "learning_rate": 9.265422393187998e-06,
      "loss": 0.8479,
      "step": 2795
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0420059285424241,
      "learning_rate": 9.259208523680457e-06,
      "loss": 0.9775,
      "step": 2796
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0156007583870381,
      "learning_rate": 9.252994941773679e-06,
      "loss": 0.9607,
      "step": 2797
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8703645889360249,
      "learning_rate": 9.24678164987998e-06,
      "loss": 0.6008,
      "step": 2798
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9456732236319642,
      "learning_rate": 9.24056865041158e-06,
      "loss": 0.9435,
      "step": 2799
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9599529709393246,
      "learning_rate": 9.234355945780581e-06,
      "loss": 0.9198,
      "step": 2800
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1797938371962102,
      "learning_rate": 9.228143538398963e-06,
      "loss": 1.0125,
      "step": 2801
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.876186744593102,
      "learning_rate": 9.221931430678598e-06,
      "loss": 0.943,
      "step": 2802
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.01042572102738,
      "learning_rate": 9.215719625031245e-06,
      "loss": 0.9058,
      "step": 2803
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8231118343365916,
      "learning_rate": 9.209508123868534e-06,
      "loss": 0.6265,
      "step": 2804
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0109665742531961,
      "learning_rate": 9.203296929601986e-06,
      "loss": 0.929,
      "step": 2805
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9346350604581256,
      "learning_rate": 9.197086044643004e-06,
      "loss": 0.9191,
      "step": 2806
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0490912601508244,
      "learning_rate": 9.190875471402865e-06,
      "loss": 0.9028,
      "step": 2807
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1012158240423038,
      "learning_rate": 9.184665212292723e-06,
      "loss": 0.8965,
      "step": 2808
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0554937932010364,
      "learning_rate": 9.178455269723623e-06,
      "loss": 0.8966,
      "step": 2809
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9641131193115875,
      "learning_rate": 9.172245646106471e-06,
      "loss": 0.9382,
      "step": 2810
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1497078804029826,
      "learning_rate": 9.166036343852061e-06,
      "loss": 0.926,
      "step": 2811
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9442903911796486,
      "learning_rate": 9.159827365371055e-06,
      "loss": 0.9163,
      "step": 2812
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7524912676545145,
      "learning_rate": 9.153618713073995e-06,
      "loss": 0.619,
      "step": 2813
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4317390104291576,
      "learning_rate": 9.14741038937129e-06,
      "loss": 0.9381,
      "step": 2814
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0064663207422744,
      "learning_rate": 9.141202396673232e-06,
      "loss": 0.9008,
      "step": 2815
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1455233224077312,
      "learning_rate": 9.13499473738997e-06,
      "loss": 0.9665,
      "step": 2816
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1353631535129873,
      "learning_rate": 9.128787413931536e-06,
      "loss": 0.928,
      "step": 2817
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1190647402012972,
      "learning_rate": 9.122580428707822e-06,
      "loss": 0.9023,
      "step": 2818
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0490440198839037,
      "learning_rate": 9.116373784128597e-06,
      "loss": 0.9721,
      "step": 2819
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9939068102880008,
      "learning_rate": 9.110167482603494e-06,
      "loss": 0.9518,
      "step": 2820
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9512791668558882,
      "learning_rate": 9.10396152654201e-06,
      "loss": 0.859,
      "step": 2821
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0199468661201034,
      "learning_rate": 9.097755918353513e-06,
      "loss": 0.9874,
      "step": 2822
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9196679859338391,
      "learning_rate": 9.091550660447236e-06,
      "loss": 0.5872,
      "step": 2823
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.151066500386334,
      "learning_rate": 9.08534575523227e-06,
      "loss": 0.9665,
      "step": 2824
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4504278169995757,
      "learning_rate": 9.079141205117573e-06,
      "loss": 0.9562,
      "step": 2825
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9596918787471271,
      "learning_rate": 9.072937012511968e-06,
      "loss": 0.9355,
      "step": 2826
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7668499185907166,
      "learning_rate": 9.066733179824134e-06,
      "loss": 0.65,
      "step": 2827
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1715382590864676,
      "learning_rate": 9.060529709462613e-06,
      "loss": 0.9182,
      "step": 2828
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9765531096006955,
      "learning_rate": 9.054326603835807e-06,
      "loss": 0.9236,
      "step": 2829
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.072544805256972,
      "learning_rate": 9.048123865351971e-06,
      "loss": 0.9605,
      "step": 2830
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1682555805725203,
      "learning_rate": 9.041921496419225e-06,
      "loss": 0.9596,
      "step": 2831
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1263795588055225,
      "learning_rate": 9.035719499445545e-06,
      "loss": 0.9422,
      "step": 2832
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1122386737371475,
      "learning_rate": 9.029517876838755e-06,
      "loss": 0.9323,
      "step": 2833
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0761154083402498,
      "learning_rate": 9.023316631006536e-06,
      "loss": 0.9025,
      "step": 2834
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0455638101309055,
      "learning_rate": 9.017115764356436e-06,
      "loss": 0.9804,
      "step": 2835
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7511577553483352,
      "learning_rate": 9.010915279295833e-06,
      "loss": 0.6152,
      "step": 2836
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0081257862241277,
      "learning_rate": 9.004715178231975e-06,
      "loss": 0.6186,
      "step": 2837
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9063819959206205,
      "learning_rate": 8.998515463571953e-06,
      "loss": 0.916,
      "step": 2838
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1703889538129992,
      "learning_rate": 8.992316137722711e-06,
      "loss": 1.0049,
      "step": 2839
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0249510487028501,
      "learning_rate": 8.986117203091042e-06,
      "loss": 0.9352,
      "step": 2840
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9676673571912595,
      "learning_rate": 8.97991866208358e-06,
      "loss": 0.9151,
      "step": 2841
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9067534574862496,
      "learning_rate": 8.973720517106814e-06,
      "loss": 0.8834,
      "step": 2842
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.094143909808958,
      "learning_rate": 8.967522770567086e-06,
      "loss": 0.9098,
      "step": 2843
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9985412613791669,
      "learning_rate": 8.961325424870561e-06,
      "loss": 0.9182,
      "step": 2844
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9874683632622933,
      "learning_rate": 8.955128482423271e-06,
      "loss": 0.9156,
      "step": 2845
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8306192572439479,
      "learning_rate": 8.948931945631082e-06,
      "loss": 0.8511,
      "step": 2846
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0369675219592676,
      "learning_rate": 8.9427358168997e-06,
      "loss": 0.973,
      "step": 2847
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8959828960475748,
      "learning_rate": 8.936540098634675e-06,
      "loss": 0.9206,
      "step": 2848
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8939195160449428,
      "learning_rate": 8.930344793241404e-06,
      "loss": 0.9058,
      "step": 2849
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0580371299691822,
      "learning_rate": 8.924149903125108e-06,
      "loss": 1.0147,
      "step": 2850
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0465705338985862,
      "learning_rate": 8.917955430690865e-06,
      "loss": 0.9369,
      "step": 2851
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0207037208364143,
      "learning_rate": 8.91176137834358e-06,
      "loss": 0.9314,
      "step": 2852
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9943647239795166,
      "learning_rate": 8.905567748487997e-06,
      "loss": 0.9705,
      "step": 2853
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9841627379837427,
      "learning_rate": 8.899374543528695e-06,
      "loss": 0.9537,
      "step": 2854
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0152758759371014,
      "learning_rate": 8.893181765870094e-06,
      "loss": 0.9583,
      "step": 2855
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0493155641849738,
      "learning_rate": 8.886989417916435e-06,
      "loss": 0.9798,
      "step": 2856
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0886666730183978,
      "learning_rate": 8.88079750207181e-06,
      "loss": 0.9413,
      "step": 2857
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9272001850470533,
      "learning_rate": 8.87460602074013e-06,
      "loss": 0.9176,
      "step": 2858
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0335597573626392,
      "learning_rate": 8.86841497632514e-06,
      "loss": 0.9503,
      "step": 2859
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8316484823668765,
      "learning_rate": 8.862224371230418e-06,
      "loss": 0.9338,
      "step": 2860
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0030566014960614,
      "learning_rate": 8.85603420785937e-06,
      "loss": 0.9532,
      "step": 2861
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0652681148649685,
      "learning_rate": 8.84984448861523e-06,
      "loss": 0.9903,
      "step": 2862
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9926845793608593,
      "learning_rate": 8.84365521590106e-06,
      "loss": 0.9305,
      "step": 2863
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8046232473891722,
      "learning_rate": 8.837466392119752e-06,
      "loss": 0.6384,
      "step": 2864
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.919948677211136,
      "learning_rate": 8.831278019674017e-06,
      "loss": 0.6511,
      "step": 2865
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0080740715343552,
      "learning_rate": 8.825090100966396e-06,
      "loss": 0.9339,
      "step": 2866
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9814361903967078,
      "learning_rate": 8.818902638399247e-06,
      "loss": 0.923,
      "step": 2867
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2514294892078848,
      "learning_rate": 8.81271563437476e-06,
      "loss": 0.8553,
      "step": 2868
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0414113590668903,
      "learning_rate": 8.806529091294948e-06,
      "loss": 0.952,
      "step": 2869
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.175783324279413,
      "learning_rate": 8.800343011561633e-06,
      "loss": 0.9197,
      "step": 2870
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4905760337176477,
      "learning_rate": 8.794157397576464e-06,
      "loss": 0.9672,
      "step": 2871
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9818965570786211,
      "learning_rate": 8.787972251740916e-06,
      "loss": 0.8893,
      "step": 2872
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0821505721348323,
      "learning_rate": 8.781787576456269e-06,
      "loss": 0.9013,
      "step": 2873
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0895432335417004,
      "learning_rate": 8.775603374123627e-06,
      "loss": 0.9423,
      "step": 2874
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1351113534456347,
      "learning_rate": 8.769419647143917e-06,
      "loss": 0.9426,
      "step": 2875
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9116104378968837,
      "learning_rate": 8.763236397917865e-06,
      "loss": 0.6464,
      "step": 2876
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0401804642077925,
      "learning_rate": 8.757053628846028e-06,
      "loss": 0.9495,
      "step": 2877
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9840426047008564,
      "learning_rate": 8.75087134232877e-06,
      "loss": 0.6694,
      "step": 2878
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0373375706918488,
      "learning_rate": 8.744689540766265e-06,
      "loss": 0.9901,
      "step": 2879
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1153529852760957,
      "learning_rate": 8.738508226558499e-06,
      "loss": 0.9704,
      "step": 2880
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.933029310680692,
      "learning_rate": 8.73232740210528e-06,
      "loss": 0.9017,
      "step": 2881
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9986808641399271,
      "learning_rate": 8.726147069806206e-06,
      "loss": 0.9722,
      "step": 2882
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0280302080937282,
      "learning_rate": 8.719967232060698e-06,
      "loss": 0.8994,
      "step": 2883
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.030556518073835,
      "learning_rate": 8.713787891267988e-06,
      "loss": 0.9223,
      "step": 2884
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7553151707331733,
      "learning_rate": 8.707609049827102e-06,
      "loss": 0.6416,
      "step": 2885
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0156953423926782,
      "learning_rate": 8.70143071013688e-06,
      "loss": 0.8787,
      "step": 2886
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0791846129016125,
      "learning_rate": 8.695252874595972e-06,
      "loss": 0.9941,
      "step": 2887
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0156364389839083,
      "learning_rate": 8.689075545602816e-06,
      "loss": 0.9905,
      "step": 2888
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1247644736686897,
      "learning_rate": 8.68289872555567e-06,
      "loss": 0.9231,
      "step": 2889
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9239393346508287,
      "learning_rate": 8.676722416852594e-06,
      "loss": 1.0058,
      "step": 2890
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2202880668727902,
      "learning_rate": 8.670546621891434e-06,
      "loss": 0.9657,
      "step": 2891
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.978347118034826,
      "learning_rate": 8.66437134306985e-06,
      "loss": 0.8967,
      "step": 2892
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9576325180906639,
      "learning_rate": 8.658196582785297e-06,
      "loss": 0.9387,
      "step": 2893
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9526581896225196,
      "learning_rate": 8.652022343435027e-06,
      "loss": 0.8928,
      "step": 2894
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9272881866681717,
      "learning_rate": 8.645848627416102e-06,
      "loss": 0.889,
      "step": 2895
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1479100342767798,
      "learning_rate": 8.63967543712536e-06,
      "loss": 0.9073,
      "step": 2896
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9510145349759942,
      "learning_rate": 8.633502774959453e-06,
      "loss": 0.9304,
      "step": 2897
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0005354803828919,
      "learning_rate": 8.627330643314818e-06,
      "loss": 0.8913,
      "step": 2898
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.980194689672904,
      "learning_rate": 8.62115904458769e-06,
      "loss": 0.9242,
      "step": 2899
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0752956664471005,
      "learning_rate": 8.614987981174093e-06,
      "loss": 0.9531,
      "step": 2900
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9499241725326928,
      "learning_rate": 8.608817455469854e-06,
      "loss": 0.952,
      "step": 2901
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.988546291393217,
      "learning_rate": 8.602647469870573e-06,
      "loss": 0.9182,
      "step": 2902
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0940688649788222,
      "learning_rate": 8.596478026771658e-06,
      "loss": 0.9417,
      "step": 2903
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0461589113115535,
      "learning_rate": 8.590309128568303e-06,
      "loss": 0.8959,
      "step": 2904
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9667497530685683,
      "learning_rate": 8.584140777655476e-06,
      "loss": 0.8614,
      "step": 2905
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0629909095460064,
      "learning_rate": 8.57797297642795e-06,
      "loss": 0.92,
      "step": 2906
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9318606456226535,
      "learning_rate": 8.571805727280278e-06,
      "loss": 0.943,
      "step": 2907
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.235698347834913,
      "learning_rate": 8.565639032606794e-06,
      "loss": 0.5988,
      "step": 2908
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8533229982873424,
      "learning_rate": 8.559472894801623e-06,
      "loss": 0.6461,
      "step": 2909
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0146118012653205,
      "learning_rate": 8.553307316258678e-06,
      "loss": 0.9241,
      "step": 2910
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.008777480437455,
      "learning_rate": 8.547142299371642e-06,
      "loss": 0.971,
      "step": 2911
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.144822477249725,
      "learning_rate": 8.540977846533986e-06,
      "loss": 0.991,
      "step": 2912
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9963277633047602,
      "learning_rate": 8.534813960138968e-06,
      "loss": 0.9431,
      "step": 2913
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7896506854366638,
      "learning_rate": 8.528650642579618e-06,
      "loss": 0.5807,
      "step": 2914
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.029861723525488,
      "learning_rate": 8.52248789624875e-06,
      "loss": 0.9203,
      "step": 2915
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.083887826425956,
      "learning_rate": 8.516325723538949e-06,
      "loss": 0.9776,
      "step": 2916
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9310821460234288,
      "learning_rate": 8.510164126842591e-06,
      "loss": 0.8624,
      "step": 2917
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0877524934348566,
      "learning_rate": 8.504003108551814e-06,
      "loss": 0.9982,
      "step": 2918
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0228843584187743,
      "learning_rate": 8.497842671058539e-06,
      "loss": 0.932,
      "step": 2919
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8794319600522968,
      "learning_rate": 8.491682816754456e-06,
      "loss": 0.8959,
      "step": 2920
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0341867396228601,
      "learning_rate": 8.485523548031044e-06,
      "loss": 0.898,
      "step": 2921
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9947066565777704,
      "learning_rate": 8.479364867279529e-06,
      "loss": 0.8685,
      "step": 2922
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0831866338147207,
      "learning_rate": 8.47320677689093e-06,
      "loss": 0.937,
      "step": 2923
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.93289361584377,
      "learning_rate": 8.467049279256034e-06,
      "loss": 0.8952,
      "step": 2924
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0198858088085125,
      "learning_rate": 8.460892376765387e-06,
      "loss": 0.9348,
      "step": 2925
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2223019776008417,
      "learning_rate": 8.45473607180931e-06,
      "loss": 0.9154,
      "step": 2926
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0794606133611455,
      "learning_rate": 8.448580366777898e-06,
      "loss": 0.9862,
      "step": 2927
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2944239941963753,
      "learning_rate": 8.442425264061e-06,
      "loss": 0.8991,
      "step": 2928
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0597753738363738,
      "learning_rate": 8.436270766048245e-06,
      "loss": 0.9722,
      "step": 2929
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0309205063943185,
      "learning_rate": 8.430116875129023e-06,
      "loss": 0.8885,
      "step": 2930
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8529391159882163,
      "learning_rate": 8.42396359369248e-06,
      "loss": 0.6515,
      "step": 2931
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0099348889268802,
      "learning_rate": 8.417810924127533e-06,
      "loss": 0.8928,
      "step": 2932
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0540136743374062,
      "learning_rate": 8.411658868822866e-06,
      "loss": 0.9404,
      "step": 2933
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9716485823631876,
      "learning_rate": 8.40550743016691e-06,
      "loss": 0.9544,
      "step": 2934
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9440385796330567,
      "learning_rate": 8.39935661054787e-06,
      "loss": 0.899,
      "step": 2935
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0559202630120577,
      "learning_rate": 8.393206412353709e-06,
      "loss": 0.9193,
      "step": 2936
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8624663040856401,
      "learning_rate": 8.38705683797214e-06,
      "loss": 0.6508,
      "step": 2937
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6716617367745346,
      "learning_rate": 8.38090788979064e-06,
      "loss": 0.3657,
      "step": 2938
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1823264614382338,
      "learning_rate": 8.374759570196448e-06,
      "loss": 0.9922,
      "step": 2939
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9904343211079352,
      "learning_rate": 8.368611881576547e-06,
      "loss": 0.912,
      "step": 2940
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2369662430155053,
      "learning_rate": 8.362464826317687e-06,
      "loss": 0.9283,
      "step": 2941
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0914725282928026,
      "learning_rate": 8.35631840680636e-06,
      "loss": 0.9481,
      "step": 2942
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8526392693607602,
      "learning_rate": 8.35017262542882e-06,
      "loss": 0.6383,
      "step": 2943
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0923354990326568,
      "learning_rate": 8.344027484571075e-06,
      "loss": 0.9424,
      "step": 2944
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4337485347440369,
      "learning_rate": 8.337882986618877e-06,
      "loss": 0.944,
      "step": 2945
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0858699604770405,
      "learning_rate": 8.331739133957729e-06,
      "loss": 0.9252,
      "step": 2946
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8958904819975253,
      "learning_rate": 8.325595928972894e-06,
      "loss": 0.9084,
      "step": 2947
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9607297134306951,
      "learning_rate": 8.319453374049367e-06,
      "loss": 0.9051,
      "step": 2948
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.061477211583545,
      "learning_rate": 8.313311471571903e-06,
      "loss": 0.9767,
      "step": 2949
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0550468388988723,
      "learning_rate": 8.307170223925003e-06,
      "loss": 0.9489,
      "step": 2950
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.973776977630793,
      "learning_rate": 8.301029633492907e-06,
      "loss": 0.908,
      "step": 2951
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0029416130010642,
      "learning_rate": 8.294889702659602e-06,
      "loss": 0.9462,
      "step": 2952
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0633798535348284,
      "learning_rate": 8.288750433808828e-06,
      "loss": 0.9482,
      "step": 2953
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8971270538393837,
      "learning_rate": 8.282611829324049e-06,
      "loss": 0.9371,
      "step": 2954
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9884628863557815,
      "learning_rate": 8.276473891588492e-06,
      "loss": 0.9196,
      "step": 2955
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8929052961032686,
      "learning_rate": 8.270336622985116e-06,
      "loss": 0.8942,
      "step": 2956
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.114610898771252,
      "learning_rate": 8.264200025896616e-06,
      "loss": 1.0077,
      "step": 2957
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.12862549195744,
      "learning_rate": 8.258064102705428e-06,
      "loss": 0.9828,
      "step": 2958
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1281374581027521,
      "learning_rate": 8.251928855793736e-06,
      "loss": 0.9345,
      "step": 2959
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1222283701882276,
      "learning_rate": 8.245794287543447e-06,
      "loss": 0.8991,
      "step": 2960
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9808697732896846,
      "learning_rate": 8.239660400336213e-06,
      "loss": 0.9578,
      "step": 2961
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1776804710027953,
      "learning_rate": 8.233527196553428e-06,
      "loss": 0.9804,
      "step": 2962
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0027475199740583,
      "learning_rate": 8.227394678576204e-06,
      "loss": 0.9166,
      "step": 2963
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8684613155566873,
      "learning_rate": 8.221262848785395e-06,
      "loss": 0.6236,
      "step": 2964
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9803970771881675,
      "learning_rate": 8.215131709561597e-06,
      "loss": 0.9073,
      "step": 2965
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0524017789689202,
      "learning_rate": 8.20900126328512e-06,
      "loss": 0.9363,
      "step": 2966
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1200256561772384,
      "learning_rate": 8.202871512336023e-06,
      "loss": 0.9739,
      "step": 2967
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3280588673507328,
      "learning_rate": 8.196742459094079e-06,
      "loss": 0.9318,
      "step": 2968
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9806989695401639,
      "learning_rate": 8.190614105938796e-06,
      "loss": 0.9058,
      "step": 2969
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9753944393508464,
      "learning_rate": 8.184486455249424e-06,
      "loss": 0.9154,
      "step": 2970
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8560928461659599,
      "learning_rate": 8.178359509404916e-06,
      "loss": 0.63,
      "step": 2971
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.006880556860581,
      "learning_rate": 8.172233270783966e-06,
      "loss": 0.8995,
      "step": 2972
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.036814396808571,
      "learning_rate": 8.166107741764997e-06,
      "loss": 0.9006,
      "step": 2973
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9446352449340697,
      "learning_rate": 8.15998292472614e-06,
      "loss": 0.9379,
      "step": 2974
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9587943716527753,
      "learning_rate": 8.153858822045267e-06,
      "loss": 0.9709,
      "step": 2975
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1640617142385032,
      "learning_rate": 8.147735436099967e-06,
      "loss": 0.9813,
      "step": 2976
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7721957929904302,
      "learning_rate": 8.141612769267543e-06,
      "loss": 0.6505,
      "step": 2977
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9858895759735187,
      "learning_rate": 8.135490823925027e-06,
      "loss": 0.9864,
      "step": 2978
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.081604804688857,
      "learning_rate": 8.129369602449176e-06,
      "loss": 0.9236,
      "step": 2979
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0571660299588357,
      "learning_rate": 8.123249107216446e-06,
      "loss": 0.9683,
      "step": 2980
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9771044190728635,
      "learning_rate": 8.117129340603032e-06,
      "loss": 0.9271,
      "step": 2981
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.010860781461526,
      "learning_rate": 8.111010304984841e-06,
      "loss": 0.9155,
      "step": 2982
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9941092723453553,
      "learning_rate": 8.104892002737488e-06,
      "loss": 0.9093,
      "step": 2983
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0494469430086524,
      "learning_rate": 8.098774436236308e-06,
      "loss": 0.9276,
      "step": 2984
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0400849998206152,
      "learning_rate": 8.092657607856356e-06,
      "loss": 0.941,
      "step": 2985
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1389824239703135,
      "learning_rate": 8.086541519972388e-06,
      "loss": 0.9984,
      "step": 2986
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.102813486355259,
      "learning_rate": 8.080426174958886e-06,
      "loss": 0.8991,
      "step": 2987
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0179809207382653,
      "learning_rate": 8.074311575190039e-06,
      "loss": 0.9427,
      "step": 2988
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0619966170454007,
      "learning_rate": 8.068197723039738e-06,
      "loss": 0.896,
      "step": 2989
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0102630808871371,
      "learning_rate": 8.062084620881598e-06,
      "loss": 0.9117,
      "step": 2990
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9742691979122123,
      "learning_rate": 8.055972271088933e-06,
      "loss": 0.8872,
      "step": 2991
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.072664682552424,
      "learning_rate": 8.049860676034762e-06,
      "loss": 0.9159,
      "step": 2992
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1295489330605442,
      "learning_rate": 8.043749838091828e-06,
      "loss": 0.963,
      "step": 2993
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9798522320291135,
      "learning_rate": 8.037639759632558e-06,
      "loss": 0.9701,
      "step": 2994
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0496168806952664,
      "learning_rate": 8.031530443029099e-06,
      "loss": 0.9065,
      "step": 2995
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.050829069120161,
      "learning_rate": 8.025421890653303e-06,
      "loss": 0.9388,
      "step": 2996
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1824469825806079,
      "learning_rate": 8.019314104876712e-06,
      "loss": 0.8789,
      "step": 2997
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9922026695007965,
      "learning_rate": 8.013207088070582e-06,
      "loss": 0.9925,
      "step": 2998
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8865884442782683,
      "learning_rate": 8.007100842605872e-06,
      "loss": 0.9282,
      "step": 2999
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1445680419522064,
      "learning_rate": 8.000995370853227e-06,
      "loss": 1.0001,
      "step": 3000
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0304645817515452,
      "learning_rate": 7.994890675183008e-06,
      "loss": 0.9184,
      "step": 3001
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1435031309160004,
      "learning_rate": 7.98878675796527e-06,
      "loss": 0.9661,
      "step": 3002
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0597841430500405,
      "learning_rate": 7.98268362156976e-06,
      "loss": 1.0321,
      "step": 3003
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7486661061149554,
      "learning_rate": 7.976581268365924e-06,
      "loss": 0.6268,
      "step": 3004
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0504621199275632,
      "learning_rate": 7.97047970072291e-06,
      "loss": 0.8861,
      "step": 3005
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9782385657540896,
      "learning_rate": 7.964378921009552e-06,
      "loss": 0.9842,
      "step": 3006
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.217151640644154,
      "learning_rate": 7.958278931594385e-06,
      "loss": 0.9225,
      "step": 3007
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0060590700877226,
      "learning_rate": 7.952179734845642e-06,
      "loss": 0.9153,
      "step": 3008
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.315187339859387,
      "learning_rate": 7.946081333131227e-06,
      "loss": 0.9287,
      "step": 3009
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9231059591656327,
      "learning_rate": 7.93998372881876e-06,
      "loss": 0.9795,
      "step": 3010
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7847966895368298,
      "learning_rate": 7.93388692427554e-06,
      "loss": 0.5951,
      "step": 3011
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1040234207594946,
      "learning_rate": 7.92779092186855e-06,
      "loss": 0.9358,
      "step": 3012
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9581729264686994,
      "learning_rate": 7.921695723964473e-06,
      "loss": 0.9192,
      "step": 3013
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9990094846718763,
      "learning_rate": 7.915601332929678e-06,
      "loss": 0.9891,
      "step": 3014
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8637559670266864,
      "learning_rate": 7.90950775113021e-06,
      "loss": 0.662,
      "step": 3015
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.991356324470182,
      "learning_rate": 7.903414980931813e-06,
      "loss": 0.9444,
      "step": 3016
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7714616311921796,
      "learning_rate": 7.897323024699907e-06,
      "loss": 0.6253,
      "step": 3017
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0936531786128176,
      "learning_rate": 7.8912318847996e-06,
      "loss": 0.9199,
      "step": 3018
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1645480613756707,
      "learning_rate": 7.885141563595685e-06,
      "loss": 0.9061,
      "step": 3019
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9714262891914535,
      "learning_rate": 7.879052063452626e-06,
      "loss": 0.8653,
      "step": 3020
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9953022394987621,
      "learning_rate": 7.872963386734584e-06,
      "loss": 0.9518,
      "step": 3021
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8101892501033535,
      "learning_rate": 7.866875535805394e-06,
      "loss": 0.6168,
      "step": 3022
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9882141423152212,
      "learning_rate": 7.860788513028566e-06,
      "loss": 0.9596,
      "step": 3023
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9126482248455503,
      "learning_rate": 7.85470232076729e-06,
      "loss": 0.8716,
      "step": 3024
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0043472092050751,
      "learning_rate": 7.848616961384442e-06,
      "loss": 0.9569,
      "step": 3025
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1620661066318931,
      "learning_rate": 7.842532437242559e-06,
      "loss": 0.8968,
      "step": 3026
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0267537754128953,
      "learning_rate": 7.83644875070387e-06,
      "loss": 0.9514,
      "step": 3027
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.117482801153417,
      "learning_rate": 7.83036590413027e-06,
      "loss": 0.9675,
      "step": 3028
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6432569532129567,
      "learning_rate": 7.824283899883327e-06,
      "loss": 0.3604,
      "step": 3029
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9231796042392658,
      "learning_rate": 7.818202740324287e-06,
      "loss": 0.9265,
      "step": 3030
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0430726330434403,
      "learning_rate": 7.812122427814068e-06,
      "loss": 0.9267,
      "step": 3031
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9880689285231574,
      "learning_rate": 7.806042964713248e-06,
      "loss": 0.8674,
      "step": 3032
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.008218627981941,
      "learning_rate": 7.79996435338209e-06,
      "loss": 0.878,
      "step": 3033
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9523654008306152,
      "learning_rate": 7.793886596180521e-06,
      "loss": 0.9403,
      "step": 3034
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9790813850604336,
      "learning_rate": 7.787809695468134e-06,
      "loss": 0.8746,
      "step": 3035
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9453190924477649,
      "learning_rate": 7.78173365360419e-06,
      "loss": 0.646,
      "step": 3036
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.95677906771797,
      "learning_rate": 7.775658472947623e-06,
      "loss": 0.9226,
      "step": 3037
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0478831594845828,
      "learning_rate": 7.769584155857019e-06,
      "loss": 0.9663,
      "step": 3038
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.201361352816128,
      "learning_rate": 7.763510704690645e-06,
      "loss": 0.9752,
      "step": 3039
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0489163028689052,
      "learning_rate": 7.757438121806414e-06,
      "loss": 0.94,
      "step": 3040
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9854962046181268,
      "learning_rate": 7.75136640956192e-06,
      "loss": 0.9264,
      "step": 3041
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.023151459538009,
      "learning_rate": 7.745295570314412e-06,
      "loss": 0.907,
      "step": 3042
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.083745896143517,
      "learning_rate": 7.739225606420793e-06,
      "loss": 0.8791,
      "step": 3043
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0341571187626728,
      "learning_rate": 7.733156520237633e-06,
      "loss": 0.8931,
      "step": 3044
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9384016010522643,
      "learning_rate": 7.727088314121165e-06,
      "loss": 0.619,
      "step": 3045
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9220991841995549,
      "learning_rate": 7.721020990427268e-06,
      "loss": 0.9407,
      "step": 3046
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.98209167921791,
      "learning_rate": 7.714954551511489e-06,
      "loss": 0.9454,
      "step": 3047
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0551344390335962,
      "learning_rate": 7.708888999729036e-06,
      "loss": 0.9006,
      "step": 3048
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7649911160438714,
      "learning_rate": 7.702824337434756e-06,
      "loss": 0.5959,
      "step": 3049
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.124154574568612,
      "learning_rate": 7.69676056698316e-06,
      "loss": 0.9263,
      "step": 3050
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.14706293122703,
      "learning_rate": 7.690697690728417e-06,
      "loss": 0.9561,
      "step": 3051
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9588340048495855,
      "learning_rate": 7.68463571102434e-06,
      "loss": 0.916,
      "step": 3052
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0471703357193078,
      "learning_rate": 7.678574630224399e-06,
      "loss": 0.9243,
      "step": 3053
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9570729747746243,
      "learning_rate": 7.672514450681721e-06,
      "loss": 0.5906,
      "step": 3054
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.95132096158119,
      "learning_rate": 7.666455174749066e-06,
      "loss": 0.8925,
      "step": 3055
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9887500759848046,
      "learning_rate": 7.66039680477886e-06,
      "loss": 0.8867,
      "step": 3056
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0274983847661532,
      "learning_rate": 7.654339343123173e-06,
      "loss": 0.8852,
      "step": 3057
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1020881665927706,
      "learning_rate": 7.648282792133711e-06,
      "loss": 0.8915,
      "step": 3058
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9170717777653694,
      "learning_rate": 7.642227154161841e-06,
      "loss": 0.9497,
      "step": 3059
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1428056676800622,
      "learning_rate": 7.636172431558575e-06,
      "loss": 0.9343,
      "step": 3060
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9621972374014511,
      "learning_rate": 7.630118626674557e-06,
      "loss": 0.8912,
      "step": 3061
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.960292332208309,
      "learning_rate": 7.6240657418600846e-06,
      "loss": 0.853,
      "step": 3062
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9204822944582083,
      "learning_rate": 7.618013779465101e-06,
      "loss": 0.6543,
      "step": 3063
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0768744298719084,
      "learning_rate": 7.611962741839178e-06,
      "loss": 0.9663,
      "step": 3064
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0497321120392509,
      "learning_rate": 7.6059126313315466e-06,
      "loss": 0.8844,
      "step": 3065
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1542118044389,
      "learning_rate": 7.599863450291056e-06,
      "loss": 0.9772,
      "step": 3066
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1692797366492187,
      "learning_rate": 7.593815201066215e-06,
      "loss": 0.903,
      "step": 3067
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9675441435666973,
      "learning_rate": 7.587767886005164e-06,
      "loss": 0.9272,
      "step": 3068
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1485371511961149,
      "learning_rate": 7.581721507455672e-06,
      "loss": 0.9538,
      "step": 3069
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0861482789816768,
      "learning_rate": 7.575676067765154e-06,
      "loss": 0.928,
      "step": 3070
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0471331919400841,
      "learning_rate": 7.569631569280662e-06,
      "loss": 0.9084,
      "step": 3071
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.289867767682808,
      "learning_rate": 7.563588014348871e-06,
      "loss": 0.9498,
      "step": 3072
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1922544416907725,
      "learning_rate": 7.5575454053161e-06,
      "loss": 0.9517,
      "step": 3073
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0863510307570559,
      "learning_rate": 7.551503744528304e-06,
      "loss": 0.9459,
      "step": 3074
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9796710114958569,
      "learning_rate": 7.545463034331054e-06,
      "loss": 0.932,
      "step": 3075
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8812380211340225,
      "learning_rate": 7.539423277069568e-06,
      "loss": 0.926,
      "step": 3076
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.803552252676909,
      "learning_rate": 7.53338447508869e-06,
      "loss": 0.6917,
      "step": 3077
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9133424229744846,
      "learning_rate": 7.52734663073288e-06,
      "loss": 0.8985,
      "step": 3078
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9652809593610318,
      "learning_rate": 7.521309746346246e-06,
      "loss": 0.9215,
      "step": 3079
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9635992147496162,
      "learning_rate": 7.515273824272516e-06,
      "loss": 0.9517,
      "step": 3080
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.042418895305631,
      "learning_rate": 7.509238866855033e-06,
      "loss": 0.9216,
      "step": 3081
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9602472726186779,
      "learning_rate": 7.503204876436785e-06,
      "loss": 0.9493,
      "step": 3082
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9938214202735962,
      "learning_rate": 7.497171855360372e-06,
      "loss": 0.9078,
      "step": 3083
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.049570536782212,
      "learning_rate": 7.491139805968018e-06,
      "loss": 0.9381,
      "step": 3084
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0523228851850406,
      "learning_rate": 7.485108730601571e-06,
      "loss": 0.938,
      "step": 3085
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7795651622070606,
      "learning_rate": 7.4790786316025125e-06,
      "loss": 0.582,
      "step": 3086
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.939252501729728,
      "learning_rate": 7.473049511311921e-06,
      "loss": 0.8499,
      "step": 3087
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8807356244875505,
      "learning_rate": 7.467021372070515e-06,
      "loss": 0.8925,
      "step": 3088
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0429456535025101,
      "learning_rate": 7.46099421621863e-06,
      "loss": 0.9118,
      "step": 3089
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0954571777696125,
      "learning_rate": 7.4549680460962044e-06,
      "loss": 0.9373,
      "step": 3090
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0783352936158495,
      "learning_rate": 7.448942864042819e-06,
      "loss": 0.9497,
      "step": 3091
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9776885373774188,
      "learning_rate": 7.4429186723976425e-06,
      "loss": 0.9047,
      "step": 3092
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.994462070649242,
      "learning_rate": 7.43689547349948e-06,
      "loss": 0.996,
      "step": 3093
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9664528676631559,
      "learning_rate": 7.43087326968675e-06,
      "loss": 0.9727,
      "step": 3094
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1550048750152262,
      "learning_rate": 7.42485206329747e-06,
      "loss": 0.967,
      "step": 3095
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9547714177786144,
      "learning_rate": 7.418831856669286e-06,
      "loss": 0.8823,
      "step": 3096
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9824453527875954,
      "learning_rate": 7.41281265213945e-06,
      "loss": 0.9152,
      "step": 3097
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.071700384683349,
      "learning_rate": 7.406794452044816e-06,
      "loss": 0.9302,
      "step": 3098
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8841218586545335,
      "learning_rate": 7.400777258721865e-06,
      "loss": 0.8587,
      "step": 3099
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.969102853639007,
      "learning_rate": 7.394761074506679e-06,
      "loss": 0.8945,
      "step": 3100
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9328201309058511,
      "learning_rate": 7.3887459017349405e-06,
      "loss": 0.9344,
      "step": 3101
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0267302123371707,
      "learning_rate": 7.382731742741953e-06,
      "loss": 0.9466,
      "step": 3102
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0254711001469659,
      "learning_rate": 7.376718599862621e-06,
      "loss": 0.9232,
      "step": 3103
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.953027584909666,
      "learning_rate": 7.370706475431446e-06,
      "loss": 0.9236,
      "step": 3104
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0070583901815462,
      "learning_rate": 7.364695371782547e-06,
      "loss": 0.9002,
      "step": 3105
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8829003644783593,
      "learning_rate": 7.358685291249644e-06,
      "loss": 0.9214,
      "step": 3106
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.989334936021155,
      "learning_rate": 7.352676236166051e-06,
      "loss": 0.9231,
      "step": 3107
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.801532908966086,
      "learning_rate": 7.346668208864695e-06,
      "loss": 0.6298,
      "step": 3108
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0551642421027916,
      "learning_rate": 7.3406612116781e-06,
      "loss": 0.851,
      "step": 3109
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1065330243339233,
      "learning_rate": 7.33465524693838e-06,
      "loss": 0.8991,
      "step": 3110
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0490575260709767,
      "learning_rate": 7.328650316977265e-06,
      "loss": 0.8997,
      "step": 3111
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0592596154493554,
      "learning_rate": 7.322646424126079e-06,
      "loss": 0.9199,
      "step": 3112
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1242660113436604,
      "learning_rate": 7.316643570715729e-06,
      "loss": 0.8734,
      "step": 3113
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.011583827615958,
      "learning_rate": 7.310641759076742e-06,
      "loss": 0.918,
      "step": 3114
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0373487173909006,
      "learning_rate": 7.304640991539216e-06,
      "loss": 0.8977,
      "step": 3115
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.063836933411038,
      "learning_rate": 7.2986412704328625e-06,
      "loss": 0.9028,
      "step": 3116
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.01701270560418,
      "learning_rate": 7.292642598086982e-06,
      "loss": 0.9334,
      "step": 3117
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0192038685964315,
      "learning_rate": 7.286644976830457e-06,
      "loss": 0.9403,
      "step": 3118
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.911259383019937,
      "learning_rate": 7.280648408991775e-06,
      "loss": 0.924,
      "step": 3119
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8655297956891403,
      "learning_rate": 7.274652896899015e-06,
      "loss": 0.656,
      "step": 3120
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9702816545667655,
      "learning_rate": 7.268658442879834e-06,
      "loss": 0.8681,
      "step": 3121
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0175897287340514,
      "learning_rate": 7.262665049261489e-06,
      "loss": 0.9081,
      "step": 3122
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0032881033177208,
      "learning_rate": 7.256672718370824e-06,
      "loss": 0.9382,
      "step": 3123
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.233918797944359,
      "learning_rate": 7.250681452534261e-06,
      "loss": 0.9755,
      "step": 3124
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9668559623492989,
      "learning_rate": 7.2446912540778196e-06,
      "loss": 0.9,
      "step": 3125
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0724077868764208,
      "learning_rate": 7.238702125327106e-06,
      "loss": 0.9506,
      "step": 3126
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0251414450376004,
      "learning_rate": 7.232714068607296e-06,
      "loss": 0.9197,
      "step": 3127
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9164086817240017,
      "learning_rate": 7.226727086243168e-06,
      "loss": 0.6293,
      "step": 3128
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5706825428254316,
      "learning_rate": 7.220741180559074e-06,
      "loss": 0.9542,
      "step": 3129
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0910927923981684,
      "learning_rate": 7.214756353878942e-06,
      "loss": 0.9295,
      "step": 3130
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1286288138298848,
      "learning_rate": 7.208772608526293e-06,
      "loss": 0.8896,
      "step": 3131
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0799958900942876,
      "learning_rate": 7.202789946824227e-06,
      "loss": 0.9581,
      "step": 3132
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9565830763888173,
      "learning_rate": 7.1968083710954075e-06,
      "loss": 0.9178,
      "step": 3133
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.965772055639199,
      "learning_rate": 7.1908278836621e-06,
      "loss": 0.8751,
      "step": 3134
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9067099961741266,
      "learning_rate": 7.184848486846128e-06,
      "loss": 0.9319,
      "step": 3135
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1788324828322996,
      "learning_rate": 7.178870182968904e-06,
      "loss": 0.9018,
      "step": 3136
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9440696492740462,
      "learning_rate": 7.1728929743514065e-06,
      "loss": 0.8759,
      "step": 3137
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.101747184516866,
      "learning_rate": 7.166916863314199e-06,
      "loss": 0.923,
      "step": 3138
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9836623094463397,
      "learning_rate": 7.1609418521774095e-06,
      "loss": 0.9312,
      "step": 3139
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0693733218595203,
      "learning_rate": 7.154967943260748e-06,
      "loss": 0.9135,
      "step": 3140
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9679830186598872,
      "learning_rate": 7.148995138883483e-06,
      "loss": 0.8815,
      "step": 3141
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2241180531723064,
      "learning_rate": 7.143023441364471e-06,
      "loss": 0.8752,
      "step": 3142
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1166415554098934,
      "learning_rate": 7.13705285302213e-06,
      "loss": 0.9293,
      "step": 3143
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.173502371457765,
      "learning_rate": 7.131083376174441e-06,
      "loss": 0.9674,
      "step": 3144
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.142268890120897,
      "learning_rate": 7.125115013138966e-06,
      "loss": 0.9108,
      "step": 3145
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8634553249407592,
      "learning_rate": 7.119147766232832e-06,
      "loss": 0.6126,
      "step": 3146
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9989404124125335,
      "learning_rate": 7.113181637772721e-06,
      "loss": 0.8818,
      "step": 3147
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0482709324081358,
      "learning_rate": 7.107216630074895e-06,
      "loss": 0.9526,
      "step": 3148
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.081862326933922,
      "learning_rate": 7.1012527454551795e-06,
      "loss": 0.9308,
      "step": 3149
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.143647533209886,
      "learning_rate": 7.09528998622895e-06,
      "loss": 0.913,
      "step": 3150
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9847784699110517,
      "learning_rate": 7.089328354711159e-06,
      "loss": 0.9048,
      "step": 3151
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.105063694033445,
      "learning_rate": 7.083367853216323e-06,
      "loss": 1.0229,
      "step": 3152
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0853166139911115,
      "learning_rate": 7.077408484058505e-06,
      "loss": 0.9557,
      "step": 3153
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.10697221983063,
      "learning_rate": 7.071450249551342e-06,
      "loss": 0.9575,
      "step": 3154
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9197084276441129,
      "learning_rate": 7.065493152008026e-06,
      "loss": 0.926,
      "step": 3155
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.06016650808104,
      "learning_rate": 7.059537193741306e-06,
      "loss": 0.9176,
      "step": 3156
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1222471049351075,
      "learning_rate": 7.053582377063489e-06,
      "loss": 0.9363,
      "step": 3157
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1061563031247057,
      "learning_rate": 7.047628704286446e-06,
      "loss": 0.9553,
      "step": 3158
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8162174873609558,
      "learning_rate": 7.041676177721588e-06,
      "loss": 0.6361,
      "step": 3159
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0311143944683905,
      "learning_rate": 7.035724799679898e-06,
      "loss": 0.9058,
      "step": 3160
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9664608436431386,
      "learning_rate": 7.029774572471904e-06,
      "loss": 0.9448,
      "step": 3161
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9944522859561733,
      "learning_rate": 7.023825498407689e-06,
      "loss": 0.8738,
      "step": 3162
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0498614020885475,
      "learning_rate": 7.0178775797968855e-06,
      "loss": 0.9266,
      "step": 3163
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0870576741795275,
      "learning_rate": 7.011930818948688e-06,
      "loss": 0.8843,
      "step": 3164
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2068854704185503,
      "learning_rate": 7.005985218171825e-06,
      "loss": 0.9445,
      "step": 3165
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1076283882565616,
      "learning_rate": 7.000040779774591e-06,
      "loss": 0.9181,
      "step": 3166
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7027065546920502,
      "learning_rate": 6.994097506064812e-06,
      "loss": 0.5956,
      "step": 3167
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.037609832467834,
      "learning_rate": 6.9881553993498805e-06,
      "loss": 0.92,
      "step": 3168
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1502934456923124,
      "learning_rate": 6.9822144619367275e-06,
      "loss": 0.9025,
      "step": 3169
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9416557085125922,
      "learning_rate": 6.97627469613182e-06,
      "loss": 0.885,
      "step": 3170
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.164163844686175,
      "learning_rate": 6.970336104241186e-06,
      "loss": 0.916,
      "step": 3171
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0672225884244575,
      "learning_rate": 6.9643986885703955e-06,
      "loss": 0.9133,
      "step": 3172
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.844298145632848,
      "learning_rate": 6.958462451424547e-06,
      "loss": 0.6117,
      "step": 3173
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0566931484974464,
      "learning_rate": 6.952527395108302e-06,
      "loss": 0.9388,
      "step": 3174
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.065607350472797,
      "learning_rate": 6.9465935219258504e-06,
      "loss": 0.8969,
      "step": 3175
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9935559420486787,
      "learning_rate": 6.9406608341809215e-06,
      "loss": 0.9117,
      "step": 3176
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0098215176222847,
      "learning_rate": 6.934729334176793e-06,
      "loss": 0.9589,
      "step": 3177
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0164216810672502,
      "learning_rate": 6.928799024216282e-06,
      "loss": 0.9094,
      "step": 3178
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1443087521039073,
      "learning_rate": 6.92286990660173e-06,
      "loss": 0.9567,
      "step": 3179
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1508542550019827,
      "learning_rate": 6.91694198363503e-06,
      "loss": 0.9591,
      "step": 3180
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1277072375437824,
      "learning_rate": 6.911015257617606e-06,
      "loss": 0.8966,
      "step": 3181
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.794938295983253,
      "learning_rate": 6.905089730850416e-06,
      "loss": 0.6234,
      "step": 3182
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9336357946435231,
      "learning_rate": 6.8991654056339505e-06,
      "loss": 0.6664,
      "step": 3183
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.941635274629767,
      "learning_rate": 6.893242284268244e-06,
      "loss": 0.924,
      "step": 3184
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0462862305623322,
      "learning_rate": 6.887320369052848e-06,
      "loss": 0.948,
      "step": 3185
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.061414719531071,
      "learning_rate": 6.8813996622868584e-06,
      "loss": 0.9338,
      "step": 3186
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8866450895160409,
      "learning_rate": 6.8754801662688964e-06,
      "loss": 0.8931,
      "step": 3187
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1763882961070398,
      "learning_rate": 6.869561883297116e-06,
      "loss": 0.9615,
      "step": 3188
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8125092681828407,
      "learning_rate": 6.863644815669197e-06,
      "loss": 0.7007,
      "step": 3189
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0299879833452583,
      "learning_rate": 6.857728965682344e-06,
      "loss": 0.9763,
      "step": 3190
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.95841666731552,
      "learning_rate": 6.851814335633298e-06,
      "loss": 0.9486,
      "step": 3191
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9292883574647439,
      "learning_rate": 6.8459009278183275e-06,
      "loss": 0.8838,
      "step": 3192
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9507585672528763,
      "learning_rate": 6.839988744533211e-06,
      "loss": 0.8921,
      "step": 3193
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9940180105671406,
      "learning_rate": 6.834077788073268e-06,
      "loss": 0.8904,
      "step": 3194
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0421192845588267,
      "learning_rate": 6.8281680607333364e-06,
      "loss": 0.9506,
      "step": 3195
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9192441829026712,
      "learning_rate": 6.822259564807768e-06,
      "loss": 0.9112,
      "step": 3196
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.098167433188225,
      "learning_rate": 6.81635230259045e-06,
      "loss": 0.9692,
      "step": 3197
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7877817315739737,
      "learning_rate": 6.810446276374789e-06,
      "loss": 0.6075,
      "step": 3198
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1369036986129795,
      "learning_rate": 6.8045414884536975e-06,
      "loss": 0.8966,
      "step": 3199
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9083514150181089,
      "learning_rate": 6.7986379411196255e-06,
      "loss": 0.8916,
      "step": 3200
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9605806597243982,
      "learning_rate": 6.7927356366645315e-06,
      "loss": 0.6195,
      "step": 3201
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0014892761834184,
      "learning_rate": 6.786834577379893e-06,
      "loss": 0.8872,
      "step": 3202
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0364922133650956,
      "learning_rate": 6.780934765556702e-06,
      "loss": 0.8858,
      "step": 3203
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8264959958703032,
      "learning_rate": 6.775036203485472e-06,
      "loss": 0.6215,
      "step": 3204
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1272273494447904,
      "learning_rate": 6.769138893456225e-06,
      "loss": 0.9252,
      "step": 3205
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.966693551389608,
      "learning_rate": 6.763242837758504e-06,
      "loss": 0.8747,
      "step": 3206
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.994267782967549,
      "learning_rate": 6.757348038681357e-06,
      "loss": 0.927,
      "step": 3207
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9238328064100274,
      "learning_rate": 6.751454498513349e-06,
      "loss": 0.8985,
      "step": 3208
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9755071674084042,
      "learning_rate": 6.745562219542554e-06,
      "loss": 0.9006,
      "step": 3209
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.134124972240751,
      "learning_rate": 6.7396712040565625e-06,
      "loss": 0.9691,
      "step": 3210
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.096433087619902,
      "learning_rate": 6.733781454342463e-06,
      "loss": 0.9341,
      "step": 3211
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1534400915292071,
      "learning_rate": 6.727892972686861e-06,
      "loss": 0.9038,
      "step": 3212
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0482654088703394,
      "learning_rate": 6.722005761375873e-06,
      "loss": 0.9727,
      "step": 3213
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.051327382873233,
      "learning_rate": 6.716119822695111e-06,
      "loss": 0.8735,
      "step": 3214
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.024673768666734,
      "learning_rate": 6.710235158929703e-06,
      "loss": 0.9236,
      "step": 3215
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0775910317260535,
      "learning_rate": 6.704351772364274e-06,
      "loss": 0.9045,
      "step": 3216
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.001041678892545,
      "learning_rate": 6.698469665282958e-06,
      "loss": 0.9271,
      "step": 3217
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1376836681997977,
      "learning_rate": 6.692588839969397e-06,
      "loss": 0.9258,
      "step": 3218
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.909694334160996,
      "learning_rate": 6.6867092987067214e-06,
      "loss": 0.9302,
      "step": 3219
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0020606376453922,
      "learning_rate": 6.680831043777579e-06,
      "loss": 0.9451,
      "step": 3220
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0364713259662162,
      "learning_rate": 6.674954077464108e-06,
      "loss": 0.9345,
      "step": 3221
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0747954108239768,
      "learning_rate": 6.6690784020479484e-06,
      "loss": 0.9381,
      "step": 3222
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0583890185999119,
      "learning_rate": 6.6632040198102364e-06,
      "loss": 0.9246,
      "step": 3223
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9946397589764213,
      "learning_rate": 6.657330933031619e-06,
      "loss": 0.8958,
      "step": 3224
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0393274512166546,
      "learning_rate": 6.651459143992221e-06,
      "loss": 0.8939,
      "step": 3225
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0170076021539454,
      "learning_rate": 6.645588654971677e-06,
      "loss": 0.9732,
      "step": 3226
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9083404810157191,
      "learning_rate": 6.639719468249115e-06,
      "loss": 0.6816,
      "step": 3227
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9510910180462167,
      "learning_rate": 6.633851586103153e-06,
      "loss": 0.9205,
      "step": 3228
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0696950490198573,
      "learning_rate": 6.627985010811903e-06,
      "loss": 0.9317,
      "step": 3229
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.259497397352002,
      "learning_rate": 6.622119744652977e-06,
      "loss": 0.9539,
      "step": 3230
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1813891624039319,
      "learning_rate": 6.616255789903467e-06,
      "loss": 0.9311,
      "step": 3231
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.013257532639054,
      "learning_rate": 6.610393148839964e-06,
      "loss": 0.8931,
      "step": 3232
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.97737140723111,
      "learning_rate": 6.6045318237385526e-06,
      "loss": 0.5941,
      "step": 3233
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9831417621469957,
      "learning_rate": 6.598671816874794e-06,
      "loss": 0.9537,
      "step": 3234
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9436083036007628,
      "learning_rate": 6.5928131305237465e-06,
      "loss": 0.9398,
      "step": 3235
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1300910314395856,
      "learning_rate": 6.586955766959958e-06,
      "loss": 0.9207,
      "step": 3236
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0074818282594,
      "learning_rate": 6.581099728457451e-06,
      "loss": 0.9383,
      "step": 3237
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.074582313656539,
      "learning_rate": 6.5752450172897466e-06,
      "loss": 0.9404,
      "step": 3238
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9825511065523188,
      "learning_rate": 6.569391635729847e-06,
      "loss": 0.9339,
      "step": 3239
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9821014775317561,
      "learning_rate": 6.563539586050233e-06,
      "loss": 0.8543,
      "step": 3240
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0148619623546928,
      "learning_rate": 6.557688870522871e-06,
      "loss": 0.8853,
      "step": 3241
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8784438591153222,
      "learning_rate": 6.551839491419213e-06,
      "loss": 0.6241,
      "step": 3242
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1108140602892322,
      "learning_rate": 6.545991451010185e-06,
      "loss": 0.9704,
      "step": 3243
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0410805637447236,
      "learning_rate": 6.5401447515662065e-06,
      "loss": 0.9016,
      "step": 3244
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1402850309863763,
      "learning_rate": 6.5342993953571556e-06,
      "loss": 0.8912,
      "step": 3245
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0837652534816082,
      "learning_rate": 6.52845538465241e-06,
      "loss": 0.9413,
      "step": 3246
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9855392726687381,
      "learning_rate": 6.522612721720813e-06,
      "loss": 0.9047,
      "step": 3247
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8990780166879591,
      "learning_rate": 6.5167714088306865e-06,
      "loss": 0.8902,
      "step": 3248
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8471669109129004,
      "learning_rate": 6.51093144824983e-06,
      "loss": 0.6521,
      "step": 3249
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8365214058517145,
      "learning_rate": 6.505092842245519e-06,
      "loss": 0.6651,
      "step": 3250
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8328973834182941,
      "learning_rate": 6.499255593084498e-06,
      "loss": 0.6035,
      "step": 3251
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.019550709817323,
      "learning_rate": 6.493419703032991e-06,
      "loss": 0.9427,
      "step": 3252
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.91346453069879,
      "learning_rate": 6.487585174356691e-06,
      "loss": 0.9403,
      "step": 3253
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8336123718918507,
      "learning_rate": 6.481752009320761e-06,
      "loss": 0.6246,
      "step": 3254
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1081412506428108,
      "learning_rate": 6.4759202101898366e-06,
      "loss": 0.9554,
      "step": 3255
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1369528175106012,
      "learning_rate": 6.4700897792280285e-06,
      "loss": 0.9415,
      "step": 3256
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0415579030257094,
      "learning_rate": 6.464260718698902e-06,
      "loss": 0.9771,
      "step": 3257
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9969393850161729,
      "learning_rate": 6.458433030865503e-06,
      "loss": 0.9429,
      "step": 3258
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0359641512223023,
      "learning_rate": 6.452606717990346e-06,
      "loss": 0.9071,
      "step": 3259
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1095481002107224,
      "learning_rate": 6.4467817823354005e-06,
      "loss": 0.9685,
      "step": 3260
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9736298470896765,
      "learning_rate": 6.440958226162104e-06,
      "loss": 0.9229,
      "step": 3261
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.940000257012102,
      "learning_rate": 6.43513605173137e-06,
      "loss": 0.9051,
      "step": 3262
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0660931167659364,
      "learning_rate": 6.4293152613035594e-06,
      "loss": 0.9817,
      "step": 3263
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9717682545280657,
      "learning_rate": 6.4234958571385095e-06,
      "loss": 0.9204,
      "step": 3264
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.082428762341039,
      "learning_rate": 6.4176778414955075e-06,
      "loss": 0.9505,
      "step": 3265
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0013178943546082,
      "learning_rate": 6.4118612166333124e-06,
      "loss": 0.8791,
      "step": 3266
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8482438426272806,
      "learning_rate": 6.4060459848101354e-06,
      "loss": 0.6159,
      "step": 3267
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1630810688262814,
      "learning_rate": 6.400232148283651e-06,
      "loss": 0.9339,
      "step": 3268
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8035822005663951,
      "learning_rate": 6.3944197093109885e-06,
      "loss": 0.6538,
      "step": 3269
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.010910591663933,
      "learning_rate": 6.388608670148741e-06,
      "loss": 0.9537,
      "step": 3270
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0860933740411727,
      "learning_rate": 6.38279903305295e-06,
      "loss": 0.956,
      "step": 3271
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9864180362200226,
      "learning_rate": 6.376990800279119e-06,
      "loss": 0.8779,
      "step": 3272
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9234253424180612,
      "learning_rate": 6.3711839740822035e-06,
      "loss": 0.9562,
      "step": 3273
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0637337924816466,
      "learning_rate": 6.3653785567166125e-06,
      "loss": 0.9797,
      "step": 3274
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9957225021248001,
      "learning_rate": 6.359574550436209e-06,
      "loss": 0.9179,
      "step": 3275
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9709210539750355,
      "learning_rate": 6.3537719574943105e-06,
      "loss": 0.9178,
      "step": 3276
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1434545587149194,
      "learning_rate": 6.347970780143678e-06,
      "loss": 0.9372,
      "step": 3277
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9199469268327689,
      "learning_rate": 6.342171020636533e-06,
      "loss": 0.8966,
      "step": 3278
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.809676109655382,
      "learning_rate": 6.336372681224543e-06,
      "loss": 0.65,
      "step": 3279
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1682625531808484,
      "learning_rate": 6.330575764158819e-06,
      "loss": 0.9531,
      "step": 3280
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0577600508999767,
      "learning_rate": 6.324780271689923e-06,
      "loss": 0.9228,
      "step": 3281
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8528552564142523,
      "learning_rate": 6.318986206067872e-06,
      "loss": 0.6164,
      "step": 3282
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7815812424941897,
      "learning_rate": 6.313193569542113e-06,
      "loss": 0.6323,
      "step": 3283
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.994342826368131,
      "learning_rate": 6.30740236436155e-06,
      "loss": 0.9015,
      "step": 3284
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9683655684581984,
      "learning_rate": 6.301612592774533e-06,
      "loss": 0.9379,
      "step": 3285
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0512887448655974,
      "learning_rate": 6.295824257028844e-06,
      "loss": 0.9323,
      "step": 3286
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.827997750754387,
      "learning_rate": 6.290037359371717e-06,
      "loss": 0.5987,
      "step": 3287
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0623214736923068,
      "learning_rate": 6.284251902049827e-06,
      "loss": 0.9362,
      "step": 3288
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0880279963652295,
      "learning_rate": 6.278467887309283e-06,
      "loss": 0.9745,
      "step": 3289
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0103575155312141,
      "learning_rate": 6.272685317395644e-06,
      "loss": 0.8428,
      "step": 3290
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.869940350177472,
      "learning_rate": 6.266904194553896e-06,
      "loss": 0.5723,
      "step": 3291
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0884438424101204,
      "learning_rate": 6.261124521028477e-06,
      "loss": 0.9191,
      "step": 3292
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.116753103176311,
      "learning_rate": 6.255346299063252e-06,
      "loss": 0.8988,
      "step": 3293
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0236493447183728,
      "learning_rate": 6.249569530901525e-06,
      "loss": 0.9373,
      "step": 3294
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.989150574813145,
      "learning_rate": 6.243794218786034e-06,
      "loss": 0.9317,
      "step": 3295
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.084686431378175,
      "learning_rate": 6.238020364958964e-06,
      "loss": 0.8759,
      "step": 3296
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1836255339959596,
      "learning_rate": 6.232247971661912e-06,
      "loss": 0.9145,
      "step": 3297
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9905117013838158,
      "learning_rate": 6.2264770411359256e-06,
      "loss": 0.9115,
      "step": 3298
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0994911591894385,
      "learning_rate": 6.22070757562148e-06,
      "loss": 0.9316,
      "step": 3299
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0686738204837256,
      "learning_rate": 6.214939577358479e-06,
      "loss": 0.9188,
      "step": 3300
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1553743619439734,
      "learning_rate": 6.209173048586253e-06,
      "loss": 0.9463,
      "step": 3301
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0872460827066888,
      "learning_rate": 6.203407991543577e-06,
      "loss": 0.9273,
      "step": 3302
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0232436375517602,
      "learning_rate": 6.197644408468635e-06,
      "loss": 0.9147,
      "step": 3303
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0285814045293118,
      "learning_rate": 6.191882301599052e-06,
      "loss": 0.8786,
      "step": 3304
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0464454900050155,
      "learning_rate": 6.186121673171882e-06,
      "loss": 0.9553,
      "step": 3305
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.222360873019979,
      "learning_rate": 6.180362525423591e-06,
      "loss": 0.9832,
      "step": 3306
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9686262504500585,
      "learning_rate": 6.174604860590081e-06,
      "loss": 0.9062,
      "step": 3307
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0381901364411414,
      "learning_rate": 6.168848680906678e-06,
      "loss": 0.9053,
      "step": 3308
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0846346598257521,
      "learning_rate": 6.163093988608127e-06,
      "loss": 0.9048,
      "step": 3309
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0007031739617083,
      "learning_rate": 6.157340785928595e-06,
      "loss": 0.9399,
      "step": 3310
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3125371445213547,
      "learning_rate": 6.151589075101681e-06,
      "loss": 0.9041,
      "step": 3311
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9903783266830389,
      "learning_rate": 6.145838858360391e-06,
      "loss": 0.9386,
      "step": 3312
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8734759492730398,
      "learning_rate": 6.140090137937158e-06,
      "loss": 0.6623,
      "step": 3313
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9825100838405381,
      "learning_rate": 6.134342916063838e-06,
      "loss": 0.9217,
      "step": 3314
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9488531012308292,
      "learning_rate": 6.128597194971691e-06,
      "loss": 0.8873,
      "step": 3315
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9623391919630828,
      "learning_rate": 6.122852976891413e-06,
      "loss": 0.9381,
      "step": 3316
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0477892858498226,
      "learning_rate": 6.117110264053101e-06,
      "loss": 0.9032,
      "step": 3317
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.004345515274,
      "learning_rate": 6.111369058686276e-06,
      "loss": 0.8885,
      "step": 3318
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.072748099759778,
      "learning_rate": 6.105629363019875e-06,
      "loss": 0.9235,
      "step": 3319
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1811962218096324,
      "learning_rate": 6.099891179282242e-06,
      "loss": 0.9316,
      "step": 3320
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0112936572544025,
      "learning_rate": 6.094154509701133e-06,
      "loss": 0.8794,
      "step": 3321
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0115773865581703,
      "learning_rate": 6.088419356503732e-06,
      "loss": 0.9357,
      "step": 3322
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9448669267459859,
      "learning_rate": 6.082685721916612e-06,
      "loss": 0.6232,
      "step": 3323
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0357501312456934,
      "learning_rate": 6.076953608165772e-06,
      "loss": 0.9136,
      "step": 3324
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0834185760363482,
      "learning_rate": 6.07122301747662e-06,
      "loss": 0.9268,
      "step": 3325
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8224039690509108,
      "learning_rate": 6.065493952073961e-06,
      "loss": 0.6278,
      "step": 3326
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0434618350888154,
      "learning_rate": 6.0597664141820176e-06,
      "loss": 0.9158,
      "step": 3327
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0417174852660929,
      "learning_rate": 6.054040406024422e-06,
      "loss": 0.936,
      "step": 3328
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0228027120480851,
      "learning_rate": 6.0483159298242e-06,
      "loss": 0.9576,
      "step": 3329
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0490416589355431,
      "learning_rate": 6.042592987803796e-06,
      "loss": 0.9395,
      "step": 3330
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7878124596873042,
      "learning_rate": 6.036871582185054e-06,
      "loss": 0.5993,
      "step": 3331
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9344024110015967,
      "learning_rate": 6.031151715189217e-06,
      "loss": 0.864,
      "step": 3332
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0056926241864108,
      "learning_rate": 6.025433389036935e-06,
      "loss": 0.9283,
      "step": 3333
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1967447333637984,
      "learning_rate": 6.019716605948261e-06,
      "loss": 0.9704,
      "step": 3334
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1206884245408626,
      "learning_rate": 6.014001368142643e-06,
      "loss": 0.8734,
      "step": 3335
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9304671849838982,
      "learning_rate": 6.008287677838937e-06,
      "loss": 0.8608,
      "step": 3336
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9469577864260029,
      "learning_rate": 6.002575537255395e-06,
      "loss": 0.9175,
      "step": 3337
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1010092115560548,
      "learning_rate": 5.996864948609662e-06,
      "loss": 0.9246,
      "step": 3338
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0570562696720591,
      "learning_rate": 5.9911559141187924e-06,
      "loss": 0.8666,
      "step": 3339
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0121409506879933,
      "learning_rate": 5.9854484359992235e-06,
      "loss": 0.9546,
      "step": 3340
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8316837259979045,
      "learning_rate": 5.979742516466793e-06,
      "loss": 0.6886,
      "step": 3341
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0935312026728652,
      "learning_rate": 5.974038157736746e-06,
      "loss": 0.862,
      "step": 3342
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0966740229741427,
      "learning_rate": 5.968335362023697e-06,
      "loss": 0.9575,
      "step": 3343
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9047979623121948,
      "learning_rate": 5.962634131541676e-06,
      "loss": 0.9485,
      "step": 3344
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9216657965398043,
      "learning_rate": 5.956934468504101e-06,
      "loss": 0.8724,
      "step": 3345
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0372693033706963,
      "learning_rate": 5.951236375123768e-06,
      "loss": 0.8757,
      "step": 3346
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0187348248095287,
      "learning_rate": 5.945539853612876e-06,
      "loss": 0.9792,
      "step": 3347
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.050751170273824,
      "learning_rate": 5.939844906183016e-06,
      "loss": 0.858,
      "step": 3348
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0605582329719692,
      "learning_rate": 5.934151535045156e-06,
      "loss": 0.9307,
      "step": 3349
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9962574623247532,
      "learning_rate": 5.92845974240966e-06,
      "loss": 0.8701,
      "step": 3350
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0687941673534798,
      "learning_rate": 5.922769530486283e-06,
      "loss": 0.9586,
      "step": 3351
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9796735456832658,
      "learning_rate": 5.917080901484156e-06,
      "loss": 0.9128,
      "step": 3352
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9801198328535211,
      "learning_rate": 5.9113938576118e-06,
      "loss": 0.945,
      "step": 3353
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0118807933071183,
      "learning_rate": 5.905708401077128e-06,
      "loss": 0.9189,
      "step": 3354
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.997933141807704,
      "learning_rate": 5.900024534087421e-06,
      "loss": 0.9281,
      "step": 3355
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.04672055753247,
      "learning_rate": 5.894342258849355e-06,
      "loss": 0.8914,
      "step": 3356
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0814168442106398,
      "learning_rate": 5.88866157756899e-06,
      "loss": 0.8794,
      "step": 3357
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9024326238895273,
      "learning_rate": 5.882982492451757e-06,
      "loss": 0.6427,
      "step": 3358
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.039802644727357,
      "learning_rate": 5.877305005702471e-06,
      "loss": 0.9046,
      "step": 3359
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0841030373791956,
      "learning_rate": 5.871629119525335e-06,
      "loss": 0.9154,
      "step": 3360
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0030829559521626,
      "learning_rate": 5.865954836123915e-06,
      "loss": 0.9027,
      "step": 3361
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0764385690616987,
      "learning_rate": 5.860282157701167e-06,
      "loss": 0.9137,
      "step": 3362
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0190649240428973,
      "learning_rate": 5.854611086459423e-06,
      "loss": 0.9055,
      "step": 3363
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0109961105514556,
      "learning_rate": 5.8489416246003814e-06,
      "loss": 0.9787,
      "step": 3364
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7866898090213479,
      "learning_rate": 5.8432737743251315e-06,
      "loss": 0.5878,
      "step": 3365
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1130272571372841,
      "learning_rate": 5.8376075378341194e-06,
      "loss": 0.8841,
      "step": 3366
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8240482057269966,
      "learning_rate": 5.831942917327172e-06,
      "loss": 0.6519,
      "step": 3367
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0064879887105598,
      "learning_rate": 5.826279915003503e-06,
      "loss": 0.9976,
      "step": 3368
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9278974074867553,
      "learning_rate": 5.8206185330616725e-06,
      "loss": 0.8798,
      "step": 3369
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0365361330700733,
      "learning_rate": 5.814958773699625e-06,
      "loss": 0.9173,
      "step": 3370
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0529913067897603,
      "learning_rate": 5.809300639114683e-06,
      "loss": 0.9391,
      "step": 3371
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2876953056146825,
      "learning_rate": 5.803644131503516e-06,
      "loss": 0.9671,
      "step": 3372
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9550539617758264,
      "learning_rate": 5.797989253062186e-06,
      "loss": 0.8793,
      "step": 3373
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0421284894361338,
      "learning_rate": 5.792336005986105e-06,
      "loss": 0.9459,
      "step": 3374
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0760656088427383,
      "learning_rate": 5.786684392470064e-06,
      "loss": 0.9693,
      "step": 3375
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2511357426524945,
      "learning_rate": 5.781034414708208e-06,
      "loss": 0.9594,
      "step": 3376
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1509061706139323,
      "learning_rate": 5.775386074894058e-06,
      "loss": 0.9289,
      "step": 3377
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.04941999805862,
      "learning_rate": 5.769739375220489e-06,
      "loss": 0.9783,
      "step": 3378
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0064642944790818,
      "learning_rate": 5.7640943178797445e-06,
      "loss": 0.9421,
      "step": 3379
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0056048772657782,
      "learning_rate": 5.7584509050634395e-06,
      "loss": 0.9312,
      "step": 3380
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0752386715415858,
      "learning_rate": 5.752809138962525e-06,
      "loss": 0.915,
      "step": 3381
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9548602493163637,
      "learning_rate": 5.747169021767342e-06,
      "loss": 0.9243,
      "step": 3382
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8654209930286221,
      "learning_rate": 5.7415305556675805e-06,
      "loss": 0.6332,
      "step": 3383
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8769719540246207,
      "learning_rate": 5.73589374285227e-06,
      "loss": 0.8906,
      "step": 3384
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.038562438989506,
      "learning_rate": 5.730258585509832e-06,
      "loss": 0.9406,
      "step": 3385
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.076084034072932,
      "learning_rate": 5.724625085828022e-06,
      "loss": 0.9464,
      "step": 3386
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0590252920440346,
      "learning_rate": 5.718993245993958e-06,
      "loss": 0.9169,
      "step": 3387
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1145747050285786,
      "learning_rate": 5.713363068194115e-06,
      "loss": 0.9288,
      "step": 3388
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9623316637752153,
      "learning_rate": 5.7077345546143235e-06,
      "loss": 0.8955,
      "step": 3389
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9246029326597627,
      "learning_rate": 5.702107707439766e-06,
      "loss": 0.6534,
      "step": 3390
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9682958998925353,
      "learning_rate": 5.6964825288549745e-06,
      "loss": 0.8573,
      "step": 3391
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0417846521723282,
      "learning_rate": 5.690859021043842e-06,
      "loss": 0.8806,
      "step": 3392
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0258842315608236,
      "learning_rate": 5.685237186189601e-06,
      "loss": 0.9063,
      "step": 3393
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9572118913457894,
      "learning_rate": 5.679617026474853e-06,
      "loss": 0.6733,
      "step": 3394
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1317110529652599,
      "learning_rate": 5.673998544081527e-06,
      "loss": 0.9763,
      "step": 3395
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0247952749257363,
      "learning_rate": 5.6683817411909114e-06,
      "loss": 0.9062,
      "step": 3396
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0510541646122256,
      "learning_rate": 5.662766619983653e-06,
      "loss": 0.9497,
      "step": 3397
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0740751876050811,
      "learning_rate": 5.65715318263972e-06,
      "loss": 0.9013,
      "step": 3398
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9717337007462356,
      "learning_rate": 5.651541431338454e-06,
      "loss": 0.9259,
      "step": 3399
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.290033258501579,
      "learning_rate": 5.645931368258527e-06,
      "loss": 1.0005,
      "step": 3400
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1426624481964374,
      "learning_rate": 5.640322995577958e-06,
      "loss": 0.923,
      "step": 3401
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.823958409468028,
      "learning_rate": 5.634716315474109e-06,
      "loss": 0.6337,
      "step": 3402
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.063331014417569,
      "learning_rate": 5.629111330123689e-06,
      "loss": 0.9179,
      "step": 3403
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2870703725303365,
      "learning_rate": 5.623508041702743e-06,
      "loss": 0.943,
      "step": 3404
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.108716825726232,
      "learning_rate": 5.617906452386659e-06,
      "loss": 0.9714,
      "step": 3405
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8908840077843557,
      "learning_rate": 5.612306564350179e-06,
      "loss": 0.583,
      "step": 3406
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.036436515058092,
      "learning_rate": 5.6067083797673535e-06,
      "loss": 0.9203,
      "step": 3407
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1272737790223992,
      "learning_rate": 5.601111900811607e-06,
      "loss": 0.926,
      "step": 3408
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.943594486642453,
      "learning_rate": 5.595517129655681e-06,
      "loss": 0.9033,
      "step": 3409
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7746995548334493,
      "learning_rate": 5.589924068471648e-06,
      "loss": 0.5978,
      "step": 3410
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.95243368057666,
      "learning_rate": 5.58433271943094e-06,
      "loss": 0.9144,
      "step": 3411
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.971795666424823,
      "learning_rate": 5.578743084704306e-06,
      "loss": 0.9088,
      "step": 3412
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8700145247853057,
      "learning_rate": 5.573155166461833e-06,
      "loss": 0.8821,
      "step": 3413
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7578838023887858,
      "learning_rate": 5.567568966872947e-06,
      "loss": 0.6458,
      "step": 3414
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0219131708489386,
      "learning_rate": 5.5619844881064e-06,
      "loss": 0.9283,
      "step": 3415
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.998841289143347,
      "learning_rate": 5.556401732330281e-06,
      "loss": 0.9353,
      "step": 3416
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9623549836552222,
      "learning_rate": 5.550820701712007e-06,
      "loss": 0.9602,
      "step": 3417
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0357162699387144,
      "learning_rate": 5.545241398418326e-06,
      "loss": 0.9073,
      "step": 3418
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1739044900497244,
      "learning_rate": 5.539663824615312e-06,
      "loss": 0.9201,
      "step": 3419
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7911792047338502,
      "learning_rate": 5.534087982468384e-06,
      "loss": 0.6231,
      "step": 3420
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8875029590771012,
      "learning_rate": 5.5285138741422615e-06,
      "loss": 0.8707,
      "step": 3421
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.023610341149873,
      "learning_rate": 5.522941501801008e-06,
      "loss": 0.8961,
      "step": 3422
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.126222692029327,
      "learning_rate": 5.517370867608021e-06,
      "loss": 0.899,
      "step": 3423
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1802440482704801,
      "learning_rate": 5.511801973725997e-06,
      "loss": 0.9357,
      "step": 3424
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.013842785581478,
      "learning_rate": 5.506234822316983e-06,
      "loss": 0.8909,
      "step": 3425
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9552436598130135,
      "learning_rate": 5.500669415542336e-06,
      "loss": 0.6217,
      "step": 3426
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8629753233659947,
      "learning_rate": 5.495105755562738e-06,
      "loss": 0.655,
      "step": 3427
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9251901780860363,
      "learning_rate": 5.4895438445381945e-06,
      "loss": 0.6794,
      "step": 3428
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9722095508869641,
      "learning_rate": 5.48398368462803e-06,
      "loss": 0.8715,
      "step": 3429
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8615092462293183,
      "learning_rate": 5.4784252779908905e-06,
      "loss": 0.8496,
      "step": 3430
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0462681435684755,
      "learning_rate": 5.4728686267847354e-06,
      "loss": 0.9259,
      "step": 3431
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1139963450893275,
      "learning_rate": 5.467313733166863e-06,
      "loss": 0.9469,
      "step": 3432
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0209526144426988,
      "learning_rate": 5.461760599293855e-06,
      "loss": 0.8564,
      "step": 3433
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9559707084775878,
      "learning_rate": 5.456209227321643e-06,
      "loss": 0.9406,
      "step": 3434
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0224544912971225,
      "learning_rate": 5.450659619405458e-06,
      "loss": 0.9252,
      "step": 3435
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9445204315421847,
      "learning_rate": 5.445111777699842e-06,
      "loss": 0.944,
      "step": 3436
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0534217922105589,
      "learning_rate": 5.439565704358667e-06,
      "loss": 0.9847,
      "step": 3437
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7542946306456507,
      "learning_rate": 5.434021401535105e-06,
      "loss": 0.6033,
      "step": 3438
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0102750249142292,
      "learning_rate": 5.428478871381646e-06,
      "loss": 0.8996,
      "step": 3439
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1323370505112125,
      "learning_rate": 5.422938116050092e-06,
      "loss": 0.9124,
      "step": 3440
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0515615863303653,
      "learning_rate": 5.417399137691552e-06,
      "loss": 0.9091,
      "step": 3441
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1584685846222151,
      "learning_rate": 5.411861938456453e-06,
      "loss": 0.9369,
      "step": 3442
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1157024039469239,
      "learning_rate": 5.406326520494522e-06,
      "loss": 0.8941,
      "step": 3443
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0591619972740303,
      "learning_rate": 5.400792885954802e-06,
      "loss": 0.9486,
      "step": 3444
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0123395287926482,
      "learning_rate": 5.395261036985635e-06,
      "loss": 0.9209,
      "step": 3445
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.041246374520419,
      "learning_rate": 5.389730975734686e-06,
      "loss": 0.9245,
      "step": 3446
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2005462822967272,
      "learning_rate": 5.384202704348902e-06,
      "loss": 0.9415,
      "step": 3447
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9650602024303365,
      "learning_rate": 5.378676224974557e-06,
      "loss": 0.8932,
      "step": 3448
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9876056522765022,
      "learning_rate": 5.373151539757224e-06,
      "loss": 0.863,
      "step": 3449
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.928207171329524,
      "learning_rate": 5.367628650841761e-06,
      "loss": 0.9311,
      "step": 3450
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9693124699485284,
      "learning_rate": 5.362107560372358e-06,
      "loss": 0.9035,
      "step": 3451
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0194108082755209,
      "learning_rate": 5.356588270492487e-06,
      "loss": 0.8912,
      "step": 3452
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9547849211917356,
      "learning_rate": 5.351070783344926e-06,
      "loss": 0.9316,
      "step": 3453
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.030435146843312,
      "learning_rate": 5.3455551010717545e-06,
      "loss": 0.9328,
      "step": 3454
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.089092201880755,
      "learning_rate": 5.34004122581435e-06,
      "loss": 0.9156,
      "step": 3455
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9846890725240073,
      "learning_rate": 5.334529159713389e-06,
      "loss": 0.8805,
      "step": 3456
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9184629005617747,
      "learning_rate": 5.329018904908841e-06,
      "loss": 0.9361,
      "step": 3457
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8191869980281364,
      "learning_rate": 5.323510463539989e-06,
      "loss": 0.6588,
      "step": 3458
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.757161373649974,
      "learning_rate": 5.318003837745382e-06,
      "loss": 0.6108,
      "step": 3459
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0882647271760764,
      "learning_rate": 5.3124990296628974e-06,
      "loss": 0.9325,
      "step": 3460
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0752250135011177,
      "learning_rate": 5.306996041429688e-06,
      "loss": 0.9588,
      "step": 3461
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1270527913351884,
      "learning_rate": 5.301494875182192e-06,
      "loss": 0.8979,
      "step": 3462
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8809308643681488,
      "learning_rate": 5.295995533056162e-06,
      "loss": 0.8576,
      "step": 3463
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2005496782301284,
      "learning_rate": 5.290498017186631e-06,
      "loss": 0.9492,
      "step": 3464
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9108819788766401,
      "learning_rate": 5.2850023297079235e-06,
      "loss": 0.8881,
      "step": 3465
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2878157254151819,
      "learning_rate": 5.279508472753654e-06,
      "loss": 0.9657,
      "step": 3466
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9320870370588603,
      "learning_rate": 5.274016448456725e-06,
      "loss": 0.9186,
      "step": 3467
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0440478820924233,
      "learning_rate": 5.2685262589493314e-06,
      "loss": 0.9401,
      "step": 3468
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0610932879755182,
      "learning_rate": 5.263037906362953e-06,
      "loss": 0.9519,
      "step": 3469
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0395830211235997,
      "learning_rate": 5.257551392828359e-06,
      "loss": 0.8759,
      "step": 3470
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0405361306290724,
      "learning_rate": 5.252066720475597e-06,
      "loss": 0.9487,
      "step": 3471
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0501201953950579,
      "learning_rate": 5.246583891434018e-06,
      "loss": 0.9465,
      "step": 3472
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.981858706357324,
      "learning_rate": 5.241102907832232e-06,
      "loss": 0.8742,
      "step": 3473
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9375338650090452,
      "learning_rate": 5.235623771798151e-06,
      "loss": 0.9117,
      "step": 3474
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.056239226167444,
      "learning_rate": 5.23014648545897e-06,
      "loss": 0.877,
      "step": 3475
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0278539118397254,
      "learning_rate": 5.224671050941146e-06,
      "loss": 0.9066,
      "step": 3476
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9875206024983622,
      "learning_rate": 5.2191974703704425e-06,
      "loss": 0.6666,
      "step": 3477
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.978976819741027,
      "learning_rate": 5.213725745871889e-06,
      "loss": 0.9449,
      "step": 3478
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0129736280618038,
      "learning_rate": 5.208255879569799e-06,
      "loss": 0.9468,
      "step": 3479
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.174611814224887,
      "learning_rate": 5.20278787358776e-06,
      "loss": 0.9117,
      "step": 3480
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9569162365351106,
      "learning_rate": 5.197321730048641e-06,
      "loss": 0.9265,
      "step": 3481
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0063449011870926,
      "learning_rate": 5.1918574510745865e-06,
      "loss": 0.894,
      "step": 3482
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0749878011818088,
      "learning_rate": 5.186395038787017e-06,
      "loss": 0.8953,
      "step": 3483
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2120704265671758,
      "learning_rate": 5.180934495306638e-06,
      "loss": 0.9314,
      "step": 3484
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8713264868073656,
      "learning_rate": 5.175475822753404e-06,
      "loss": 0.6272,
      "step": 3485
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1078183107482635,
      "learning_rate": 5.170019023246574e-06,
      "loss": 0.9362,
      "step": 3486
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0609778625978834,
      "learning_rate": 5.16456409890466e-06,
      "loss": 0.9486,
      "step": 3487
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9480943517915481,
      "learning_rate": 5.159111051845451e-06,
      "loss": 0.866,
      "step": 3488
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0027604590603985,
      "learning_rate": 5.153659884186013e-06,
      "loss": 0.8991,
      "step": 3489
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0055758264094272,
      "learning_rate": 5.148210598042665e-06,
      "loss": 0.9313,
      "step": 3490
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8134291641925318,
      "learning_rate": 5.142763195531017e-06,
      "loss": 0.6111,
      "step": 3491
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.149495694770898,
      "learning_rate": 5.137317678765939e-06,
      "loss": 0.8467,
      "step": 3492
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.936637199759753,
      "learning_rate": 5.131874049861563e-06,
      "loss": 0.9133,
      "step": 3493
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0045821344312125,
      "learning_rate": 5.126432310931295e-06,
      "loss": 0.9274,
      "step": 3494
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8448733024539461,
      "learning_rate": 5.120992464087807e-06,
      "loss": 0.6026,
      "step": 3495
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.053541972681646,
      "learning_rate": 5.115554511443033e-06,
      "loss": 0.93,
      "step": 3496
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1803307565664498,
      "learning_rate": 5.1101184551081705e-06,
      "loss": 0.9813,
      "step": 3497
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0612506088142337,
      "learning_rate": 5.104684297193694e-06,
      "loss": 0.9372,
      "step": 3498
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9687394530448985,
      "learning_rate": 5.099252039809317e-06,
      "loss": 0.9132,
      "step": 3499
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1080232230923115,
      "learning_rate": 5.09382168506404e-06,
      "loss": 0.9278,
      "step": 3500
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9595865030425236,
      "learning_rate": 5.088393235066114e-06,
      "loss": 0.9441,
      "step": 3501
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.043063211159357,
      "learning_rate": 5.082966691923037e-06,
      "loss": 0.6374,
      "step": 3502
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.190626501707572,
      "learning_rate": 5.077542057741592e-06,
      "loss": 0.9275,
      "step": 3503
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0935100113316565,
      "learning_rate": 5.0721193346278066e-06,
      "loss": 0.9572,
      "step": 3504
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.867520720049299,
      "learning_rate": 5.066698524686966e-06,
      "loss": 0.6369,
      "step": 3505
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.066424917662022,
      "learning_rate": 5.061279630023618e-06,
      "loss": 0.9532,
      "step": 3506
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0261432633007002,
      "learning_rate": 5.055862652741562e-06,
      "loss": 0.9124,
      "step": 3507
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0723777631583034,
      "learning_rate": 5.050447594943856e-06,
      "loss": 0.9534,
      "step": 3508
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0571220556305883,
      "learning_rate": 5.045034458732808e-06,
      "loss": 0.9027,
      "step": 3509
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0191098625579929,
      "learning_rate": 5.0396232462099945e-06,
      "loss": 0.9037,
      "step": 3510
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9745511859903125,
      "learning_rate": 5.034213959476222e-06,
      "loss": 0.8671,
      "step": 3511
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1131734127595114,
      "learning_rate": 5.028806600631569e-06,
      "loss": 0.9481,
      "step": 3512
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1823957725796876,
      "learning_rate": 5.023401171775357e-06,
      "loss": 0.9735,
      "step": 3513
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.285883233366866,
      "learning_rate": 5.017997675006161e-06,
      "loss": 0.9278,
      "step": 3514
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9451142793653223,
      "learning_rate": 5.012596112421806e-06,
      "loss": 0.8813,
      "step": 3515
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.099945628768244,
      "learning_rate": 5.007196486119355e-06,
      "loss": 0.9035,
      "step": 3516
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0244735910940357,
      "learning_rate": 5.001798798195136e-06,
      "loss": 0.8841,
      "step": 3517
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.836843892581038,
      "learning_rate": 4.996403050744719e-06,
      "loss": 0.6107,
      "step": 3518
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.230040119650963,
      "learning_rate": 4.991009245862917e-06,
      "loss": 0.9541,
      "step": 3519
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.991223089256402,
      "learning_rate": 4.985617385643789e-06,
      "loss": 0.962,
      "step": 3520
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9937326351971314,
      "learning_rate": 4.980227472180643e-06,
      "loss": 0.9497,
      "step": 3521
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9253183592883293,
      "learning_rate": 4.974839507566027e-06,
      "loss": 0.9129,
      "step": 3522
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0945724195663455,
      "learning_rate": 4.969453493891733e-06,
      "loss": 0.9291,
      "step": 3523
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9959223244453173,
      "learning_rate": 4.9640694332488075e-06,
      "loss": 0.9517,
      "step": 3524
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0944460157150848,
      "learning_rate": 4.958687327727511e-06,
      "loss": 0.9331,
      "step": 3525
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1838865500260192,
      "learning_rate": 4.953307179417376e-06,
      "loss": 0.9445,
      "step": 3526
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.790916964182585,
      "learning_rate": 4.947928990407156e-06,
      "loss": 0.6289,
      "step": 3527
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.127804660473253,
      "learning_rate": 4.94255276278485e-06,
      "loss": 0.9318,
      "step": 3528
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0977684561601109,
      "learning_rate": 4.937178498637696e-06,
      "loss": 1.0089,
      "step": 3529
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9888206706538772,
      "learning_rate": 4.931806200052165e-06,
      "loss": 0.9217,
      "step": 3530
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.025652678132815,
      "learning_rate": 4.926435869113971e-06,
      "loss": 0.8833,
      "step": 3531
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1054022768173428,
      "learning_rate": 4.92106750790806e-06,
      "loss": 0.9245,
      "step": 3532
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6049928626965018,
      "learning_rate": 4.915701118518616e-06,
      "loss": 1.0107,
      "step": 3533
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0005330922343663,
      "learning_rate": 4.910336703029055e-06,
      "loss": 0.8841,
      "step": 3534
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.029729921583388,
      "learning_rate": 4.904974263522025e-06,
      "loss": 0.9086,
      "step": 3535
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1026984289452568,
      "learning_rate": 4.899613802079419e-06,
      "loss": 0.9378,
      "step": 3536
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0437877417237647,
      "learning_rate": 4.8942553207823395e-06,
      "loss": 0.9348,
      "step": 3537
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0388019659091383,
      "learning_rate": 4.888898821711144e-06,
      "loss": 0.9262,
      "step": 3538
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0831330133919737,
      "learning_rate": 4.883544306945407e-06,
      "loss": 0.9139,
      "step": 3539
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.844194948905651,
      "learning_rate": 4.878191778563934e-06,
      "loss": 0.6008,
      "step": 3540
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7498461831334154,
      "learning_rate": 4.872841238644766e-06,
      "loss": 0.6133,
      "step": 3541
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.048922129410617,
      "learning_rate": 4.867492689265154e-06,
      "loss": 0.9174,
      "step": 3542
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9794659474367562,
      "learning_rate": 4.8621461325016015e-06,
      "loss": 0.9679,
      "step": 3543
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9948679662937489,
      "learning_rate": 4.856801570429822e-06,
      "loss": 0.8956,
      "step": 3544
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0450369793784842,
      "learning_rate": 4.851459005124759e-06,
      "loss": 0.8907,
      "step": 3545
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9517503304718755,
      "learning_rate": 4.846118438660578e-06,
      "loss": 0.9007,
      "step": 3546
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0124638035540239,
      "learning_rate": 4.840779873110675e-06,
      "loss": 0.9731,
      "step": 3547
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0644924059234524,
      "learning_rate": 4.83544331054766e-06,
      "loss": 0.8872,
      "step": 3548
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9700019521982034,
      "learning_rate": 4.83010875304337e-06,
      "loss": 0.5852,
      "step": 3549
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7903420361882865,
      "learning_rate": 4.824776202668875e-06,
      "loss": 0.6349,
      "step": 3550
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.150355240952844,
      "learning_rate": 4.819445661494437e-06,
      "loss": 0.9508,
      "step": 3551
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9991386170985649,
      "learning_rate": 4.8141171315895694e-06,
      "loss": 0.9792,
      "step": 3552
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7890126910753315,
      "learning_rate": 4.808790615022987e-06,
      "loss": 0.6482,
      "step": 3553
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2987173599681916,
      "learning_rate": 4.803466113862626e-06,
      "loss": 0.9537,
      "step": 3554
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8001888294960606,
      "learning_rate": 4.798143630175642e-06,
      "loss": 0.6166,
      "step": 3555
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8317960728015974,
      "learning_rate": 4.792823166028405e-06,
      "loss": 0.6259,
      "step": 3556
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9524652347754058,
      "learning_rate": 4.787504723486505e-06,
      "loss": 0.9174,
      "step": 3557
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1163029319920617,
      "learning_rate": 4.7821883046147414e-06,
      "loss": 0.9376,
      "step": 3558
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0230672447289417,
      "learning_rate": 4.776873911477133e-06,
      "loss": 0.9084,
      "step": 3559
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9059246881862605,
      "learning_rate": 4.771561546136908e-06,
      "loss": 0.5935,
      "step": 3560
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0031990241671782,
      "learning_rate": 4.766251210656509e-06,
      "loss": 0.9765,
      "step": 3561
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9397437021482135,
      "learning_rate": 4.760942907097601e-06,
      "loss": 0.8807,
      "step": 3562
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9307886197416508,
      "learning_rate": 4.755636637521035e-06,
      "loss": 0.8207,
      "step": 3563
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9793686793646159,
      "learning_rate": 4.750332403986902e-06,
      "loss": 0.9303,
      "step": 3564
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8843788850137885,
      "learning_rate": 4.7450302085544735e-06,
      "loss": 0.6308,
      "step": 3565
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0699131283188414,
      "learning_rate": 4.739730053282255e-06,
      "loss": 0.91,
      "step": 3566
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0421749223342616,
      "learning_rate": 4.734431940227951e-06,
      "loss": 0.9603,
      "step": 3567
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0665615786536013,
      "learning_rate": 4.7291358714484594e-06,
      "loss": 0.9563,
      "step": 3568
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0890610284997173,
      "learning_rate": 4.723841848999907e-06,
      "loss": 0.9371,
      "step": 3569
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.013186012747761,
      "learning_rate": 4.718549874937612e-06,
      "loss": 0.9268,
      "step": 3570
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8391003429014535,
      "learning_rate": 4.713259951316103e-06,
      "loss": 0.6458,
      "step": 3571
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.076956304272901,
      "learning_rate": 4.707972080189106e-06,
      "loss": 0.9361,
      "step": 3572
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.989558697237506,
      "learning_rate": 4.702686263609559e-06,
      "loss": 0.8606,
      "step": 3573
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1577160504229052,
      "learning_rate": 4.697402503629596e-06,
      "loss": 0.9395,
      "step": 3574
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0981571333328748,
      "learning_rate": 4.69212080230055e-06,
      "loss": 0.9176,
      "step": 3575
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0165748118583033,
      "learning_rate": 4.686841161672974e-06,
      "loss": 0.8796,
      "step": 3576
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0609851784572386,
      "learning_rate": 4.681563583796587e-06,
      "loss": 0.9373,
      "step": 3577
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1665340829253712,
      "learning_rate": 4.67628807072034e-06,
      "loss": 0.9225,
      "step": 3578
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9566851969478088,
      "learning_rate": 4.6710146244923645e-06,
      "loss": 0.902,
      "step": 3579
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9836841622039194,
      "learning_rate": 4.665743247159995e-06,
      "loss": 0.8411,
      "step": 3580
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.118080740222947,
      "learning_rate": 4.660473940769761e-06,
      "loss": 0.9469,
      "step": 3581
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1448448590844913,
      "learning_rate": 4.655206707367388e-06,
      "loss": 0.9461,
      "step": 3582
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0655630759840868,
      "learning_rate": 4.649941548997797e-06,
      "loss": 0.9031,
      "step": 3583
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0403599839433098,
      "learning_rate": 4.644678467705101e-06,
      "loss": 0.9142,
      "step": 3584
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.932384901678091,
      "learning_rate": 4.639417465532622e-06,
      "loss": 0.9427,
      "step": 3585
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.187950778091702,
      "learning_rate": 4.634158544522849e-06,
      "loss": 0.9678,
      "step": 3586
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0057806694848532,
      "learning_rate": 4.628901706717476e-06,
      "loss": 0.8952,
      "step": 3587
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.936478221903647,
      "learning_rate": 4.623646954157399e-06,
      "loss": 0.9477,
      "step": 3588
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1789434819096531,
      "learning_rate": 4.618394288882681e-06,
      "loss": 0.9295,
      "step": 3589
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9721001707277575,
      "learning_rate": 4.613143712932603e-06,
      "loss": 0.8681,
      "step": 3590
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.141701559927908,
      "learning_rate": 4.607895228345603e-06,
      "loss": 0.8873,
      "step": 3591
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9899317906411225,
      "learning_rate": 4.602648837159333e-06,
      "loss": 0.8763,
      "step": 3592
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0111624115975077,
      "learning_rate": 4.597404541410622e-06,
      "loss": 1.0112,
      "step": 3593
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0697748089222494,
      "learning_rate": 4.592162343135483e-06,
      "loss": 0.8952,
      "step": 3594
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9865910650546973,
      "learning_rate": 4.586922244369122e-06,
      "loss": 0.8791,
      "step": 3595
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9935450618996401,
      "learning_rate": 4.5816842471459224e-06,
      "loss": 0.8924,
      "step": 3596
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7969292062433364,
      "learning_rate": 4.576448353499457e-06,
      "loss": 0.5965,
      "step": 3597
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9831125584692775,
      "learning_rate": 4.571214565462477e-06,
      "loss": 0.8501,
      "step": 3598
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9497596797225953,
      "learning_rate": 4.565982885066923e-06,
      "loss": 0.8828,
      "step": 3599
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8152612694972816,
      "learning_rate": 4.560753314343912e-06,
      "loss": 0.5717,
      "step": 3600
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0930551761468486,
      "learning_rate": 4.555525855323738e-06,
      "loss": 0.9056,
      "step": 3601
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0555624229207063,
      "learning_rate": 4.5503005100358945e-06,
      "loss": 0.9138,
      "step": 3602
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0839069222796949,
      "learning_rate": 4.545077280509022e-06,
      "loss": 0.904,
      "step": 3603
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9744735302501725,
      "learning_rate": 4.539856168770974e-06,
      "loss": 0.8772,
      "step": 3604
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.048722020368384,
      "learning_rate": 4.534637176848758e-06,
      "loss": 0.8895,
      "step": 3605
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9279776141376326,
      "learning_rate": 4.52942030676857e-06,
      "loss": 0.6142,
      "step": 3606
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1248910502653306,
      "learning_rate": 4.524205560555774e-06,
      "loss": 0.9614,
      "step": 3607
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0632135790305965,
      "learning_rate": 4.5189929402349175e-06,
      "loss": 0.9303,
      "step": 3608
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0724966141519408,
      "learning_rate": 4.513782447829717e-06,
      "loss": 0.864,
      "step": 3609
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0639787687700764,
      "learning_rate": 4.508574085363065e-06,
      "loss": 0.9233,
      "step": 3610
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0821507920237277,
      "learning_rate": 4.503367854857035e-06,
      "loss": 0.9217,
      "step": 3611
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9633414147124854,
      "learning_rate": 4.498163758332853e-06,
      "loss": 0.6813,
      "step": 3612
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.159988665759171,
      "learning_rate": 4.492961797810932e-06,
      "loss": 0.8779,
      "step": 3613
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1911898400795957,
      "learning_rate": 4.4877619753108605e-06,
      "loss": 0.9562,
      "step": 3614
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0025063222870978,
      "learning_rate": 4.4825642928513746e-06,
      "loss": 0.9073,
      "step": 3615
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8889357544473605,
      "learning_rate": 4.477368752450409e-06,
      "loss": 0.6429,
      "step": 3616
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9856144429896984,
      "learning_rate": 4.472175356125036e-06,
      "loss": 0.9215,
      "step": 3617
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1202351941154194,
      "learning_rate": 4.466984105891521e-06,
      "loss": 0.9103,
      "step": 3618
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0562056908469588,
      "learning_rate": 4.461795003765285e-06,
      "loss": 0.9362,
      "step": 3619
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0464116177865164,
      "learning_rate": 4.456608051760914e-06,
      "loss": 0.9354,
      "step": 3620
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1861487216846736,
      "learning_rate": 4.45142325189216e-06,
      "loss": 0.9336,
      "step": 3621
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0267712428840203,
      "learning_rate": 4.446240606171945e-06,
      "loss": 0.9075,
      "step": 3622
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9927117237539859,
      "learning_rate": 4.4410601166123475e-06,
      "loss": 0.9042,
      "step": 3623
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.020452296079357,
      "learning_rate": 4.4358817852246124e-06,
      "loss": 0.9125,
      "step": 3624
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0155648300797224,
      "learning_rate": 4.430705614019147e-06,
      "loss": 0.9666,
      "step": 3625
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1269053398625677,
      "learning_rate": 4.425531605005519e-06,
      "loss": 0.9463,
      "step": 3626
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.02000514774033,
      "learning_rate": 4.420359760192452e-06,
      "loss": 0.9152,
      "step": 3627
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0663446986771585,
      "learning_rate": 4.4151900815878455e-06,
      "loss": 0.9102,
      "step": 3628
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0194288641758429,
      "learning_rate": 4.410022571198734e-06,
      "loss": 0.9711,
      "step": 3629
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0230691652830086,
      "learning_rate": 4.404857231031332e-06,
      "loss": 0.9088,
      "step": 3630
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9515463906366796,
      "learning_rate": 4.399694063090999e-06,
      "loss": 0.8692,
      "step": 3631
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0093005783938955,
      "learning_rate": 4.394533069382255e-06,
      "loss": 0.9809,
      "step": 3632
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0195043906742483,
      "learning_rate": 4.3893742519087754e-06,
      "loss": 0.9196,
      "step": 3633
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.937425239010968,
      "learning_rate": 4.3842176126733914e-06,
      "loss": 0.8638,
      "step": 3634
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9939428954414576,
      "learning_rate": 4.379063153678087e-06,
      "loss": 0.9342,
      "step": 3635
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0069070008412042,
      "learning_rate": 4.373910876923997e-06,
      "loss": 0.9334,
      "step": 3636
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9427236377993401,
      "learning_rate": 4.368760784411423e-06,
      "loss": 0.9414,
      "step": 3637
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9853364689997033,
      "learning_rate": 4.363612878139799e-06,
      "loss": 0.8649,
      "step": 3638
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9911805821219578,
      "learning_rate": 4.3584671601077224e-06,
      "loss": 0.9303,
      "step": 3639
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9200621237824667,
      "learning_rate": 4.353323632312938e-06,
      "loss": 0.6197,
      "step": 3640
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0355064072017148,
      "learning_rate": 4.348182296752336e-06,
      "loss": 0.8882,
      "step": 3641
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0493572165264673,
      "learning_rate": 4.343043155421971e-06,
      "loss": 0.9095,
      "step": 3642
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9954736187546296,
      "learning_rate": 4.3379062103170214e-06,
      "loss": 0.8474,
      "step": 3643
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1719434795636896,
      "learning_rate": 4.332771463431837e-06,
      "loss": 0.9135,
      "step": 3644
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0105495508878866,
      "learning_rate": 4.327638916759898e-06,
      "loss": 0.9493,
      "step": 3645
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0691278113575382,
      "learning_rate": 4.322508572293836e-06,
      "loss": 0.9463,
      "step": 3646
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.046590448341932,
      "learning_rate": 4.317380432025428e-06,
      "loss": 0.9281,
      "step": 3647
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.187021344897877,
      "learning_rate": 4.312254497945595e-06,
      "loss": 0.9515,
      "step": 3648
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0771351694135487,
      "learning_rate": 4.3071307720444015e-06,
      "loss": 0.888,
      "step": 3649
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1551201020086916,
      "learning_rate": 4.3020092563110485e-06,
      "loss": 0.9003,
      "step": 3650
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0353599750480704,
      "learning_rate": 4.2968899527338984e-06,
      "loss": 0.9379,
      "step": 3651
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0485662626472008,
      "learning_rate": 4.291772863300428e-06,
      "loss": 0.9085,
      "step": 3652
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1983536860378656,
      "learning_rate": 4.2866579899972686e-06,
      "loss": 0.9733,
      "step": 3653
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2023897939736912,
      "learning_rate": 4.281545334810201e-06,
      "loss": 0.958,
      "step": 3654
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.04603214537902,
      "learning_rate": 4.276434899724119e-06,
      "loss": 0.9143,
      "step": 3655
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1033107195987684,
      "learning_rate": 4.27132668672308e-06,
      "loss": 0.8906,
      "step": 3656
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.152079521613587,
      "learning_rate": 4.266220697790266e-06,
      "loss": 0.9389,
      "step": 3657
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9801623140542315,
      "learning_rate": 4.2611169349079985e-06,
      "loss": 0.9127,
      "step": 3658
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4497193165403433,
      "learning_rate": 4.25601540005773e-06,
      "loss": 0.8947,
      "step": 3659
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.024729298135464,
      "learning_rate": 4.250916095220056e-06,
      "loss": 0.9463,
      "step": 3660
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1180702542846763,
      "learning_rate": 4.2458190223747e-06,
      "loss": 0.9734,
      "step": 3661
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9942291132323737,
      "learning_rate": 4.240724183500518e-06,
      "loss": 0.8985,
      "step": 3662
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9783277609650683,
      "learning_rate": 4.2356315805755135e-06,
      "loss": 0.9162,
      "step": 3663
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0182368306238831,
      "learning_rate": 4.230541215576798e-06,
      "loss": 0.8891,
      "step": 3664
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.142607781848587,
      "learning_rate": 4.225453090480631e-06,
      "loss": 0.9273,
      "step": 3665
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9724709071379809,
      "learning_rate": 4.220367207262398e-06,
      "loss": 0.9338,
      "step": 3666
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9437997328989596,
      "learning_rate": 4.21528356789661e-06,
      "loss": 0.8841,
      "step": 3667
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0554046146624947,
      "learning_rate": 4.210202174356922e-06,
      "loss": 0.9142,
      "step": 3668
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0115680138130836,
      "learning_rate": 4.20512302861609e-06,
      "loss": 0.897,
      "step": 3669
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1298041836647084,
      "learning_rate": 4.2000461326460274e-06,
      "loss": 0.9395,
      "step": 3670
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9787764903462035,
      "learning_rate": 4.194971488417753e-06,
      "loss": 0.9288,
      "step": 3671
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.10532113398033,
      "learning_rate": 4.189899097901421e-06,
      "loss": 0.9579,
      "step": 3672
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.630560308483505,
      "learning_rate": 4.184828963066305e-06,
      "loss": 0.3343,
      "step": 3673
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.023465521534386,
      "learning_rate": 4.179761085880809e-06,
      "loss": 0.8995,
      "step": 3674
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9308215547919393,
      "learning_rate": 4.174695468312456e-06,
      "loss": 0.937,
      "step": 3675
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.133463670142457,
      "learning_rate": 4.16963211232789e-06,
      "loss": 0.9318,
      "step": 3676
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0249751654875017,
      "learning_rate": 4.16457101989289e-06,
      "loss": 0.9079,
      "step": 3677
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.127350865921223,
      "learning_rate": 4.159512192972337e-06,
      "loss": 0.9433,
      "step": 3678
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8638025658103662,
      "learning_rate": 4.15445563353024e-06,
      "loss": 0.8795,
      "step": 3679
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1332460073410175,
      "learning_rate": 4.149401343529742e-06,
      "loss": 0.9092,
      "step": 3680
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8854659857992038,
      "learning_rate": 4.144349324933077e-06,
      "loss": 0.8703,
      "step": 3681
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1525469166462685,
      "learning_rate": 4.139299579701623e-06,
      "loss": 0.9414,
      "step": 3682
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0338332873266125,
      "learning_rate": 4.134252109795863e-06,
      "loss": 0.896,
      "step": 3683
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1407183656300752,
      "learning_rate": 4.129206917175397e-06,
      "loss": 0.913,
      "step": 3684
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8273056572889257,
      "learning_rate": 4.124164003798944e-06,
      "loss": 0.6181,
      "step": 3685
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7905389528382277,
      "learning_rate": 4.119123371624335e-06,
      "loss": 0.6207,
      "step": 3686
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.910704935054587,
      "learning_rate": 4.114085022608517e-06,
      "loss": 0.8504,
      "step": 3687
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0094182745864215,
      "learning_rate": 4.109048958707552e-06,
      "loss": 0.8796,
      "step": 3688
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.775935092077598,
      "learning_rate": 4.104015181876613e-06,
      "loss": 0.5825,
      "step": 3689
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1161904120372823,
      "learning_rate": 4.09898369406998e-06,
      "loss": 0.909,
      "step": 3690
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0723432897537302,
      "learning_rate": 4.0939544972410636e-06,
      "loss": 0.9165,
      "step": 3691
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0328758459752982,
      "learning_rate": 4.0889275933423576e-06,
      "loss": 0.8705,
      "step": 3692
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0719108475584813,
      "learning_rate": 4.0839029843254815e-06,
      "loss": 0.8967,
      "step": 3693
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9072302287236059,
      "learning_rate": 4.078880672141171e-06,
      "loss": 0.9499,
      "step": 3694
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9548843558750175,
      "learning_rate": 4.073860658739246e-06,
      "loss": 0.9172,
      "step": 3695
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9261720686006928,
      "learning_rate": 4.068842946068661e-06,
      "loss": 0.895,
      "step": 3696
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0120420435330542,
      "learning_rate": 4.063827536077459e-06,
      "loss": 0.6473,
      "step": 3697
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1238554316091436,
      "learning_rate": 4.058814430712796e-06,
      "loss": 0.9396,
      "step": 3698
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9379008447006095,
      "learning_rate": 4.0538036319209325e-06,
      "loss": 0.9203,
      "step": 3699
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0684989453598317,
      "learning_rate": 4.0487951416472324e-06,
      "loss": 0.9237,
      "step": 3700
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0176243583917381,
      "learning_rate": 4.043788961836164e-06,
      "loss": 0.941,
      "step": 3701
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0077141143815782,
      "learning_rate": 4.038785094431295e-06,
      "loss": 0.9229,
      "step": 3702
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.092938746868086,
      "learning_rate": 4.0337835413753116e-06,
      "loss": 0.9116,
      "step": 3703
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1217487333383087,
      "learning_rate": 4.0287843046099765e-06,
      "loss": 0.9585,
      "step": 3704
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2243467563440447,
      "learning_rate": 4.0237873860761645e-06,
      "loss": 1.0162,
      "step": 3705
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9451833776688563,
      "learning_rate": 4.018792787713865e-06,
      "loss": 0.8544,
      "step": 3706
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0236830737510818,
      "learning_rate": 4.013800511462135e-06,
      "loss": 0.9356,
      "step": 3707
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1403391270401633,
      "learning_rate": 4.008810559259162e-06,
      "loss": 0.92,
      "step": 3708
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8035302715213775,
      "learning_rate": 4.003822933042213e-06,
      "loss": 0.5892,
      "step": 3709
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1093314665577305,
      "learning_rate": 3.998837634747655e-06,
      "loss": 0.9203,
      "step": 3710
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2658443934082553,
      "learning_rate": 3.993854666310955e-06,
      "loss": 0.9074,
      "step": 3711
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0795303795913564,
      "learning_rate": 3.98887402966667e-06,
      "loss": 0.8658,
      "step": 3712
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0174951961153358,
      "learning_rate": 3.983895726748455e-06,
      "loss": 0.9881,
      "step": 3713
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1170595941358847,
      "learning_rate": 3.97891975948906e-06,
      "loss": 0.9587,
      "step": 3714
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0496629481774657,
      "learning_rate": 3.973946129820326e-06,
      "loss": 0.9195,
      "step": 3715
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.112715252564029,
      "learning_rate": 3.968974839673186e-06,
      "loss": 0.896,
      "step": 3716
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.068974692257008,
      "learning_rate": 3.964005890977672e-06,
      "loss": 0.9195,
      "step": 3717
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0010571263619588,
      "learning_rate": 3.9590392856628946e-06,
      "loss": 0.8993,
      "step": 3718
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9835823337221532,
      "learning_rate": 3.954075025657058e-06,
      "loss": 0.9277,
      "step": 3719
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0478178305053736,
      "learning_rate": 3.949113112887471e-06,
      "loss": 0.9243,
      "step": 3720
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9880201810984787,
      "learning_rate": 3.944153549280506e-06,
      "loss": 0.9265,
      "step": 3721
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3472693523710457,
      "learning_rate": 3.939196336761645e-06,
      "loss": 0.9124,
      "step": 3722
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9422397694586887,
      "learning_rate": 3.934241477255445e-06,
      "loss": 0.9464,
      "step": 3723
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0233581344756562,
      "learning_rate": 3.929288972685555e-06,
      "loss": 0.8847,
      "step": 3724
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2179548263059794,
      "learning_rate": 3.924338824974705e-06,
      "loss": 0.9902,
      "step": 3725
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8850528129105575,
      "learning_rate": 3.919391036044715e-06,
      "loss": 0.6589,
      "step": 3726
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1247475364812662,
      "learning_rate": 3.914445607816486e-06,
      "loss": 0.8963,
      "step": 3727
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9534441716931435,
      "learning_rate": 3.909502542210001e-06,
      "loss": 0.8899,
      "step": 3728
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9580527058366938,
      "learning_rate": 3.904561841144338e-06,
      "loss": 0.8734,
      "step": 3729
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9926945346349934,
      "learning_rate": 3.899623506537635e-06,
      "loss": 0.8906,
      "step": 3730
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9653260338644719,
      "learning_rate": 3.894687540307127e-06,
      "loss": 0.871,
      "step": 3731
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.993414289606311,
      "learning_rate": 3.8897539443691355e-06,
      "loss": 0.9512,
      "step": 3732
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8664974938278649,
      "learning_rate": 3.884822720639036e-06,
      "loss": 0.6677,
      "step": 3733
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9854920328965385,
      "learning_rate": 3.879893871031314e-06,
      "loss": 0.9106,
      "step": 3734
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8995843075408165,
      "learning_rate": 3.874967397459511e-06,
      "loss": 0.6304,
      "step": 3735
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2035822102171025,
      "learning_rate": 3.870043301836256e-06,
      "loss": 0.9329,
      "step": 3736
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9843185858985555,
      "learning_rate": 3.86512158607325e-06,
      "loss": 0.917,
      "step": 3737
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1968422068259836,
      "learning_rate": 3.860202252081276e-06,
      "loss": 0.9205,
      "step": 3738
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0868052383570248,
      "learning_rate": 3.855285301770188e-06,
      "loss": 0.8873,
      "step": 3739
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.110935482058188,
      "learning_rate": 3.850370737048913e-06,
      "loss": 0.9181,
      "step": 3740
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0602176503882066,
      "learning_rate": 3.8454585598254565e-06,
      "loss": 0.94,
      "step": 3741
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8909863781383108,
      "learning_rate": 3.840548772006891e-06,
      "loss": 0.6564,
      "step": 3742
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6569405052935211,
      "learning_rate": 3.835641375499375e-06,
      "loss": 0.3279,
      "step": 3743
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0180058006159007,
      "learning_rate": 3.830736372208118e-06,
      "loss": 0.9582,
      "step": 3744
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.057825675445437,
      "learning_rate": 3.8258337640374125e-06,
      "loss": 0.9508,
      "step": 3745
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.032814400408038,
      "learning_rate": 3.820933552890629e-06,
      "loss": 0.8714,
      "step": 3746
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9820473406849715,
      "learning_rate": 3.816035740670185e-06,
      "loss": 0.8601,
      "step": 3747
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1072373092378465,
      "learning_rate": 3.811140329277591e-06,
      "loss": 0.9479,
      "step": 3748
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.113410219242858,
      "learning_rate": 3.8062473206134088e-06,
      "loss": 0.9062,
      "step": 3749
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0154347851414882,
      "learning_rate": 3.8013567165772735e-06,
      "loss": 0.9419,
      "step": 3750
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0626571669197389,
      "learning_rate": 3.7964685190678874e-06,
      "loss": 0.9706,
      "step": 3751
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9762759959041286,
      "learning_rate": 3.7915827299830154e-06,
      "loss": 0.8898,
      "step": 3752
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1367169987281602,
      "learning_rate": 3.7866993512194895e-06,
      "loss": 0.9432,
      "step": 3753
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2080855679910019,
      "learning_rate": 3.7818183846732024e-06,
      "loss": 0.9171,
      "step": 3754
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9440961925955698,
      "learning_rate": 3.776939832239125e-06,
      "loss": 0.9056,
      "step": 3755
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.189058743378041,
      "learning_rate": 3.7720636958112623e-06,
      "loss": 0.8826,
      "step": 3756
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0419419056630954,
      "learning_rate": 3.7671899772827113e-06,
      "loss": 0.9549,
      "step": 3757
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8317486757200215,
      "learning_rate": 3.7623186785456156e-06,
      "loss": 0.6556,
      "step": 3758
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8334735429860796,
      "learning_rate": 3.757449801491172e-06,
      "loss": 0.669,
      "step": 3759
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8357332661957803,
      "learning_rate": 3.7525833480096575e-06,
      "loss": 0.6104,
      "step": 3760
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9554905415410357,
      "learning_rate": 3.7477193199903903e-06,
      "loss": 0.9327,
      "step": 3761
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0582625952812885,
      "learning_rate": 3.7428577193217563e-06,
      "loss": 0.9529,
      "step": 3762
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9477090052316399,
      "learning_rate": 3.737998547891195e-06,
      "loss": 0.8708,
      "step": 3763
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.985222479211933,
      "learning_rate": 3.7331418075852053e-06,
      "loss": 0.9068,
      "step": 3764
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0088579153436183,
      "learning_rate": 3.728287500289339e-06,
      "loss": 0.9105,
      "step": 3765
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0762969625187797,
      "learning_rate": 3.7234356278882076e-06,
      "loss": 0.8761,
      "step": 3766
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1458037933873135,
      "learning_rate": 3.718586192265473e-06,
      "loss": 0.9693,
      "step": 3767
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9545190676381095,
      "learning_rate": 3.7137391953038516e-06,
      "loss": 0.8151,
      "step": 3768
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7381616663019568,
      "learning_rate": 3.7088946388851223e-06,
      "loss": 0.5871,
      "step": 3769
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0212752776000775,
      "learning_rate": 3.7040525248901003e-06,
      "loss": 0.9374,
      "step": 3770
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.976464246575537,
      "learning_rate": 3.6992128551986617e-06,
      "loss": 0.9217,
      "step": 3771
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.738758001975026,
      "learning_rate": 3.6943756316897406e-06,
      "loss": 0.6268,
      "step": 3772
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1022651331436628,
      "learning_rate": 3.6895408562413027e-06,
      "loss": 0.958,
      "step": 3773
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9658350529958529,
      "learning_rate": 3.684708530730382e-06,
      "loss": 0.9403,
      "step": 3774
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9819244121613551,
      "learning_rate": 3.6798786570330526e-06,
      "loss": 0.9216,
      "step": 3775
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9980425671262395,
      "learning_rate": 3.6750512370244363e-06,
      "loss": 0.8898,
      "step": 3776
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0049692701546673,
      "learning_rate": 3.670226272578704e-06,
      "loss": 0.9528,
      "step": 3777
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0041980089842333,
      "learning_rate": 3.6654037655690732e-06,
      "loss": 0.9122,
      "step": 3778
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.017256085173691,
      "learning_rate": 3.660583717867807e-06,
      "loss": 0.9067,
      "step": 3779
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9474084762389943,
      "learning_rate": 3.655766131346211e-06,
      "loss": 0.8974,
      "step": 3780
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0594848283001157,
      "learning_rate": 3.650951007874648e-06,
      "loss": 0.8578,
      "step": 3781
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9505170025704761,
      "learning_rate": 3.6461383493225012e-06,
      "loss": 0.9151,
      "step": 3782
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.069485049106708,
      "learning_rate": 3.6413281575582194e-06,
      "loss": 0.9351,
      "step": 3783
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9922786530211287,
      "learning_rate": 3.6365204344492867e-06,
      "loss": 0.8705,
      "step": 3784
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8207542435599744,
      "learning_rate": 3.6317151818622154e-06,
      "loss": 0.6018,
      "step": 3785
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9959847785676506,
      "learning_rate": 3.62691240166258e-06,
      "loss": 0.9152,
      "step": 3786
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0051819310500028,
      "learning_rate": 3.6221120957149826e-06,
      "loss": 0.9002,
      "step": 3787
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9949584838830868,
      "learning_rate": 3.617314265883066e-06,
      "loss": 0.9219,
      "step": 3788
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1797430902845172,
      "learning_rate": 3.612518914029515e-06,
      "loss": 0.8957,
      "step": 3789
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.969407584131677,
      "learning_rate": 3.6077260420160487e-06,
      "loss": 0.9234,
      "step": 3790
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.07168757816564,
      "learning_rate": 3.602935651703424e-06,
      "loss": 0.9205,
      "step": 3791
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0087737375851307,
      "learning_rate": 3.598147744951438e-06,
      "loss": 0.9082,
      "step": 3792
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1345987796413497,
      "learning_rate": 3.5933623236189198e-06,
      "loss": 0.9685,
      "step": 3793
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0568429099712853,
      "learning_rate": 3.58857938956373e-06,
      "loss": 0.865,
      "step": 3794
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0154531936489202,
      "learning_rate": 3.58379894464278e-06,
      "loss": 0.9036,
      "step": 3795
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9692177442445584,
      "learning_rate": 3.57902099071199e-06,
      "loss": 0.9107,
      "step": 3796
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7724895178128922,
      "learning_rate": 3.5742455296263346e-06,
      "loss": 0.5756,
      "step": 3797
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.234148679655556,
      "learning_rate": 3.569472563239814e-06,
      "loss": 0.8709,
      "step": 3798
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8721308043774094,
      "learning_rate": 3.5647020934054465e-06,
      "loss": 0.9264,
      "step": 3799
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7584965033251975,
      "learning_rate": 3.559934121975304e-06,
      "loss": 0.6148,
      "step": 3800
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.120570394594565,
      "learning_rate": 3.5551686508004735e-06,
      "loss": 0.9686,
      "step": 3801
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0528926394226001,
      "learning_rate": 3.550405681731074e-06,
      "loss": 0.9366,
      "step": 3802
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9274854223111951,
      "learning_rate": 3.5456452166162547e-06,
      "loss": 0.8666,
      "step": 3803
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.056878328468985,
      "learning_rate": 3.540887257304193e-06,
      "loss": 0.9575,
      "step": 3804
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.982843521387432,
      "learning_rate": 3.5361318056420925e-06,
      "loss": 0.8962,
      "step": 3805
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1380672861812804,
      "learning_rate": 3.531378863476178e-06,
      "loss": 0.9595,
      "step": 3806
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0905580926903424,
      "learning_rate": 3.5266284326517165e-06,
      "loss": 0.9024,
      "step": 3807
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1059377992823085,
      "learning_rate": 3.5218805150129755e-06,
      "loss": 0.8945,
      "step": 3808
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0076111678588628,
      "learning_rate": 3.5171351124032703e-06,
      "loss": 0.8849,
      "step": 3809
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0696301364743588,
      "learning_rate": 3.51239222666493e-06,
      "loss": 0.9335,
      "step": 3810
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.060518611374156,
      "learning_rate": 3.507651859639295e-06,
      "loss": 0.8886,
      "step": 3811
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9686009375038918,
      "learning_rate": 3.5029140131667493e-06,
      "loss": 0.9352,
      "step": 3812
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7973260054661627,
      "learning_rate": 3.4981786890866853e-06,
      "loss": 0.5668,
      "step": 3813
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9237817555811254,
      "learning_rate": 3.493445889237518e-06,
      "loss": 0.887,
      "step": 3814
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9051500864463589,
      "learning_rate": 3.4887156154566847e-06,
      "loss": 0.6558,
      "step": 3815
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0574429252662076,
      "learning_rate": 3.4839878695806385e-06,
      "loss": 0.9396,
      "step": 3816
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0498971564576622,
      "learning_rate": 3.4792626534448547e-06,
      "loss": 0.8826,
      "step": 3817
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3125759770479277,
      "learning_rate": 3.4745399688838243e-06,
      "loss": 0.9724,
      "step": 3818
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0100437732122995,
      "learning_rate": 3.469819817731056e-06,
      "loss": 0.8877,
      "step": 3819
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9501715099170336,
      "learning_rate": 3.4651022018190715e-06,
      "loss": 0.8737,
      "step": 3820
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7890695645941431,
      "learning_rate": 3.460387122979423e-06,
      "loss": 0.6374,
      "step": 3821
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9956765872087197,
      "learning_rate": 3.455674583042652e-06,
      "loss": 0.8877,
      "step": 3822
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9896394232698329,
      "learning_rate": 3.4509645838383386e-06,
      "loss": 0.8277,
      "step": 3823
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1143766299311104,
      "learning_rate": 3.4462571271950674e-06,
      "loss": 0.9213,
      "step": 3824
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0300429157469781,
      "learning_rate": 3.4415522149404233e-06,
      "loss": 0.8834,
      "step": 3825
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9599797780825947,
      "learning_rate": 3.436849848901028e-06,
      "loss": 0.9001,
      "step": 3826
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9887656935864608,
      "learning_rate": 3.432150030902497e-06,
      "loss": 0.8823,
      "step": 3827
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1384496660142693,
      "learning_rate": 3.427452762769462e-06,
      "loss": 0.9186,
      "step": 3828
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9846391610792423,
      "learning_rate": 3.4227580463255628e-06,
      "loss": 0.9189,
      "step": 3829
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0536589899153492,
      "learning_rate": 3.4180658833934523e-06,
      "loss": 0.8975,
      "step": 3830
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9248728669099012,
      "learning_rate": 3.4133762757947873e-06,
      "loss": 0.9249,
      "step": 3831
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1711399041758237,
      "learning_rate": 3.4086892253502344e-06,
      "loss": 0.9726,
      "step": 3832
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0936523941033895,
      "learning_rate": 3.4040047338794756e-06,
      "loss": 0.9494,
      "step": 3833
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0288210231245636,
      "learning_rate": 3.3993228032011784e-06,
      "loss": 0.8886,
      "step": 3834
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9738344927731457,
      "learning_rate": 3.3946434351330415e-06,
      "loss": 0.9382,
      "step": 3835
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9746023963342003,
      "learning_rate": 3.3899666314917512e-06,
      "loss": 0.9014,
      "step": 3836
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9866676191266367,
      "learning_rate": 3.385292394093006e-06,
      "loss": 0.9175,
      "step": 3837
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9936892472146279,
      "learning_rate": 3.3806207247515068e-06,
      "loss": 0.8854,
      "step": 3838
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9534123928647845,
      "learning_rate": 3.375951625280948e-06,
      "loss": 0.8645,
      "step": 3839
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0990135137256716,
      "learning_rate": 3.3712850974940437e-06,
      "loss": 0.9397,
      "step": 3840
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9556784977077044,
      "learning_rate": 3.3666211432024974e-06,
      "loss": 0.8418,
      "step": 3841
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9738750878519636,
      "learning_rate": 3.361959764217018e-06,
      "loss": 0.8678,
      "step": 3842
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.916080280667621,
      "learning_rate": 3.357300962347313e-06,
      "loss": 0.8365,
      "step": 3843
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.003955732231916,
      "learning_rate": 3.3526447394020887e-06,
      "loss": 0.9029,
      "step": 3844
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1813712875145088,
      "learning_rate": 3.3479910971890516e-06,
      "loss": 0.9595,
      "step": 3845
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.100969880884694,
      "learning_rate": 3.343340037514903e-06,
      "loss": 0.9136,
      "step": 3846
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9636300034534063,
      "learning_rate": 3.3386915621853533e-06,
      "loss": 0.9089,
      "step": 3847
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0082837008675984,
      "learning_rate": 3.3340456730050887e-06,
      "loss": 0.9277,
      "step": 3848
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9303889661077134,
      "learning_rate": 3.3294023717778122e-06,
      "loss": 0.8769,
      "step": 3849
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0661888593091824,
      "learning_rate": 3.324761660306215e-06,
      "loss": 0.9105,
      "step": 3850
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.074833771469655,
      "learning_rate": 3.3201235403919683e-06,
      "loss": 0.9663,
      "step": 3851
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9898152881188874,
      "learning_rate": 3.3154880138357626e-06,
      "loss": 0.9112,
      "step": 3852
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7140563733400975,
      "learning_rate": 3.3108550824372632e-06,
      "loss": 0.5729,
      "step": 3853
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.028719625600199,
      "learning_rate": 3.306224747995136e-06,
      "loss": 0.8595,
      "step": 3854
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2758210749755423,
      "learning_rate": 3.301597012307034e-06,
      "loss": 0.9568,
      "step": 3855
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0876688726927382,
      "learning_rate": 3.2969718771696047e-06,
      "loss": 0.8686,
      "step": 3856
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1009448411045335,
      "learning_rate": 3.292349344378486e-06,
      "loss": 0.878,
      "step": 3857
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7865235515643659,
      "learning_rate": 3.287729415728298e-06,
      "loss": 0.6056,
      "step": 3858
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8365113754035193,
      "learning_rate": 3.283112093012669e-06,
      "loss": 0.6269,
      "step": 3859
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0758984953229311,
      "learning_rate": 3.278497378024187e-06,
      "loss": 0.8711,
      "step": 3860
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0441722721294493,
      "learning_rate": 3.2738852725544547e-06,
      "loss": 0.8884,
      "step": 3861
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9877377720693093,
      "learning_rate": 3.2692757783940467e-06,
      "loss": 0.9035,
      "step": 3862
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0105278165588294,
      "learning_rate": 3.264668897332527e-06,
      "loss": 0.9293,
      "step": 3863
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1722802595908048,
      "learning_rate": 3.2600646311584494e-06,
      "loss": 0.9849,
      "step": 3864
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2382519873461706,
      "learning_rate": 3.2554629816593375e-06,
      "loss": 0.8833,
      "step": 3865
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9751553883699261,
      "learning_rate": 3.250863950621721e-06,
      "loss": 0.6039,
      "step": 3866
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1198823052949411,
      "learning_rate": 3.2462675398310984e-06,
      "loss": 0.9481,
      "step": 3867
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9468227340427695,
      "learning_rate": 3.241673751071954e-06,
      "loss": 0.8637,
      "step": 3868
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9364933745229933,
      "learning_rate": 3.2370825861277567e-06,
      "loss": 0.8752,
      "step": 3869
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0082868149046436,
      "learning_rate": 3.2324940467809527e-06,
      "loss": 0.8655,
      "step": 3870
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.10060100896276,
      "learning_rate": 3.2279081348129713e-06,
      "loss": 0.9198,
      "step": 3871
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9761752755642812,
      "learning_rate": 3.223324852004219e-06,
      "loss": 0.9332,
      "step": 3872
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1346990661010603,
      "learning_rate": 3.2187442001340942e-06,
      "loss": 0.8828,
      "step": 3873
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8863915189370944,
      "learning_rate": 3.21416618098095e-06,
      "loss": 0.6301,
      "step": 3874
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0557272093339853,
      "learning_rate": 3.2095907963221396e-06,
      "loss": 0.8918,
      "step": 3875
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9827848810369441,
      "learning_rate": 3.2050180479339865e-06,
      "loss": 0.9171,
      "step": 3876
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9034538497681587,
      "learning_rate": 3.2004479375917783e-06,
      "loss": 0.8719,
      "step": 3877
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0614078546756829,
      "learning_rate": 3.1958804670698008e-06,
      "loss": 0.9081,
      "step": 3878
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.995943007563171,
      "learning_rate": 3.191315638141297e-06,
      "loss": 0.848,
      "step": 3879
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1986972975171624,
      "learning_rate": 3.1867534525784937e-06,
      "loss": 0.9912,
      "step": 3880
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9961808166599223,
      "learning_rate": 3.182193912152586e-06,
      "loss": 0.9271,
      "step": 3881
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9768540396627956,
      "learning_rate": 3.177637018633746e-06,
      "loss": 0.943,
      "step": 3882
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2197232273077587,
      "learning_rate": 3.1730827737911163e-06,
      "loss": 0.8729,
      "step": 3883
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0943261923620746,
      "learning_rate": 3.1685311793928077e-06,
      "loss": 0.9326,
      "step": 3884
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2351657026497411,
      "learning_rate": 3.163982237205917e-06,
      "loss": 0.9572,
      "step": 3885
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9771707507366445,
      "learning_rate": 3.1594359489964853e-06,
      "loss": 0.8225,
      "step": 3886
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9983858823539049,
      "learning_rate": 3.15489231652955e-06,
      "loss": 0.9088,
      "step": 3887
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.03818759191714,
      "learning_rate": 3.150351341569101e-06,
      "loss": 0.8886,
      "step": 3888
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9469343139405406,
      "learning_rate": 3.1458130258781006e-06,
      "loss": 0.8776,
      "step": 3889
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9149366152734936,
      "learning_rate": 3.141277371218484e-06,
      "loss": 0.8687,
      "step": 3890
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.149241793078626,
      "learning_rate": 3.136744379351139e-06,
      "loss": 0.9319,
      "step": 3891
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.058080808698075,
      "learning_rate": 3.1322140520359366e-06,
      "loss": 0.8945,
      "step": 3892
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6071368875940729,
      "learning_rate": 3.1276863910317057e-06,
      "loss": 0.3703,
      "step": 3893
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9361805760586414,
      "learning_rate": 3.1231613980962373e-06,
      "loss": 0.8925,
      "step": 3894
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0092726766407343,
      "learning_rate": 3.1186390749862904e-06,
      "loss": 0.8291,
      "step": 3895
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0508918927750366,
      "learning_rate": 3.1141194234575878e-06,
      "loss": 0.9351,
      "step": 3896
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8904972690666436,
      "learning_rate": 3.1096024452648123e-06,
      "loss": 0.6517,
      "step": 3897
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1281564633177008,
      "learning_rate": 3.1050881421616076e-06,
      "loss": 0.9388,
      "step": 3898
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0037324305911184,
      "learning_rate": 3.100576515900591e-06,
      "loss": 0.8921,
      "step": 3899
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2133551364044137,
      "learning_rate": 3.0960675682333186e-06,
      "loss": 0.862,
      "step": 3900
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1294356656943194,
      "learning_rate": 3.0915613009103296e-06,
      "loss": 0.8588,
      "step": 3901
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9946925015828254,
      "learning_rate": 3.0870577156811077e-06,
      "loss": 0.9036,
      "step": 3902
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0267192212356102,
      "learning_rate": 3.0825568142940998e-06,
      "loss": 0.9208,
      "step": 3903
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.011752816357624,
      "learning_rate": 3.0780585984967113e-06,
      "loss": 0.8713,
      "step": 3904
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1652921116438641,
      "learning_rate": 3.073563070035305e-06,
      "loss": 0.8708,
      "step": 3905
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0082421555921282,
      "learning_rate": 3.069070230655198e-06,
      "loss": 0.7874,
      "step": 3906
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0035472085889454,
      "learning_rate": 3.0645800821006667e-06,
      "loss": 0.9218,
      "step": 3907
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9773011269325954,
      "learning_rate": 3.060092626114941e-06,
      "loss": 0.8966,
      "step": 3908
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9647190712284474,
      "learning_rate": 3.0556078644402066e-06,
      "loss": 0.9427,
      "step": 3909
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.040824347416807,
      "learning_rate": 3.051125798817598e-06,
      "loss": 0.9307,
      "step": 3910
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0069417193766308,
      "learning_rate": 3.0466464309872167e-06,
      "loss": 0.9033,
      "step": 3911
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0063428948080149,
      "learning_rate": 3.042169762688096e-06,
      "loss": 0.9191,
      "step": 3912
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.08521746788204,
      "learning_rate": 3.0376957956582452e-06,
      "loss": 0.9066,
      "step": 3913
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1947492280537226,
      "learning_rate": 3.0332245316346e-06,
      "loss": 0.9163,
      "step": 3914
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.399333075371447,
      "learning_rate": 3.0287559723530667e-06,
      "loss": 0.9565,
      "step": 3915
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1050198320835782,
      "learning_rate": 3.024290119548495e-06,
      "loss": 0.9104,
      "step": 3916
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0312518582408594,
      "learning_rate": 3.019826974954674e-06,
      "loss": 0.9339,
      "step": 3917
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0971149617247584,
      "learning_rate": 3.0153665403043586e-06,
      "loss": 0.8922,
      "step": 3918
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9133252760906979,
      "learning_rate": 3.01090881732924e-06,
      "loss": 0.843,
      "step": 3919
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.201294255418934,
      "learning_rate": 3.0064538077599603e-06,
      "loss": 0.9785,
      "step": 3920
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0562600582623671,
      "learning_rate": 3.002001513326107e-06,
      "loss": 0.9392,
      "step": 3921
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0469750958778885,
      "learning_rate": 2.9975519357562155e-06,
      "loss": 0.8679,
      "step": 3922
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9606426248641049,
      "learning_rate": 2.9931050767777626e-06,
      "loss": 0.9039,
      "step": 3923
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.812080710112981,
      "learning_rate": 2.9886609381171703e-06,
      "loss": 0.623,
      "step": 3924
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0784787454192337,
      "learning_rate": 2.984219521499816e-06,
      "loss": 0.8828,
      "step": 3925
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.017710973822789,
      "learning_rate": 2.9797808286499976e-06,
      "loss": 0.931,
      "step": 3926
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8396200200866951,
      "learning_rate": 2.9753448612909775e-06,
      "loss": 0.568,
      "step": 3927
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.050576066015807,
      "learning_rate": 2.9709116211449484e-06,
      "loss": 0.9475,
      "step": 3928
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.059224921287094,
      "learning_rate": 2.966481109933047e-06,
      "loss": 0.9239,
      "step": 3929
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1058950171020763,
      "learning_rate": 2.9620533293753495e-06,
      "loss": 0.9321,
      "step": 3930
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7539721868394745,
      "learning_rate": 2.957628281190873e-06,
      "loss": 0.5849,
      "step": 3931
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9284964169801184,
      "learning_rate": 2.9532059670975732e-06,
      "loss": 0.8619,
      "step": 3932
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0918721636265576,
      "learning_rate": 2.948786388812346e-06,
      "loss": 0.9793,
      "step": 3933
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.044420499037396,
      "learning_rate": 2.9443695480510225e-06,
      "loss": 0.9102,
      "step": 3934
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9758138082625821,
      "learning_rate": 2.9399554465283742e-06,
      "loss": 0.8944,
      "step": 3935
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8449560246519416,
      "learning_rate": 2.935544085958102e-06,
      "loss": 0.5945,
      "step": 3936
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.112353782328085,
      "learning_rate": 2.931135468052858e-06,
      "loss": 0.9614,
      "step": 3937
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9525990152401131,
      "learning_rate": 2.926729594524207e-06,
      "loss": 0.903,
      "step": 3938
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1053282737779286,
      "learning_rate": 2.9223264670826746e-06,
      "loss": 0.9262,
      "step": 3939
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.823255070487115,
      "learning_rate": 2.9179260874376915e-06,
      "loss": 0.6456,
      "step": 3940
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0120301605063848,
      "learning_rate": 2.9135284572976486e-06,
      "loss": 0.966,
      "step": 3941
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9024803298997393,
      "learning_rate": 2.9091335783698517e-06,
      "loss": 0.6325,
      "step": 3942
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.885215238856573,
      "learning_rate": 2.9047414523605467e-06,
      "loss": 0.8742,
      "step": 3943
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0801481857780628,
      "learning_rate": 2.9003520809749053e-06,
      "loss": 0.9294,
      "step": 3944
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9922526470286144,
      "learning_rate": 2.8959654659170354e-06,
      "loss": 0.9095,
      "step": 3945
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0363428724681454,
      "learning_rate": 2.8915816088899696e-06,
      "loss": 0.9446,
      "step": 3946
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0901289215567678,
      "learning_rate": 2.8872005115956746e-06,
      "loss": 0.9099,
      "step": 3947
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.10005765390113,
      "learning_rate": 2.8828221757350406e-06,
      "loss": 0.8906,
      "step": 3948
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0598171458619838,
      "learning_rate": 2.8784466030078905e-06,
      "loss": 0.9087,
      "step": 3949
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.126186864854367,
      "learning_rate": 2.874073795112967e-06,
      "loss": 0.8752,
      "step": 3950
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4309083306937818,
      "learning_rate": 2.8697037537479565e-06,
      "loss": 0.9455,
      "step": 3951
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1221595954607522,
      "learning_rate": 2.8653364806094454e-06,
      "loss": 0.893,
      "step": 3952
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1447684019749573,
      "learning_rate": 2.86097197739297e-06,
      "loss": 0.8666,
      "step": 3953
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0450254143906763,
      "learning_rate": 2.856610245792976e-06,
      "loss": 0.9097,
      "step": 3954
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0351120204968245,
      "learning_rate": 2.8522512875028396e-06,
      "loss": 0.6007,
      "step": 3955
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9603291613706928,
      "learning_rate": 2.847895104214856e-06,
      "loss": 0.872,
      "step": 3956
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0678178497492472,
      "learning_rate": 2.843541697620249e-06,
      "loss": 0.8998,
      "step": 3957
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0452035660158898,
      "learning_rate": 2.8391910694091584e-06,
      "loss": 0.9442,
      "step": 3958
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1025830843764002,
      "learning_rate": 2.8348432212706443e-06,
      "loss": 0.9272,
      "step": 3959
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9626058248949436,
      "learning_rate": 2.8304981548927025e-06,
      "loss": 0.9467,
      "step": 3960
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.80585696930903,
      "learning_rate": 2.826155871962227e-06,
      "loss": 0.6263,
      "step": 3961
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9425773463056282,
      "learning_rate": 2.8218163741650415e-06,
      "loss": 0.9569,
      "step": 3962
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0670947334426784,
      "learning_rate": 2.817479663185898e-06,
      "loss": 0.8456,
      "step": 3963
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.00484330676119,
      "learning_rate": 2.813145740708445e-06,
      "loss": 0.5935,
      "step": 3964
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9509091991239017,
      "learning_rate": 2.808814608415271e-06,
      "loss": 0.9102,
      "step": 3965
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1112947184514526,
      "learning_rate": 2.8044862679878605e-06,
      "loss": 0.916,
      "step": 3966
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1076485277442947,
      "learning_rate": 2.800160721106633e-06,
      "loss": 0.9133,
      "step": 3967
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0543516993180995,
      "learning_rate": 2.7958379694509108e-06,
      "loss": 0.6208,
      "step": 3968
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1157100398187147,
      "learning_rate": 2.791518014698935e-06,
      "loss": 0.9,
      "step": 3969
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0207316009510736,
      "learning_rate": 2.787200858527862e-06,
      "loss": 0.9075,
      "step": 3970
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2786976036738849,
      "learning_rate": 2.7828865026137584e-06,
      "loss": 0.9831,
      "step": 3971
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.062718707800612,
      "learning_rate": 2.7785749486316085e-06,
      "loss": 0.8697,
      "step": 3972
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0537164918206925,
      "learning_rate": 2.774266198255303e-06,
      "loss": 0.9427,
      "step": 3973
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1836089276451687,
      "learning_rate": 2.7699602531576496e-06,
      "loss": 0.9034,
      "step": 3974
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0225081915624923,
      "learning_rate": 2.765657115010364e-06,
      "loss": 0.9,
      "step": 3975
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9726081510939719,
      "learning_rate": 2.7613567854840685e-06,
      "loss": 0.8901,
      "step": 3976
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9985984418442745,
      "learning_rate": 2.7570592662483086e-06,
      "loss": 0.9131,
      "step": 3977
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.970179888820817,
      "learning_rate": 2.752764558971517e-06,
      "loss": 0.9161,
      "step": 3978
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1531895668177528,
      "learning_rate": 2.748472665321056e-06,
      "loss": 0.9576,
      "step": 3979
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0393572172888383,
      "learning_rate": 2.744183586963185e-06,
      "loss": 0.8735,
      "step": 3980
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9466234146453497,
      "learning_rate": 2.739897325563069e-06,
      "loss": 0.8136,
      "step": 3981
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.049050073653042,
      "learning_rate": 2.7356138827847856e-06,
      "loss": 0.9201,
      "step": 3982
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0174619943391627,
      "learning_rate": 2.731333260291311e-06,
      "loss": 0.9019,
      "step": 3983
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0328569507222083,
      "learning_rate": 2.7270554597445343e-06,
      "loss": 0.9087,
      "step": 3984
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.104002411203161,
      "learning_rate": 2.7227804828052384e-06,
      "loss": 0.9176,
      "step": 3985
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1169254165000342,
      "learning_rate": 2.7185083311331283e-06,
      "loss": 0.9646,
      "step": 3986
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0481557413459754,
      "learning_rate": 2.7142390063867896e-06,
      "loss": 0.9552,
      "step": 3987
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0159361845132708,
      "learning_rate": 2.709972510223725e-06,
      "loss": 0.9371,
      "step": 3988
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0578494274879502,
      "learning_rate": 2.7057088443003343e-06,
      "loss": 0.9044,
      "step": 3989
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.135214910350177,
      "learning_rate": 2.7014480102719174e-06,
      "loss": 0.8806,
      "step": 3990
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0569747938010488,
      "learning_rate": 2.697190009792685e-06,
      "loss": 0.9161,
      "step": 3991
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9632051746755175,
      "learning_rate": 2.692934844515729e-06,
      "loss": 0.8515,
      "step": 3992
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1132915630197517,
      "learning_rate": 2.6886825160930587e-06,
      "loss": 0.8784,
      "step": 3993
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0212965650386856,
      "learning_rate": 2.6844330261755715e-06,
      "loss": 0.8662,
      "step": 3994
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9846528572705646,
      "learning_rate": 2.6801863764130653e-06,
      "loss": 0.9092,
      "step": 3995
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0764058102448184,
      "learning_rate": 2.675942568454236e-06,
      "loss": 0.8795,
      "step": 3996
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0338068413597516,
      "learning_rate": 2.671701603946678e-06,
      "loss": 0.8855,
      "step": 3997
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0340163502200446,
      "learning_rate": 2.667463484536876e-06,
      "loss": 0.9163,
      "step": 3998
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3191079501091962,
      "learning_rate": 2.6632282118702147e-06,
      "loss": 0.9381,
      "step": 3999
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0531334760780053,
      "learning_rate": 2.65899578759098e-06,
      "loss": 0.8734,
      "step": 4000
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0736988383027897,
      "learning_rate": 2.654766213342335e-06,
      "loss": 0.9017,
      "step": 4001
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1097917565183246,
      "learning_rate": 2.650539490766346e-06,
      "loss": 0.9174,
      "step": 4002
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8219729084292325,
      "learning_rate": 2.646315621503983e-06,
      "loss": 0.6463,
      "step": 4003
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0038810663738216,
      "learning_rate": 2.642094607195085e-06,
      "loss": 0.8783,
      "step": 4004
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0092217310554168,
      "learning_rate": 2.6378764494784027e-06,
      "loss": 0.9293,
      "step": 4005
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9551296830887883,
      "learning_rate": 2.633661149991569e-06,
      "loss": 0.9058,
      "step": 4006
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0639677542252317,
      "learning_rate": 2.6294487103711064e-06,
      "loss": 0.937,
      "step": 4007
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0672950318077068,
      "learning_rate": 2.6252391322524297e-06,
      "loss": 0.8708,
      "step": 4008
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.145964728122434,
      "learning_rate": 2.6210324172698432e-06,
      "loss": 0.8544,
      "step": 4009
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1235307370529566,
      "learning_rate": 2.6168285670565374e-06,
      "loss": 0.8894,
      "step": 4010
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.966554449252857,
      "learning_rate": 2.6126275832445892e-06,
      "loss": 0.8887,
      "step": 4011
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8508317597097584,
      "learning_rate": 2.6084294674649734e-06,
      "loss": 0.6559,
      "step": 4012
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8260534518107675,
      "learning_rate": 2.6042342213475346e-06,
      "loss": 0.6192,
      "step": 4013
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0353077986732389,
      "learning_rate": 2.6000418465210143e-06,
      "loss": 0.919,
      "step": 4014
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1214853784004595,
      "learning_rate": 2.595852344613038e-06,
      "loss": 0.9024,
      "step": 4015
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9028249109007008,
      "learning_rate": 2.5916657172501103e-06,
      "loss": 0.9158,
      "step": 4016
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1074009658716633,
      "learning_rate": 2.587481966057633e-06,
      "loss": 0.943,
      "step": 4017
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0672105075768532,
      "learning_rate": 2.583301092659872e-06,
      "loss": 0.9328,
      "step": 4018
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0144775114022035,
      "learning_rate": 2.5791230986799944e-06,
      "loss": 0.8632,
      "step": 4019
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9582198971045599,
      "learning_rate": 2.5749479857400383e-06,
      "loss": 0.8884,
      "step": 4020
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0263473104345207,
      "learning_rate": 2.5707757554609247e-06,
      "loss": 0.8895,
      "step": 4021
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9406600353884168,
      "learning_rate": 2.56660640946246e-06,
      "loss": 0.9342,
      "step": 4022
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0482436492665226,
      "learning_rate": 2.5624399493633257e-06,
      "loss": 0.903,
      "step": 4023
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8975162145331986,
      "learning_rate": 2.558276376781086e-06,
      "loss": 0.8708,
      "step": 4024
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0627690721352712,
      "learning_rate": 2.55411569333218e-06,
      "loss": 0.9525,
      "step": 4025
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0611805307147912,
      "learning_rate": 2.5499579006319365e-06,
      "loss": 0.9204,
      "step": 4026
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0444607057926898,
      "learning_rate": 2.5458030002945457e-06,
      "loss": 0.9232,
      "step": 4027
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.013881962473236,
      "learning_rate": 2.5416509939330836e-06,
      "loss": 0.8992,
      "step": 4028
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9433273992193668,
      "learning_rate": 2.537501883159509e-06,
      "loss": 0.8934,
      "step": 4029
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.803606827297757,
      "learning_rate": 2.5333556695846384e-06,
      "loss": 0.615,
      "step": 4030
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.136139458469156,
      "learning_rate": 2.5292123548181847e-06,
      "loss": 0.9187,
      "step": 4031
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8934634642859859,
      "learning_rate": 2.525071940468722e-06,
      "loss": 0.6348,
      "step": 4032
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.038010858837895,
      "learning_rate": 2.520934428143701e-06,
      "loss": 0.9471,
      "step": 4033
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0311586596260032,
      "learning_rate": 2.5167998194494468e-06,
      "loss": 0.9033,
      "step": 4034
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0074174143763444,
      "learning_rate": 2.5126681159911558e-06,
      "loss": 0.8611,
      "step": 4035
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0851157285726571,
      "learning_rate": 2.5085393193729e-06,
      "loss": 0.877,
      "step": 4036
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9840609027127136,
      "learning_rate": 2.5044134311976156e-06,
      "loss": 0.8882,
      "step": 4037
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1226128740946901,
      "learning_rate": 2.5002904530671236e-06,
      "loss": 0.8703,
      "step": 4038
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0562388915491452,
      "learning_rate": 2.4961703865820974e-06,
      "loss": 0.9498,
      "step": 4039
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.799778219252,
      "learning_rate": 2.492053233342091e-06,
      "loss": 0.6013,
      "step": 4040
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9806664346409082,
      "learning_rate": 2.487938994945527e-06,
      "loss": 0.5966,
      "step": 4041
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0167632482832503,
      "learning_rate": 2.4838276729896884e-06,
      "loss": 0.8964,
      "step": 4042
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.011357261724962,
      "learning_rate": 2.479719269070743e-06,
      "loss": 0.8977,
      "step": 4043
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7325678667162047,
      "learning_rate": 2.4756137847837025e-06,
      "loss": 0.5943,
      "step": 4044
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.15816334148304,
      "learning_rate": 2.4715112217224657e-06,
      "loss": 0.9421,
      "step": 4045
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0310306286394006,
      "learning_rate": 2.467411581479786e-06,
      "loss": 0.8712,
      "step": 4046
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8064712898414313,
      "learning_rate": 2.463314865647286e-06,
      "loss": 0.6469,
      "step": 4047
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.047430953816197,
      "learning_rate": 2.45922107581545e-06,
      "loss": 0.9436,
      "step": 4048
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8950855850666907,
      "learning_rate": 2.4551302135736287e-06,
      "loss": 0.8446,
      "step": 4049
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.111872611512001,
      "learning_rate": 2.4510422805100366e-06,
      "loss": 0.8887,
      "step": 4050
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1246621457392976,
      "learning_rate": 2.446957278211746e-06,
      "loss": 0.9212,
      "step": 4051
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.008026103644824,
      "learning_rate": 2.4428752082647044e-06,
      "loss": 0.9711,
      "step": 4052
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2277222760660578,
      "learning_rate": 2.438796072253704e-06,
      "loss": 0.9402,
      "step": 4053
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9976702908604783,
      "learning_rate": 2.4347198717624054e-06,
      "loss": 0.9445,
      "step": 4054
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0012294105685078,
      "learning_rate": 2.4306466083733392e-06,
      "loss": 0.6431,
      "step": 4055
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9598375027154555,
      "learning_rate": 2.426576283667873e-06,
      "loss": 0.6203,
      "step": 4056
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0227095663491719,
      "learning_rate": 2.422508899226258e-06,
      "loss": 0.8741,
      "step": 4057
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1085562488619247,
      "learning_rate": 2.418444456627589e-06,
      "loss": 0.897,
      "step": 4058
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8907021100403844,
      "learning_rate": 2.4143829574498224e-06,
      "loss": 0.5912,
      "step": 4059
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0735195517848144,
      "learning_rate": 2.4103244032697717e-06,
      "loss": 0.9452,
      "step": 4060
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9683926874172935,
      "learning_rate": 2.406268795663108e-06,
      "loss": 0.9405,
      "step": 4061
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.193675740273384,
      "learning_rate": 2.4022161362043574e-06,
      "loss": 0.8904,
      "step": 4062
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9902374847822246,
      "learning_rate": 2.3981664264669025e-06,
      "loss": 0.8885,
      "step": 4063
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7786225895006489,
      "learning_rate": 2.3941196680229794e-06,
      "loss": 0.6498,
      "step": 4064
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0366870421842584,
      "learning_rate": 2.3900758624436772e-06,
      "loss": 0.8927,
      "step": 4065
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.952445988500816,
      "learning_rate": 2.3860350112989473e-06,
      "loss": 0.8927,
      "step": 4066
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9087421735124158,
      "learning_rate": 2.3819971161575807e-06,
      "loss": 0.8952,
      "step": 4067
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0981186585539726,
      "learning_rate": 2.3779621785872252e-06,
      "loss": 0.9743,
      "step": 4068
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9725396824218162,
      "learning_rate": 2.3739302001543918e-06,
      "loss": 0.8752,
      "step": 4069
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0260843110380136,
      "learning_rate": 2.3699011824244234e-06,
      "loss": 0.8949,
      "step": 4070
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0993382240720018,
      "learning_rate": 2.365875126961531e-06,
      "loss": 0.8505,
      "step": 4071
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0921594433794624,
      "learning_rate": 2.3618520353287644e-06,
      "loss": 0.951,
      "step": 4072
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8477034780863942,
      "learning_rate": 2.3578319090880263e-06,
      "loss": 0.6245,
      "step": 4073
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9085914245753653,
      "learning_rate": 2.3538147498000695e-06,
      "loss": 0.8754,
      "step": 4074
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.038492629873231,
      "learning_rate": 2.349800559024492e-06,
      "loss": 0.9003,
      "step": 4075
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1306069708834987,
      "learning_rate": 2.3457893383197415e-06,
      "loss": 0.9306,
      "step": 4076
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0019827851335459,
      "learning_rate": 2.3417810892431104e-06,
      "loss": 0.8654,
      "step": 4077
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9973313952329481,
      "learning_rate": 2.3377758133507455e-06,
      "loss": 0.8509,
      "step": 4078
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8564846751281183,
      "learning_rate": 2.3337735121976247e-06,
      "loss": 0.5978,
      "step": 4079
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9303736259785956,
      "learning_rate": 2.32977418733758e-06,
      "loss": 0.5964,
      "step": 4080
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9617360887138171,
      "learning_rate": 2.3257778403232954e-06,
      "loss": 0.9266,
      "step": 4081
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9813655190982628,
      "learning_rate": 2.321784472706279e-06,
      "loss": 0.9231,
      "step": 4082
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1156573879592169,
      "learning_rate": 2.317794086036901e-06,
      "loss": 0.8982,
      "step": 4083
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9637796800442201,
      "learning_rate": 2.3138066818643647e-06,
      "loss": 0.9074,
      "step": 4084
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9566582641420079,
      "learning_rate": 2.3098222617367184e-06,
      "loss": 0.8721,
      "step": 4085
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1129364251369354,
      "learning_rate": 2.30584082720085e-06,
      "loss": 0.8945,
      "step": 4086
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0825484008062467,
      "learning_rate": 2.301862379802492e-06,
      "loss": 0.8948,
      "step": 4087
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7523383773060265,
      "learning_rate": 2.297886921086211e-06,
      "loss": 0.5937,
      "step": 4088
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9609032474956132,
      "learning_rate": 2.2939144525954194e-06,
      "loss": 0.9146,
      "step": 4089
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9841364820391175,
      "learning_rate": 2.2899449758723657e-06,
      "loss": 0.906,
      "step": 4090
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0866671255551803,
      "learning_rate": 2.285978492458134e-06,
      "loss": 0.8848,
      "step": 4091
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9329027971855751,
      "learning_rate": 2.282015003892659e-06,
      "loss": 0.9057,
      "step": 4092
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0771504530321494,
      "learning_rate": 2.2780545117146947e-06,
      "loss": 0.8677,
      "step": 4093
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9405442421357277,
      "learning_rate": 2.2740970174618405e-06,
      "loss": 0.8876,
      "step": 4094
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9011362137371268,
      "learning_rate": 2.270142522670541e-06,
      "loss": 0.8878,
      "step": 4095
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0985905342444267,
      "learning_rate": 2.2661910288760545e-06,
      "loss": 0.9058,
      "step": 4096
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0559711878519156,
      "learning_rate": 2.262242537612497e-06,
      "loss": 0.9068,
      "step": 4097
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0771553502116071,
      "learning_rate": 2.258297050412804e-06,
      "loss": 0.9086,
      "step": 4098
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0589913866508545,
      "learning_rate": 2.254354568808752e-06,
      "loss": 0.9368,
      "step": 4099
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9713957103222209,
      "learning_rate": 2.2504150943309455e-06,
      "loss": 0.8747,
      "step": 4100
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1643275492817091,
      "learning_rate": 2.246478628508827e-06,
      "loss": 0.9688,
      "step": 4101
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1216001781096634,
      "learning_rate": 2.242545172870665e-06,
      "loss": 0.9027,
      "step": 4102
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8769578577818737,
      "learning_rate": 2.238614728943561e-06,
      "loss": 0.6278,
      "step": 4103
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9306260041874101,
      "learning_rate": 2.2346872982534584e-06,
      "loss": 0.8237,
      "step": 4104
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9447915884785565,
      "learning_rate": 2.2307628823251083e-06,
      "loss": 0.9096,
      "step": 4105
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9741022549148441,
      "learning_rate": 2.2268414826821117e-06,
      "loss": 0.857,
      "step": 4106
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0530262844926228,
      "learning_rate": 2.222923100846893e-06,
      "loss": 0.8973,
      "step": 4107
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0850271250710877,
      "learning_rate": 2.2190077383406938e-06,
      "loss": 0.9084,
      "step": 4108
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8902981465787057,
      "learning_rate": 2.2150953966835996e-06,
      "loss": 0.6297,
      "step": 4109
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0794604597886779,
      "learning_rate": 2.211186077394516e-06,
      "loss": 0.9195,
      "step": 4110
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9891031292365894,
      "learning_rate": 2.207279781991173e-06,
      "loss": 0.8497,
      "step": 4111
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9169414335338597,
      "learning_rate": 2.2033765119901294e-06,
      "loss": 0.9045,
      "step": 4112
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0780029003137692,
      "learning_rate": 2.1994762689067705e-06,
      "loss": 0.8469,
      "step": 4113
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9069818071848453,
      "learning_rate": 2.1955790542553036e-06,
      "loss": 0.6278,
      "step": 4114
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3226437297474625,
      "learning_rate": 2.1916848695487615e-06,
      "loss": 0.89,
      "step": 4115
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9734341738972776,
      "learning_rate": 2.1877937162990015e-06,
      "loss": 0.8726,
      "step": 4116
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0564847952893994,
      "learning_rate": 2.1839055960167e-06,
      "loss": 0.8621,
      "step": 4117
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.084254840989928,
      "learning_rate": 2.180020510211367e-06,
      "loss": 0.9451,
      "step": 4118
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0410346562648143,
      "learning_rate": 2.1761384603913203e-06,
      "loss": 0.9128,
      "step": 4119
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8919928133341943,
      "learning_rate": 2.172259448063704e-06,
      "loss": 0.6456,
      "step": 4120
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0351653185258258,
      "learning_rate": 2.1683834747344913e-06,
      "loss": 0.9147,
      "step": 4121
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0266485001119963,
      "learning_rate": 2.1645105419084587e-06,
      "loss": 0.8881,
      "step": 4122
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.97467964240133,
      "learning_rate": 2.160640651089221e-06,
      "loss": 0.8722,
      "step": 4123
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8983554333543877,
      "learning_rate": 2.1567738037791998e-06,
      "loss": 0.6178,
      "step": 4124
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0334687871694925,
      "learning_rate": 2.152910001479638e-06,
      "loss": 0.8977,
      "step": 4125
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9392981691410256,
      "learning_rate": 2.1490492456905964e-06,
      "loss": 0.8388,
      "step": 4126
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9847314507116718,
      "learning_rate": 2.1451915379109546e-06,
      "loss": 0.8646,
      "step": 4127
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0146058744162487,
      "learning_rate": 2.141336879638406e-06,
      "loss": 0.8798,
      "step": 4128
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0653693780210893,
      "learning_rate": 2.1374852723694595e-06,
      "loss": 0.9389,
      "step": 4129
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1264542553110117,
      "learning_rate": 2.133636717599451e-06,
      "loss": 0.8755,
      "step": 4130
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9961639260367643,
      "learning_rate": 2.1297912168225086e-06,
      "loss": 0.8671,
      "step": 4131
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1165200368259363,
      "learning_rate": 2.1259487715316e-06,
      "loss": 0.8906,
      "step": 4132
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9084039428917705,
      "learning_rate": 2.1221093832184903e-06,
      "loss": 0.7086,
      "step": 4133
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9082143986679057,
      "learning_rate": 2.118273053373757e-06,
      "loss": 0.9378,
      "step": 4134
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1445467930770155,
      "learning_rate": 2.1144397834868034e-06,
      "loss": 0.6216,
      "step": 4135
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9839345215195816,
      "learning_rate": 2.1106095750458332e-06,
      "loss": 0.8979,
      "step": 4136
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0566093817884863,
      "learning_rate": 2.106782429537866e-06,
      "loss": 0.9103,
      "step": 4137
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.825253975069698,
      "learning_rate": 2.1029583484487315e-06,
      "loss": 0.6143,
      "step": 4138
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1113740259137597,
      "learning_rate": 2.0991373332630683e-06,
      "loss": 0.9419,
      "step": 4139
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1770584790769236,
      "learning_rate": 2.0953193854643274e-06,
      "loss": 0.8142,
      "step": 4140
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.165379921028142,
      "learning_rate": 2.0915045065347673e-06,
      "loss": 0.8705,
      "step": 4141
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1085518415727649,
      "learning_rate": 2.0876926979554545e-06,
      "loss": 0.9681,
      "step": 4142
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1523772917731279,
      "learning_rate": 2.0838839612062633e-06,
      "loss": 0.8647,
      "step": 4143
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0048575184959407,
      "learning_rate": 2.080078297765884e-06,
      "loss": 0.9038,
      "step": 4144
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.034802827016932,
      "learning_rate": 2.0762757091117937e-06,
      "loss": 0.903,
      "step": 4145
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1200863837909616,
      "learning_rate": 2.0724761967202987e-06,
      "loss": 0.9377,
      "step": 4146
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9900599068279656,
      "learning_rate": 2.0686797620664987e-06,
      "loss": 0.9094,
      "step": 4147
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0449735613098885,
      "learning_rate": 2.0648864066242937e-06,
      "loss": 0.8697,
      "step": 4148
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9722907505164504,
      "learning_rate": 2.0610961318664013e-06,
      "loss": 0.89,
      "step": 4149
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9820997055211688,
      "learning_rate": 2.0573089392643362e-06,
      "loss": 0.6639,
      "step": 4150
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2487324488662839,
      "learning_rate": 2.0535248302884147e-06,
      "loss": 0.9232,
      "step": 4151
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7711366345818977,
      "learning_rate": 2.0497438064077603e-06,
      "loss": 0.592,
      "step": 4152
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.981690360075734,
      "learning_rate": 2.045965869090295e-06,
      "loss": 0.9306,
      "step": 4153
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9895404404177346,
      "learning_rate": 2.0421910198027452e-06,
      "loss": 0.8985,
      "step": 4154
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9734184749969119,
      "learning_rate": 2.0384192600106335e-06,
      "loss": 0.879,
      "step": 4155
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.002104452089228,
      "learning_rate": 2.0346505911782956e-06,
      "loss": 0.8795,
      "step": 4156
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9590907037915993,
      "learning_rate": 2.0308850147688484e-06,
      "loss": 0.9441,
      "step": 4157
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9914230614156283,
      "learning_rate": 2.0271225322442255e-06,
      "loss": 0.8635,
      "step": 4158
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9950797389342997,
      "learning_rate": 2.0233631450651525e-06,
      "loss": 0.9038,
      "step": 4159
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9481769086523646,
      "learning_rate": 2.019606854691145e-06,
      "loss": 0.9077,
      "step": 4160
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.120908050471793,
      "learning_rate": 2.0158536625805325e-06,
      "loss": 0.8922,
      "step": 4161
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0940712533618069,
      "learning_rate": 2.01210357019043e-06,
      "loss": 0.9213,
      "step": 4162
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.071609866758406,
      "learning_rate": 2.008356578976752e-06,
      "loss": 0.8891,
      "step": 4163
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9758110543201206,
      "learning_rate": 2.004612690394212e-06,
      "loss": 0.8884,
      "step": 4164
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9985475416977603,
      "learning_rate": 2.0008719058963144e-06,
      "loss": 0.9499,
      "step": 4165
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1927158642740023,
      "learning_rate": 1.997134226935361e-06,
      "loss": 0.9548,
      "step": 4166
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9807381963800587,
      "learning_rate": 1.9933996549624468e-06,
      "loss": 0.8856,
      "step": 4167
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1307480996979833,
      "learning_rate": 1.9896681914274616e-06,
      "loss": 0.9362,
      "step": 4168
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0672343198569991,
      "learning_rate": 1.9859398377790872e-06,
      "loss": 0.8802,
      "step": 4169
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9774952858679483,
      "learning_rate": 1.982214595464804e-06,
      "loss": 0.9053,
      "step": 4170
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9784836721057297,
      "learning_rate": 1.97849246593087e-06,
      "loss": 0.9341,
      "step": 4171
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.199848137639609,
      "learning_rate": 1.9747734506223525e-06,
      "loss": 0.9659,
      "step": 4172
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0779383957217863,
      "learning_rate": 1.9710575509831008e-06,
      "loss": 0.9431,
      "step": 4173
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9371731011852148,
      "learning_rate": 1.967344768455747e-06,
      "loss": 0.9102,
      "step": 4174
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.068716218429112,
      "learning_rate": 1.9636351044817292e-06,
      "loss": 0.9382,
      "step": 4175
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.79247589868014,
      "learning_rate": 1.9599285605012643e-06,
      "loss": 0.6277,
      "step": 4176
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9875008060959048,
      "learning_rate": 1.9562251379533593e-06,
      "loss": 0.9003,
      "step": 4177
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.116122308184402,
      "learning_rate": 1.952524838275811e-06,
      "loss": 0.9171,
      "step": 4178
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0177513291364881,
      "learning_rate": 1.9488276629052026e-06,
      "loss": 0.9105,
      "step": 4179
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0528233452721614,
      "learning_rate": 1.945133613276907e-06,
      "loss": 0.9206,
      "step": 4180
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9149093047385927,
      "learning_rate": 1.941442690825076e-06,
      "loss": 0.8708,
      "step": 4181
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1851763986975772,
      "learning_rate": 1.937754896982663e-06,
      "loss": 0.9179,
      "step": 4182
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0329015279003593,
      "learning_rate": 1.9340702331813842e-06,
      "loss": 0.8408,
      "step": 4183
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1073631501479901,
      "learning_rate": 1.9303887008517618e-06,
      "loss": 0.8999,
      "step": 4184
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9882192778381163,
      "learning_rate": 1.9267103014230935e-06,
      "loss": 0.8953,
      "step": 4185
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0136356595355989,
      "learning_rate": 1.923035036323452e-06,
      "loss": 0.9266,
      "step": 4186
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1230740224625415,
      "learning_rate": 1.91936290697971e-06,
      "loss": 0.9034,
      "step": 4187
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.873142261222719,
      "learning_rate": 1.9156939148175125e-06,
      "loss": 0.6321,
      "step": 4188
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.090335578481334,
      "learning_rate": 1.9120280612612873e-06,
      "loss": 0.9041,
      "step": 4189
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.046874604229532,
      "learning_rate": 1.9083653477342467e-06,
      "loss": 0.9263,
      "step": 4190
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8984609157682861,
      "learning_rate": 1.904705775658381e-06,
      "loss": 0.8525,
      "step": 4191
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1148903734991216,
      "learning_rate": 1.9010493464544621e-06,
      "loss": 0.8958,
      "step": 4192
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0558973978845838,
      "learning_rate": 1.8973960615420416e-06,
      "loss": 0.877,
      "step": 4193
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8503600490143437,
      "learning_rate": 1.8937459223394517e-06,
      "loss": 0.5806,
      "step": 4194
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.002031967688926,
      "learning_rate": 1.8900989302637985e-06,
      "loss": 0.9285,
      "step": 4195
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7550359468229526,
      "learning_rate": 1.8864550867309771e-06,
      "loss": 0.5881,
      "step": 4196
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1207503587680443,
      "learning_rate": 1.8828143931556442e-06,
      "loss": 0.9104,
      "step": 4197
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1388520436448646,
      "learning_rate": 1.8791768509512487e-06,
      "loss": 0.869,
      "step": 4198
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.037286949535774,
      "learning_rate": 1.875542461530011e-06,
      "loss": 0.8503,
      "step": 4199
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0708562133851207,
      "learning_rate": 1.871911226302917e-06,
      "loss": 0.9306,
      "step": 4200
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1432934425177512,
      "learning_rate": 1.868283146679747e-06,
      "loss": 0.948,
      "step": 4201
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7985551409642415,
      "learning_rate": 1.8646582240690414e-06,
      "loss": 0.5997,
      "step": 4202
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8348765884146081,
      "learning_rate": 1.8610364598781227e-06,
      "loss": 0.6486,
      "step": 4203
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.025269031354228,
      "learning_rate": 1.8574178555130818e-06,
      "loss": 0.8941,
      "step": 4204
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9841691293375314,
      "learning_rate": 1.8538024123787868e-06,
      "loss": 0.8999,
      "step": 4205
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0824634892811242,
      "learning_rate": 1.8501901318788773e-06,
      "loss": 0.9079,
      "step": 4206
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9060121439410255,
      "learning_rate": 1.8465810154157626e-06,
      "loss": 0.9287,
      "step": 4207
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0097076410435295,
      "learning_rate": 1.8429750643906331e-06,
      "loss": 0.9134,
      "step": 4208
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0906369733759473,
      "learning_rate": 1.8393722802034331e-06,
      "loss": 0.8755,
      "step": 4209
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0393505488165178,
      "learning_rate": 1.835772664252895e-06,
      "loss": 0.9061,
      "step": 4210
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0278748419537818,
      "learning_rate": 1.832176217936511e-06,
      "loss": 0.8922,
      "step": 4211
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9650711164987777,
      "learning_rate": 1.8285829426505453e-06,
      "loss": 0.9006,
      "step": 4212
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9920955283517207,
      "learning_rate": 1.8249928397900351e-06,
      "loss": 0.9613,
      "step": 4213
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0838969397292793,
      "learning_rate": 1.8214059107487726e-06,
      "loss": 0.902,
      "step": 4214
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.004397923116562,
      "learning_rate": 1.8178221569193343e-06,
      "loss": 0.8961,
      "step": 4215
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0156747229261185,
      "learning_rate": 1.8142415796930568e-06,
      "loss": 0.8963,
      "step": 4216
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0194436543568717,
      "learning_rate": 1.8106641804600411e-06,
      "loss": 0.9326,
      "step": 4217
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0658157865831197,
      "learning_rate": 1.8070899606091586e-06,
      "loss": 0.9668,
      "step": 4218
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1602370307474692,
      "learning_rate": 1.8035189215280423e-06,
      "loss": 0.8934,
      "step": 4219
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0566761732291547,
      "learning_rate": 1.799951064603095e-06,
      "loss": 0.9594,
      "step": 4220
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9404711220279678,
      "learning_rate": 1.7963863912194768e-06,
      "loss": 0.8082,
      "step": 4221
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0567816214526264,
      "learning_rate": 1.7928249027611255e-06,
      "loss": 0.8771,
      "step": 4222
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9665747741417998,
      "learning_rate": 1.789266600610724e-06,
      "loss": 0.9024,
      "step": 4223
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0453512298995702,
      "learning_rate": 1.7857114861497337e-06,
      "loss": 0.9321,
      "step": 4224
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.957613490550938,
      "learning_rate": 1.782159560758373e-06,
      "loss": 0.6292,
      "step": 4225
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0804801284624643,
      "learning_rate": 1.7786108258156154e-06,
      "loss": 0.8469,
      "step": 4226
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9814347760028723,
      "learning_rate": 1.7750652826992077e-06,
      "loss": 0.8817,
      "step": 4227
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.11913329100411,
      "learning_rate": 1.7715229327856498e-06,
      "loss": 0.9603,
      "step": 4228
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1470619074977793,
      "learning_rate": 1.7679837774502052e-06,
      "loss": 0.9238,
      "step": 4229
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0677655770324916,
      "learning_rate": 1.7644478180668945e-06,
      "loss": 0.8196,
      "step": 4230
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.014288945301486,
      "learning_rate": 1.7609150560084986e-06,
      "loss": 0.8794,
      "step": 4231
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0366767049922103,
      "learning_rate": 1.7573854926465582e-06,
      "loss": 0.9267,
      "step": 4232
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9605855903137765,
      "learning_rate": 1.7538591293513685e-06,
      "loss": 0.8726,
      "step": 4233
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0900384092901418,
      "learning_rate": 1.7503359674919929e-06,
      "loss": 0.924,
      "step": 4234
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2527981774209505,
      "learning_rate": 1.746816008436234e-06,
      "loss": 0.8922,
      "step": 4235
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9573783386219634,
      "learning_rate": 1.7432992535506687e-06,
      "loss": 0.9218,
      "step": 4236
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.167074465544599,
      "learning_rate": 1.7397857042006194e-06,
      "loss": 0.8913,
      "step": 4237
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9181086116629613,
      "learning_rate": 1.736275361750167e-06,
      "loss": 0.8621,
      "step": 4238
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9571814832714667,
      "learning_rate": 1.7327682275621506e-06,
      "loss": 0.9162,
      "step": 4239
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0426815621822354,
      "learning_rate": 1.7292643029981525e-06,
      "loss": 0.8781,
      "step": 4240
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0766923873644239,
      "learning_rate": 1.7257635894185232e-06,
      "loss": 0.8771,
      "step": 4241
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.03936576372661,
      "learning_rate": 1.7222660881823594e-06,
      "loss": 0.9205,
      "step": 4242
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0988013602349083,
      "learning_rate": 1.7187718006475117e-06,
      "loss": 0.9075,
      "step": 4243
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0841287997835765,
      "learning_rate": 1.7152807281705809e-06,
      "loss": 0.9343,
      "step": 4244
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0096350314251343,
      "learning_rate": 1.7117928721069233e-06,
      "loss": 0.9107,
      "step": 4245
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1789788107301185,
      "learning_rate": 1.708308233810644e-06,
      "loss": 0.9166,
      "step": 4246
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2687666235457793,
      "learning_rate": 1.704826814634597e-06,
      "loss": 0.8794,
      "step": 4247
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1265747517301148,
      "learning_rate": 1.701348615930397e-06,
      "loss": 0.9149,
      "step": 4248
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.157170827545514,
      "learning_rate": 1.6978736390483896e-06,
      "loss": 0.9444,
      "step": 4249
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1784516419107398,
      "learning_rate": 1.6944018853376898e-06,
      "loss": 0.9307,
      "step": 4250
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0593431054480849,
      "learning_rate": 1.6909333561461471e-06,
      "loss": 0.9414,
      "step": 4251
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7834574942279199,
      "learning_rate": 1.6874680528203657e-06,
      "loss": 0.5987,
      "step": 4252
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0761652506427115,
      "learning_rate": 1.6840059767056949e-06,
      "loss": 0.8494,
      "step": 4253
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.999406908081238,
      "learning_rate": 1.6805471291462316e-06,
      "loss": 0.9104,
      "step": 4254
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0121138141491295,
      "learning_rate": 1.6770915114848197e-06,
      "loss": 0.8329,
      "step": 4255
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.087165011391694,
      "learning_rate": 1.67363912506305e-06,
      "loss": 0.9028,
      "step": 4256
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.038724284725589,
      "learning_rate": 1.6701899712212565e-06,
      "loss": 0.9098,
      "step": 4257
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1431892476484038,
      "learning_rate": 1.66674405129852e-06,
      "loss": 0.8996,
      "step": 4258
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.924052085051398,
      "learning_rate": 1.6633013666326636e-06,
      "loss": 0.8465,
      "step": 4259
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8584696121240826,
      "learning_rate": 1.6598619185602616e-06,
      "loss": 0.5916,
      "step": 4260
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3108743117850277,
      "learning_rate": 1.656425708416617e-06,
      "loss": 0.8346,
      "step": 4261
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9977470142747754,
      "learning_rate": 1.6529927375357957e-06,
      "loss": 0.8818,
      "step": 4262
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9298677744257378,
      "learning_rate": 1.6495630072505841e-06,
      "loss": 0.8412,
      "step": 4263
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0026378349718952,
      "learning_rate": 1.6461365188925304e-06,
      "loss": 0.6391,
      "step": 4264
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1103882569663752,
      "learning_rate": 1.642713273791914e-06,
      "loss": 0.983,
      "step": 4265
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9669384833385677,
      "learning_rate": 1.6392932732777489e-06,
      "loss": 0.8857,
      "step": 4266
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9287063720612971,
      "learning_rate": 1.6358765186778057e-06,
      "loss": 0.8959,
      "step": 4267
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0641879532309617,
      "learning_rate": 1.6324630113185835e-06,
      "loss": 0.8763,
      "step": 4268
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7919685117549424,
      "learning_rate": 1.629052752525323e-06,
      "loss": 0.6115,
      "step": 4269
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1733151854735069,
      "learning_rate": 1.625645743622003e-06,
      "loss": 0.9215,
      "step": 4270
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1334837242382292,
      "learning_rate": 1.6222419859313443e-06,
      "loss": 0.9479,
      "step": 4271
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3023080199614436,
      "learning_rate": 1.6188414807747999e-06,
      "loss": 0.9458,
      "step": 4272
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.002431319107499,
      "learning_rate": 1.6154442294725636e-06,
      "loss": 0.9258,
      "step": 4273
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9983865346338592,
      "learning_rate": 1.6120502333435695e-06,
      "loss": 0.9516,
      "step": 4274
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2038654937504474,
      "learning_rate": 1.6086594937054767e-06,
      "loss": 0.9168,
      "step": 4275
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9952984318195612,
      "learning_rate": 1.6052720118746923e-06,
      "loss": 0.8869,
      "step": 4276
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9728143460121127,
      "learning_rate": 1.6018877891663521e-06,
      "loss": 0.926,
      "step": 4277
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1247010299421152,
      "learning_rate": 1.5985068268943283e-06,
      "loss": 0.9576,
      "step": 4278
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0826352803468013,
      "learning_rate": 1.5951291263712255e-06,
      "loss": 0.8671,
      "step": 4279
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8691689869721448,
      "learning_rate": 1.5917546889083834e-06,
      "loss": 0.618,
      "step": 4280
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0602467799063602,
      "learning_rate": 1.5883835158158767e-06,
      "loss": 0.9766,
      "step": 4281
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0674401640426312,
      "learning_rate": 1.5850156084025091e-06,
      "loss": 0.902,
      "step": 4282
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0415370046611625,
      "learning_rate": 1.5816509679758185e-06,
      "loss": 0.8645,
      "step": 4283
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.060381645627684,
      "learning_rate": 1.578289595842074e-06,
      "loss": 0.8963,
      "step": 4284
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9646792546679264,
      "learning_rate": 1.5749314933062754e-06,
      "loss": 0.8709,
      "step": 4285
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0417797371223865,
      "learning_rate": 1.5715766616721584e-06,
      "loss": 0.9346,
      "step": 4286
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0283627084721354,
      "learning_rate": 1.5682251022421757e-06,
      "loss": 0.8799,
      "step": 4287
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.983801461423383,
      "learning_rate": 1.5648768163175277e-06,
      "loss": 0.8779,
      "step": 4288
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0666625630545734,
      "learning_rate": 1.5615318051981243e-06,
      "loss": 0.8154,
      "step": 4289
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0568912649758915,
      "learning_rate": 1.5581900701826226e-06,
      "loss": 0.9128,
      "step": 4290
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8663076907959371,
      "learning_rate": 1.5548516125683976e-06,
      "loss": 0.6164,
      "step": 4291
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.028025259217129,
      "learning_rate": 1.5515164336515465e-06,
      "loss": 0.9351,
      "step": 4292
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1994032463542041,
      "learning_rate": 1.5481845347269077e-06,
      "loss": 0.9171,
      "step": 4293
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2290488694475632,
      "learning_rate": 1.5448559170880373e-06,
      "loss": 0.9025,
      "step": 4294
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8861494652585603,
      "learning_rate": 1.5415305820272198e-06,
      "loss": 0.6545,
      "step": 4295
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1710621193044262,
      "learning_rate": 1.5382085308354633e-06,
      "loss": 0.9016,
      "step": 4296
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0538268722652508,
      "learning_rate": 1.534889764802503e-06,
      "loss": 0.9304,
      "step": 4297
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9002409309416475,
      "learning_rate": 1.5315742852167992e-06,
      "loss": 0.8512,
      "step": 4298
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.040137638202304,
      "learning_rate": 1.5282620933655312e-06,
      "loss": 0.889,
      "step": 4299
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.07420552874145,
      "learning_rate": 1.5249531905346138e-06,
      "loss": 0.9578,
      "step": 4300
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0253331456649835,
      "learning_rate": 1.521647578008667e-06,
      "loss": 0.9235,
      "step": 4301
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9318687530283165,
      "learning_rate": 1.5183452570710522e-06,
      "loss": 0.8559,
      "step": 4302
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.120974554519099,
      "learning_rate": 1.5150462290038392e-06,
      "loss": 0.9301,
      "step": 4303
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.133790909061686,
      "learning_rate": 1.511750495087827e-06,
      "loss": 0.9065,
      "step": 4304
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9751541557571837,
      "learning_rate": 1.5084580566025309e-06,
      "loss": 0.9481,
      "step": 4305
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0670139398684555,
      "learning_rate": 1.5051689148261895e-06,
      "loss": 0.903,
      "step": 4306
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1150648406585868,
      "learning_rate": 1.5018830710357612e-06,
      "loss": 0.9035,
      "step": 4307
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8216314573042651,
      "learning_rate": 1.4986005265069204e-06,
      "loss": 0.6178,
      "step": 4308
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0443916045964674,
      "learning_rate": 1.4953212825140728e-06,
      "loss": 0.9328,
      "step": 4309
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1207145477844962,
      "learning_rate": 1.4920453403303249e-06,
      "loss": 0.9008,
      "step": 4310
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.090861017497361,
      "learning_rate": 1.4887727012275112e-06,
      "loss": 0.8895,
      "step": 4311
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0120113936374682,
      "learning_rate": 1.4855033664761898e-06,
      "loss": 0.8158,
      "step": 4312
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.796187115970241,
      "learning_rate": 1.48223733734562e-06,
      "loss": 0.623,
      "step": 4313
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9575534000709501,
      "learning_rate": 1.4789746151037942e-06,
      "loss": 0.8745,
      "step": 4314
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9694563651539907,
      "learning_rate": 1.475715201017407e-06,
      "loss": 0.885,
      "step": 4315
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.201367406902439,
      "learning_rate": 1.4724590963518803e-06,
      "loss": 0.933,
      "step": 4316
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0867448869608887,
      "learning_rate": 1.4692063023713444e-06,
      "loss": 0.9054,
      "step": 4317
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9072393435962733,
      "learning_rate": 1.4659568203386464e-06,
      "loss": 0.8182,
      "step": 4318
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8756588421589964,
      "learning_rate": 1.4627106515153456e-06,
      "loss": 0.8376,
      "step": 4319
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7858287233511809,
      "learning_rate": 1.4594677971617178e-06,
      "loss": 0.6335,
      "step": 4320
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9860324999654573,
      "learning_rate": 1.4562282585367493e-06,
      "loss": 0.9029,
      "step": 4321
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0029556869760252,
      "learning_rate": 1.452992036898142e-06,
      "loss": 0.9051,
      "step": 4322
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9575182564817868,
      "learning_rate": 1.4497591335023087e-06,
      "loss": 0.8957,
      "step": 4323
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9764847819874302,
      "learning_rate": 1.446529549604373e-06,
      "loss": 0.939,
      "step": 4324
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.073516083782765,
      "learning_rate": 1.4433032864581687e-06,
      "loss": 0.9506,
      "step": 4325
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9532199044349577,
      "learning_rate": 1.4400803453162482e-06,
      "loss": 0.9089,
      "step": 4326
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9994639100808458,
      "learning_rate": 1.4368607274298596e-06,
      "loss": 0.9394,
      "step": 4327
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0358131614025623,
      "learning_rate": 1.4336444340489775e-06,
      "loss": 0.9714,
      "step": 4328
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1157663057165441,
      "learning_rate": 1.430431466422273e-06,
      "loss": 0.8985,
      "step": 4329
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.15224255040679,
      "learning_rate": 1.4272218257971327e-06,
      "loss": 0.9378,
      "step": 4330
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8589193593817445,
      "learning_rate": 1.4240155134196499e-06,
      "loss": 0.6389,
      "step": 4331
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9839805816196406,
      "learning_rate": 1.4208125305346232e-06,
      "loss": 0.9314,
      "step": 4332
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9981234833911323,
      "learning_rate": 1.4176128783855636e-06,
      "loss": 0.9042,
      "step": 4333
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9311124155958618,
      "learning_rate": 1.4144165582146819e-06,
      "loss": 0.8602,
      "step": 4334
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0151620293361436,
      "learning_rate": 1.4112235712629063e-06,
      "loss": 0.9224,
      "step": 4335
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9537998163273043,
      "learning_rate": 1.40803391876986e-06,
      "loss": 0.9114,
      "step": 4336
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1378524389351663,
      "learning_rate": 1.4048476019738756e-06,
      "loss": 0.8779,
      "step": 4337
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0722096698697319,
      "learning_rate": 1.4016646221119912e-06,
      "loss": 0.9111,
      "step": 4338
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1699172065240504,
      "learning_rate": 1.3984849804199485e-06,
      "loss": 0.896,
      "step": 4339
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.010088912535938,
      "learning_rate": 1.395308678132199e-06,
      "loss": 0.8274,
      "step": 4340
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.057801456460681,
      "learning_rate": 1.392135716481885e-06,
      "loss": 0.9314,
      "step": 4341
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.970019014539459,
      "learning_rate": 1.3889660967008656e-06,
      "loss": 0.8887,
      "step": 4342
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0800017931119796,
      "learning_rate": 1.3857998200196943e-06,
      "loss": 0.8794,
      "step": 4343
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8977336977540412,
      "learning_rate": 1.3826368876676278e-06,
      "loss": 0.8824,
      "step": 4344
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1212250836342217,
      "learning_rate": 1.379477300872626e-06,
      "loss": 0.94,
      "step": 4345
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0481620701362304,
      "learning_rate": 1.3763210608613497e-06,
      "loss": 0.9376,
      "step": 4346
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.171544520350041,
      "learning_rate": 1.3731681688591593e-06,
      "loss": 0.9504,
      "step": 4347
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0219406946746012,
      "learning_rate": 1.370018626090116e-06,
      "loss": 0.8767,
      "step": 4348
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1319345743120608,
      "learning_rate": 1.3668724337769823e-06,
      "loss": 0.8991,
      "step": 4349
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.037821425150721,
      "learning_rate": 1.3637295931412153e-06,
      "loss": 0.8945,
      "step": 4350
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9910350833831223,
      "learning_rate": 1.3605901054029746e-06,
      "loss": 0.8961,
      "step": 4351
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8713557343594515,
      "learning_rate": 1.3574539717811231e-06,
      "loss": 0.8997,
      "step": 4352
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0552468902567742,
      "learning_rate": 1.3543211934932065e-06,
      "loss": 0.855,
      "step": 4353
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9951283526332372,
      "learning_rate": 1.3511917717554846e-06,
      "loss": 0.9293,
      "step": 4354
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7881318030843186,
      "learning_rate": 1.348065707782904e-06,
      "loss": 0.6174,
      "step": 4355
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1298228861422248,
      "learning_rate": 1.3449430027891096e-06,
      "loss": 0.8687,
      "step": 4356
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1661469313316561,
      "learning_rate": 1.3418236579864452e-06,
      "loss": 0.9772,
      "step": 4357
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0649717663080183,
      "learning_rate": 1.338707674585945e-06,
      "loss": 0.8997,
      "step": 4358
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2330067096332444,
      "learning_rate": 1.3355950537973438e-06,
      "loss": 0.9589,
      "step": 4359
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.025530376788193,
      "learning_rate": 1.332485796829065e-06,
      "loss": 0.9005,
      "step": 4360
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0663418559353985,
      "learning_rate": 1.329379904888235e-06,
      "loss": 0.9039,
      "step": 4361
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8942909823830291,
      "learning_rate": 1.3262773791806617e-06,
      "loss": 0.607,
      "step": 4362
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.871156592411638,
      "learning_rate": 1.3231782209108546e-06,
      "loss": 0.6582,
      "step": 4363
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.03879200019969,
      "learning_rate": 1.3200824312820137e-06,
      "loss": 0.9189,
      "step": 4364
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1644910791076455,
      "learning_rate": 1.3169900114960298e-06,
      "loss": 0.9841,
      "step": 4365
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.119794350281723,
      "learning_rate": 1.3139009627534927e-06,
      "loss": 0.9564,
      "step": 4366
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9549578786503653,
      "learning_rate": 1.3108152862536683e-06,
      "loss": 0.8715,
      "step": 4367
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.793759900390019,
      "learning_rate": 1.3077329831945295e-06,
      "loss": 0.5969,
      "step": 4368
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.076892031463776,
      "learning_rate": 1.3046540547727305e-06,
      "loss": 0.9199,
      "step": 4369
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0245378310854083,
      "learning_rate": 1.3015785021836159e-06,
      "loss": 0.8694,
      "step": 4370
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0542850697956425,
      "learning_rate": 1.2985063266212229e-06,
      "loss": 0.8887,
      "step": 4371
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9330110427292647,
      "learning_rate": 1.295437529278275e-06,
      "loss": 0.8998,
      "step": 4372
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.091142860304607,
      "learning_rate": 1.2923721113461852e-06,
      "loss": 0.8924,
      "step": 4373
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1395866991038572,
      "learning_rate": 1.2893100740150522e-06,
      "loss": 0.974,
      "step": 4374
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0722678326495432,
      "learning_rate": 1.2862514184736695e-06,
      "loss": 0.9177,
      "step": 4375
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9148516356147077,
      "learning_rate": 1.2831961459095088e-06,
      "loss": 0.6451,
      "step": 4376
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0728909916466705,
      "learning_rate": 1.2801442575087296e-06,
      "loss": 0.8314,
      "step": 4377
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0680687344919753,
      "learning_rate": 1.2770957544561868e-06,
      "loss": 0.9363,
      "step": 4378
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0002113143230829,
      "learning_rate": 1.274050637935408e-06,
      "loss": 0.891,
      "step": 4379
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0420795497228559,
      "learning_rate": 1.2710089091286148e-06,
      "loss": 0.923,
      "step": 4380
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1468095456383682,
      "learning_rate": 1.2679705692167122e-06,
      "loss": 0.8993,
      "step": 4381
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.89924145243646,
      "learning_rate": 1.2649356193792873e-06,
      "loss": 0.8462,
      "step": 4382
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8231176008298885,
      "learning_rate": 1.261904060794612e-06,
      "loss": 0.6291,
      "step": 4383
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0247007305044622,
      "learning_rate": 1.2588758946396417e-06,
      "loss": 0.9065,
      "step": 4384
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0663901518210992,
      "learning_rate": 1.2558511220900138e-06,
      "loss": 0.9472,
      "step": 4385
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3802615271670196,
      "learning_rate": 1.2528297443200489e-06,
      "loss": 0.9858,
      "step": 4386
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7960452325429909,
      "learning_rate": 1.2498117625027562e-06,
      "loss": 0.5705,
      "step": 4387
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0219671770505747,
      "learning_rate": 1.246797177809812e-06,
      "loss": 0.8828,
      "step": 4388
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8219292566005391,
      "learning_rate": 1.2437859914115847e-06,
      "loss": 0.5809,
      "step": 4389
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9198294505079788,
      "learning_rate": 1.2407782044771222e-06,
      "loss": 0.6162,
      "step": 4390
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0680828661782578,
      "learning_rate": 1.237773818174146e-06,
      "loss": 0.9258,
      "step": 4391
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.130274075676359,
      "learning_rate": 1.23477283366907e-06,
      "loss": 0.8988,
      "step": 4392
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9625053696571623,
      "learning_rate": 1.2317752521269722e-06,
      "loss": 0.6101,
      "step": 4393
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.112963588781592,
      "learning_rate": 1.2287810747116224e-06,
      "loss": 0.918,
      "step": 4394
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0195829319689045,
      "learning_rate": 1.225790302585461e-06,
      "loss": 0.9353,
      "step": 4395
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0998486529336988,
      "learning_rate": 1.2228029369096094e-06,
      "loss": 0.9266,
      "step": 4396
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9265226132510522,
      "learning_rate": 1.2198189788438652e-06,
      "loss": 0.8446,
      "step": 4397
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9335799630196535,
      "learning_rate": 1.216838429546704e-06,
      "loss": 0.9028,
      "step": 4398
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9695264704329611,
      "learning_rate": 1.2138612901752777e-06,
      "loss": 0.8785,
      "step": 4399
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8064520025934323,
      "learning_rate": 1.2108875618854122e-06,
      "loss": 0.6368,
      "step": 4400
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9610260441932177,
      "learning_rate": 1.2079172458316168e-06,
      "loss": 0.8612,
      "step": 4401
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1597362439987753,
      "learning_rate": 1.204950343167065e-06,
      "loss": 0.8929,
      "step": 4402
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9738046361544066,
      "learning_rate": 1.2019868550436099e-06,
      "loss": 0.8844,
      "step": 4403
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9839933431085314,
      "learning_rate": 1.1990267826117874e-06,
      "loss": 0.8805,
      "step": 4404
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1357738566513857,
      "learning_rate": 1.1960701270207885e-06,
      "loss": 0.8673,
      "step": 4405
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.068215792254832,
      "learning_rate": 1.1931168894184974e-06,
      "loss": 0.9116,
      "step": 4406
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9853452740422275,
      "learning_rate": 1.19016707095146e-06,
      "loss": 0.8852,
      "step": 4407
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0635968957731612,
      "learning_rate": 1.187220672764897e-06,
      "loss": 0.967,
      "step": 4408
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9129580936078456,
      "learning_rate": 1.1842776960027014e-06,
      "loss": 0.89,
      "step": 4409
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9681986335194501,
      "learning_rate": 1.1813381418074388e-06,
      "loss": 0.9182,
      "step": 4410
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9293093674933672,
      "learning_rate": 1.1784020113203453e-06,
      "loss": 0.616,
      "step": 4411
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0842895604441476,
      "learning_rate": 1.1754693056813272e-06,
      "loss": 0.9328,
      "step": 4412
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9583401506983105,
      "learning_rate": 1.172540026028962e-06,
      "loss": 0.9227,
      "step": 4413
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0611110945130355,
      "learning_rate": 1.169614173500494e-06,
      "loss": 0.863,
      "step": 4414
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9415436858055104,
      "learning_rate": 1.1666917492318486e-06,
      "loss": 0.8567,
      "step": 4415
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0488398518430633,
      "learning_rate": 1.1637727543576027e-06,
      "loss": 0.8764,
      "step": 4416
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0016534872670777,
      "learning_rate": 1.1608571900110122e-06,
      "loss": 0.8598,
      "step": 4417
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9659288533898834,
      "learning_rate": 1.1579450573240058e-06,
      "loss": 0.9096,
      "step": 4418
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9816807387419867,
      "learning_rate": 1.1550363574271638e-06,
      "loss": 0.8948,
      "step": 4419
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.875452234092018,
      "learning_rate": 1.1521310914497518e-06,
      "loss": 0.8625,
      "step": 4420
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.027449300324319,
      "learning_rate": 1.149229260519691e-06,
      "loss": 0.9289,
      "step": 4421
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2360341090086155,
      "learning_rate": 1.1463308657635718e-06,
      "loss": 0.8625,
      "step": 4422
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9189738559039795,
      "learning_rate": 1.1434359083066515e-06,
      "loss": 0.9227,
      "step": 4423
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9536157630950618,
      "learning_rate": 1.140544389272853e-06,
      "loss": 0.9031,
      "step": 4424
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9561870570256966,
      "learning_rate": 1.1376563097847616e-06,
      "loss": 0.8436,
      "step": 4425
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9507930098257346,
      "learning_rate": 1.1347716709636282e-06,
      "loss": 0.8671,
      "step": 4426
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1465652233622272,
      "learning_rate": 1.1318904739293745e-06,
      "loss": 0.9124,
      "step": 4427
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8416592007989387,
      "learning_rate": 1.129012719800575e-06,
      "loss": 0.6225,
      "step": 4428
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1183714880388906,
      "learning_rate": 1.1261384096944728e-06,
      "loss": 0.9042,
      "step": 4429
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8457203048663587,
      "learning_rate": 1.1232675447269803e-06,
      "loss": 0.6114,
      "step": 4430
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1631146246465534,
      "learning_rate": 1.1204001260126574e-06,
      "loss": 0.8242,
      "step": 4431
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.147445921089448,
      "learning_rate": 1.1175361546647413e-06,
      "loss": 0.8963,
      "step": 4432
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.004959553127549,
      "learning_rate": 1.1146756317951224e-06,
      "loss": 0.8934,
      "step": 4433
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1781912343767271,
      "learning_rate": 1.1118185585143536e-06,
      "loss": 0.8714,
      "step": 4434
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9394518689644837,
      "learning_rate": 1.1089649359316501e-06,
      "loss": 0.8851,
      "step": 4435
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8797674751791585,
      "learning_rate": 1.1061147651548855e-06,
      "loss": 0.6578,
      "step": 4436
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9744895195317607,
      "learning_rate": 1.1032680472905932e-06,
      "loss": 0.8695,
      "step": 4437
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1196043688310024,
      "learning_rate": 1.1004247834439697e-06,
      "loss": 0.9146,
      "step": 4438
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9798902871386853,
      "learning_rate": 1.097584974718866e-06,
      "loss": 0.918,
      "step": 4439
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0447044324572439,
      "learning_rate": 1.0947486222177928e-06,
      "loss": 0.856,
      "step": 4440
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0091375498439825,
      "learning_rate": 1.0919157270419257e-06,
      "loss": 0.878,
      "step": 4441
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5495770842536851,
      "learning_rate": 1.0890862902910849e-06,
      "loss": 0.8401,
      "step": 4442
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.151784861132038,
      "learning_rate": 1.0862603130637562e-06,
      "loss": 0.9033,
      "step": 4443
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.121563481155465,
      "learning_rate": 1.0834377964570863e-06,
      "loss": 0.9343,
      "step": 4444
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0409518199068521,
      "learning_rate": 1.0806187415668668e-06,
      "loss": 0.8949,
      "step": 4445
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.127374047323631,
      "learning_rate": 1.0778031494875574e-06,
      "loss": 0.9272,
      "step": 4446
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0160772058507082,
      "learning_rate": 1.0749910213122649e-06,
      "loss": 0.9137,
      "step": 4447
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.99845490401282,
      "learning_rate": 1.072182358132755e-06,
      "loss": 0.8771,
      "step": 4448
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0529066360230948,
      "learning_rate": 1.0693771610394477e-06,
      "loss": 0.8645,
      "step": 4449
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1056832437800581,
      "learning_rate": 1.066575431121417e-06,
      "loss": 0.9145,
      "step": 4450
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0514697919767946,
      "learning_rate": 1.06377716946639e-06,
      "loss": 0.9037,
      "step": 4451
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0560067470409402,
      "learning_rate": 1.0609823771607487e-06,
      "loss": 0.9436,
      "step": 4452
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9905115527077397,
      "learning_rate": 1.0581910552895302e-06,
      "loss": 0.8499,
      "step": 4453
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.071858868072649,
      "learning_rate": 1.055403204936416e-06,
      "loss": 0.9321,
      "step": 4454
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.096312572265225,
      "learning_rate": 1.0526188271837512e-06,
      "loss": 0.9306,
      "step": 4455
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1192086641822994,
      "learning_rate": 1.0498379231125278e-06,
      "loss": 0.912,
      "step": 4456
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0046462227918822,
      "learning_rate": 1.047060493802381e-06,
      "loss": 0.8686,
      "step": 4457
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0215699477302045,
      "learning_rate": 1.0442865403316117e-06,
      "loss": 0.8476,
      "step": 4458
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7555610816051058,
      "learning_rate": 1.0415160637771604e-06,
      "loss": 0.5538,
      "step": 4459
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9883785897216651,
      "learning_rate": 1.0387490652146236e-06,
      "loss": 0.8947,
      "step": 4460
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0972306654213513,
      "learning_rate": 1.0359855457182455e-06,
      "loss": 0.9239,
      "step": 4461
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0166884215943761,
      "learning_rate": 1.0332255063609177e-06,
      "loss": 0.9317,
      "step": 4462
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1489753072093827,
      "learning_rate": 1.0304689482141839e-06,
      "loss": 0.864,
      "step": 4463
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0197159351636438,
      "learning_rate": 1.027715872348234e-06,
      "loss": 0.8641,
      "step": 4464
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8596063927542426,
      "learning_rate": 1.0249662798319072e-06,
      "loss": 0.8665,
      "step": 4465
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9629645313186856,
      "learning_rate": 1.0222201717326885e-06,
      "loss": 0.9595,
      "step": 4466
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1085991343739803,
      "learning_rate": 1.0194775491167164e-06,
      "loss": 0.8762,
      "step": 4467
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.968093558452783,
      "learning_rate": 1.0167384130487667e-06,
      "loss": 0.8869,
      "step": 4468
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9987424338034246,
      "learning_rate": 1.0140027645922656e-06,
      "loss": 0.9078,
      "step": 4469
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0843783289915978,
      "learning_rate": 1.0112706048092924e-06,
      "loss": 0.886,
      "step": 4470
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9562431707101327,
      "learning_rate": 1.0085419347605575e-06,
      "loss": 0.9206,
      "step": 4471
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2430632923859508,
      "learning_rate": 1.00581675550543e-06,
      "loss": 0.946,
      "step": 4472
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.021989420025645,
      "learning_rate": 1.003095068101917e-06,
      "loss": 0.9258,
      "step": 4473
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1365724081045976,
      "learning_rate": 1.0003768736066722e-06,
      "loss": 0.9633,
      "step": 4474
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0097786511248767,
      "learning_rate": 9.976621730749892e-07,
      "loss": 0.8794,
      "step": 4475
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0408051825590685,
      "learning_rate": 9.949509675608115e-07,
      "loss": 0.915,
      "step": 4476
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.034242565505704,
      "learning_rate": 9.922432581167207e-07,
      "loss": 0.9433,
      "step": 4477
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0061248433611734,
      "learning_rate": 9.895390457939414e-07,
      "loss": 0.9063,
      "step": 4478
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9771055222188889,
      "learning_rate": 9.86838331642348e-07,
      "loss": 0.8741,
      "step": 4479
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0274499755096296,
      "learning_rate": 9.84141116710442e-07,
      "loss": 0.9839,
      "step": 4480
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9985716316444357,
      "learning_rate": 9.814474020453824e-07,
      "loss": 0.8728,
      "step": 4481
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9442175361448587,
      "learning_rate": 9.787571886929604e-07,
      "loss": 0.8591,
      "step": 4482
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8002254333651062,
      "learning_rate": 9.76070477697605e-07,
      "loss": 0.652,
      "step": 4483
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2399391348551287,
      "learning_rate": 9.733872701023938e-07,
      "loss": 0.934,
      "step": 4484
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0771971709492276,
      "learning_rate": 9.707075669490407e-07,
      "loss": 0.9174,
      "step": 4485
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1002812886280482,
      "learning_rate": 9.680313692778976e-07,
      "loss": 0.8834,
      "step": 4486
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.062876375346258,
      "learning_rate": 9.653586781279567e-07,
      "loss": 0.9097,
      "step": 4487
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0169861988399,
      "learning_rate": 9.626894945368492e-07,
      "loss": 0.9226,
      "step": 4488
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8848076552761509,
      "learning_rate": 9.600238195408428e-07,
      "loss": 0.8942,
      "step": 4489
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1652039957130602,
      "learning_rate": 9.573616541748464e-07,
      "loss": 0.938,
      "step": 4490
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0370506291984134,
      "learning_rate": 9.547029994724023e-07,
      "loss": 0.9033,
      "step": 4491
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0973484992579439,
      "learning_rate": 9.520478564656898e-07,
      "loss": 0.8254,
      "step": 4492
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.120417730233836,
      "learning_rate": 9.49396226185535e-07,
      "loss": 0.8722,
      "step": 4493
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0652476745514188,
      "learning_rate": 9.467481096613829e-07,
      "loss": 0.8976,
      "step": 4494
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0818106273822947,
      "learning_rate": 9.441035079213267e-07,
      "loss": 0.9123,
      "step": 4495
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9256503886877292,
      "learning_rate": 9.414624219920953e-07,
      "loss": 0.8854,
      "step": 4496
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0235123045078884,
      "learning_rate": 9.38824852899043e-07,
      "loss": 0.9013,
      "step": 4497
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.035913484866418,
      "learning_rate": 9.361908016661703e-07,
      "loss": 0.857,
      "step": 4498
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0284075774690333,
      "learning_rate": 9.335602693161039e-07,
      "loss": 0.9414,
      "step": 4499
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.005238044831685,
      "learning_rate": 9.309332568701079e-07,
      "loss": 0.9149,
      "step": 4500
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9820580659674556,
      "learning_rate": 9.283097653480788e-07,
      "loss": 0.916,
      "step": 4501
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.164660448632621,
      "learning_rate": 9.256897957685463e-07,
      "loss": 0.8841,
      "step": 4502
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9994523620591376,
      "learning_rate": 9.230733491486721e-07,
      "loss": 0.88,
      "step": 4503
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9446052129260166,
      "learning_rate": 9.204604265042505e-07,
      "loss": 0.8714,
      "step": 4504
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.017961959832689,
      "learning_rate": 9.178510288497123e-07,
      "loss": 0.9176,
      "step": 4505
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.117518845498169,
      "learning_rate": 9.15245157198108e-07,
      "loss": 0.8445,
      "step": 4506
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1920131530946227,
      "learning_rate": 9.126428125611342e-07,
      "loss": 0.9655,
      "step": 4507
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0442503697756849,
      "learning_rate": 9.10043995949108e-07,
      "loss": 0.9314,
      "step": 4508
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.967234217820385,
      "learning_rate": 9.074487083709759e-07,
      "loss": 0.8688,
      "step": 4509
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9652590031802943,
      "learning_rate": 9.04856950834323e-07,
      "loss": 0.8548,
      "step": 4510
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0438360965808524,
      "learning_rate": 9.022687243453554e-07,
      "loss": 0.9294,
      "step": 4511
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0266773168649463,
      "learning_rate": 8.996840299089149e-07,
      "loss": 0.8738,
      "step": 4512
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9472937642420901,
      "learning_rate": 8.971028685284655e-07,
      "loss": 0.8942,
      "step": 4513
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0164004810878378,
      "learning_rate": 8.945252412061056e-07,
      "loss": 0.911,
      "step": 4514
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.999140036025587,
      "learning_rate": 8.91951148942557e-07,
      "loss": 0.9108,
      "step": 4515
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8996979365639106,
      "learning_rate": 8.893805927371724e-07,
      "loss": 0.6228,
      "step": 4516
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1706272448296977,
      "learning_rate": 8.868135735879291e-07,
      "loss": 0.884,
      "step": 4517
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0457504275255127,
      "learning_rate": 8.842500924914299e-07,
      "loss": 0.8762,
      "step": 4518
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9231095137694252,
      "learning_rate": 8.816901504429143e-07,
      "loss": 0.8782,
      "step": 4519
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9735615547814849,
      "learning_rate": 8.791337484362305e-07,
      "loss": 0.8913,
      "step": 4520
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0904575133475805,
      "learning_rate": 8.765808874638682e-07,
      "loss": 0.9274,
      "step": 4521
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9519189578842728,
      "learning_rate": 8.740315685169364e-07,
      "loss": 0.876,
      "step": 4522
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9529573409614002,
      "learning_rate": 8.714857925851617e-07,
      "loss": 0.9037,
      "step": 4523
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8119723735700376,
      "learning_rate": 8.689435606569086e-07,
      "loss": 0.5651,
      "step": 4524
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1252117539253963,
      "learning_rate": 8.664048737191566e-07,
      "loss": 0.9239,
      "step": 4525
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2164887567057454,
      "learning_rate": 8.638697327575108e-07,
      "loss": 0.9227,
      "step": 4526
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.018833552430835,
      "learning_rate": 8.613381387562015e-07,
      "loss": 0.8849,
      "step": 4527
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.239989208972835,
      "learning_rate": 8.588100926980802e-07,
      "loss": 0.9941,
      "step": 4528
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.08068908832291,
      "learning_rate": 8.56285595564621e-07,
      "loss": 0.8918,
      "step": 4529
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0980559978484414,
      "learning_rate": 8.537646483359185e-07,
      "loss": 0.9039,
      "step": 4530
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7780187363876847,
      "learning_rate": 8.512472519906978e-07,
      "loss": 0.6459,
      "step": 4531
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8538699175839308,
      "learning_rate": 8.487334075062914e-07,
      "loss": 0.6354,
      "step": 4532
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1464600614288019,
      "learning_rate": 8.462231158586654e-07,
      "loss": 0.9253,
      "step": 4533
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0782634561546707,
      "learning_rate": 8.437163780224011e-07,
      "loss": 0.9875,
      "step": 4534
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.168785491014986,
      "learning_rate": 8.412131949706958e-07,
      "loss": 0.8928,
      "step": 4535
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0706399217528597,
      "learning_rate": 8.387135676753755e-07,
      "loss": 0.874,
      "step": 4536
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0130494339144924,
      "learning_rate": 8.362174971068804e-07,
      "loss": 0.8698,
      "step": 4537
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9835761308891018,
      "learning_rate": 8.337249842342721e-07,
      "loss": 0.9428,
      "step": 4538
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8773978629317433,
      "learning_rate": 8.312360300252287e-07,
      "loss": 0.605,
      "step": 4539
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0664257273318822,
      "learning_rate": 8.287506354460484e-07,
      "loss": 0.951,
      "step": 4540
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7321304451978552,
      "learning_rate": 8.26268801461646e-07,
      "loss": 0.6206,
      "step": 4541
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0134308570734674,
      "learning_rate": 8.237905290355563e-07,
      "loss": 0.8329,
      "step": 4542
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0841671651006395,
      "learning_rate": 8.213158191299297e-07,
      "loss": 0.9342,
      "step": 4543
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6014428155071319,
      "learning_rate": 8.188446727055311e-07,
      "loss": 0.9198,
      "step": 4544
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0154840027615541,
      "learning_rate": 8.163770907217506e-07,
      "loss": 0.9272,
      "step": 4545
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9644719582229524,
      "learning_rate": 8.139130741365819e-07,
      "loss": 0.8491,
      "step": 4546
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8979035162846682,
      "learning_rate": 8.114526239066456e-07,
      "loss": 0.9072,
      "step": 4547
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1164380391737432,
      "learning_rate": 8.08995740987173e-07,
      "loss": 0.8972,
      "step": 4548
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0245515476095732,
      "learning_rate": 8.065424263320054e-07,
      "loss": 0.9388,
      "step": 4549
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0267847297997672,
      "learning_rate": 8.040926808936112e-07,
      "loss": 0.9534,
      "step": 4550
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0306153484887877,
      "learning_rate": 8.016465056230616e-07,
      "loss": 0.863,
      "step": 4551
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9511685598989286,
      "learning_rate": 7.99203901470047e-07,
      "loss": 0.8671,
      "step": 4552
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0777534834056817,
      "learning_rate": 7.967648693828712e-07,
      "loss": 0.8968,
      "step": 4553
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9322624705336471,
      "learning_rate": 7.943294103084487e-07,
      "loss": 0.8478,
      "step": 4554
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8238039470116394,
      "learning_rate": 7.9189752519231e-07,
      "loss": 0.6232,
      "step": 4555
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.982184532629386,
      "learning_rate": 7.894692149785954e-07,
      "loss": 0.9089,
      "step": 4556
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0303066777114553,
      "learning_rate": 7.870444806100619e-07,
      "loss": 0.9307,
      "step": 4557
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.100274716633259,
      "learning_rate": 7.846233230280698e-07,
      "loss": 0.8828,
      "step": 4558
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.00332501403189,
      "learning_rate": 7.822057431725994e-07,
      "loss": 0.8947,
      "step": 4559
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0995045793238685,
      "learning_rate": 7.797917419822377e-07,
      "loss": 0.9035,
      "step": 4560
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0120758144336663,
      "learning_rate": 7.773813203941827e-07,
      "loss": 0.8942,
      "step": 4561
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.81594205708904,
      "learning_rate": 7.749744793442448e-07,
      "loss": 0.6005,
      "step": 4562
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0528083638808532,
      "learning_rate": 7.725712197668378e-07,
      "loss": 0.8983,
      "step": 4563
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9572792567559577,
      "learning_rate": 7.701715425949952e-07,
      "loss": 0.8758,
      "step": 4564
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0973830082379612,
      "learning_rate": 7.677754487603517e-07,
      "loss": 0.8756,
      "step": 4565
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0033922389367644,
      "learning_rate": 7.653829391931533e-07,
      "loss": 0.9067,
      "step": 4566
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9745696908768477,
      "learning_rate": 7.629940148222559e-07,
      "loss": 0.9231,
      "step": 4567
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0383448941671385,
      "learning_rate": 7.606086765751209e-07,
      "loss": 0.8305,
      "step": 4568
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0871926647449361,
      "learning_rate": 7.582269253778185e-07,
      "loss": 0.9323,
      "step": 4569
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0067329879042106,
      "learning_rate": 7.55848762155027e-07,
      "loss": 0.9382,
      "step": 4570
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9848680283333223,
      "learning_rate": 7.534741878300333e-07,
      "loss": 0.898,
      "step": 4571
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8638759565533612,
      "learning_rate": 7.511032033247256e-07,
      "loss": 0.64,
      "step": 4572
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9191002074201696,
      "learning_rate": 7.487358095596031e-07,
      "loss": 0.8865,
      "step": 4573
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0775994420572186,
      "learning_rate": 7.463720074537728e-07,
      "loss": 0.92,
      "step": 4574
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0284590937674551,
      "learning_rate": 7.440117979249362e-07,
      "loss": 0.9285,
      "step": 4575
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0488441624116787,
      "learning_rate": 7.416551818894158e-07,
      "loss": 0.8888,
      "step": 4576
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.915600333038182,
      "learning_rate": 7.393021602621264e-07,
      "loss": 0.8823,
      "step": 4577
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0128575456473734,
      "learning_rate": 7.369527339565951e-07,
      "loss": 0.8709,
      "step": 4578
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8599427469176333,
      "learning_rate": 7.346069038849469e-07,
      "loss": 0.5787,
      "step": 4579
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0282707095975387,
      "learning_rate": 7.322646709579173e-07,
      "loss": 0.8843,
      "step": 4580
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0605321270837573,
      "learning_rate": 7.299260360848382e-07,
      "loss": 0.9143,
      "step": 4581
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0108592767204627,
      "learning_rate": 7.275910001736497e-07,
      "loss": 0.8882,
      "step": 4582
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1689018420292598,
      "learning_rate": 7.252595641308957e-07,
      "loss": 0.8886,
      "step": 4583
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.081266047845805,
      "learning_rate": 7.229317288617144e-07,
      "loss": 0.8915,
      "step": 4584
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.136224550585053,
      "learning_rate": 7.20607495269856e-07,
      "loss": 0.9439,
      "step": 4585
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8067079614388996,
      "learning_rate": 7.182868642576679e-07,
      "loss": 0.6066,
      "step": 4586
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1569264078999812,
      "learning_rate": 7.15969836726097e-07,
      "loss": 0.845,
      "step": 4587
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0125239361973497,
      "learning_rate": 7.13656413574696e-07,
      "loss": 0.8607,
      "step": 4588
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.873928144257595,
      "learning_rate": 7.113465957016097e-07,
      "loss": 0.6426,
      "step": 4589
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9954701903991341,
      "learning_rate": 7.090403840035942e-07,
      "loss": 0.9058,
      "step": 4590
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9241651887672204,
      "learning_rate": 7.067377793759999e-07,
      "loss": 0.8853,
      "step": 4591
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1086352454562323,
      "learning_rate": 7.044387827127752e-07,
      "loss": 0.9279,
      "step": 4592
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1138352360508106,
      "learning_rate": 7.021433949064704e-07,
      "loss": 0.9003,
      "step": 4593
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1088373420133093,
      "learning_rate": 6.99851616848235e-07,
      "loss": 0.9151,
      "step": 4594
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1320655008199765,
      "learning_rate": 6.975634494278149e-07,
      "loss": 0.9043,
      "step": 4595
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9419130187051961,
      "learning_rate": 6.952788935335541e-07,
      "loss": 0.9117,
      "step": 4596
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.120907492753938,
      "learning_rate": 6.92997950052402e-07,
      "loss": 0.9506,
      "step": 4597
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0231703278014006,
      "learning_rate": 6.907206198698912e-07,
      "loss": 0.9646,
      "step": 4598
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0780319266770877,
      "learning_rate": 6.884469038701646e-07,
      "loss": 0.9382,
      "step": 4599
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0222127309389886,
      "learning_rate": 6.861768029359595e-07,
      "loss": 0.9514,
      "step": 4600
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2505391057331317,
      "learning_rate": 6.839103179485995e-07,
      "loss": 0.981,
      "step": 4601
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9903828392757551,
      "learning_rate": 6.816474497880177e-07,
      "loss": 0.9106,
      "step": 4602
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0675096836712792,
      "learning_rate": 6.793881993327366e-07,
      "loss": 0.9367,
      "step": 4603
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9753494622856725,
      "learning_rate": 6.77132567459875e-07,
      "loss": 0.8842,
      "step": 4604
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1176007448941958,
      "learning_rate": 6.748805550451453e-07,
      "loss": 0.8915,
      "step": 4605
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0097907402633972,
      "learning_rate": 6.726321629628585e-07,
      "loss": 0.9061,
      "step": 4606
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9925968099584016,
      "learning_rate": 6.703873920859161e-07,
      "loss": 0.881,
      "step": 4607
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0758557721883357,
      "learning_rate": 6.681462432858154e-07,
      "loss": 0.8997,
      "step": 4608
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1057745540891049,
      "learning_rate": 6.659087174326506e-07,
      "loss": 0.9264,
      "step": 4609
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0336175382238177,
      "learning_rate": 6.636748153951e-07,
      "loss": 0.9018,
      "step": 4610
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0180055512048976,
      "learning_rate": 6.614445380404478e-07,
      "loss": 0.9029,
      "step": 4611
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.084938051435608,
      "learning_rate": 6.592178862345622e-07,
      "loss": 0.912,
      "step": 4612
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9432070329102814,
      "learning_rate": 6.569948608419041e-07,
      "loss": 0.9263,
      "step": 4613
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8360942689120173,
      "learning_rate": 6.547754627255332e-07,
      "loss": 0.6469,
      "step": 4614
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9773002181718474,
      "learning_rate": 6.52559692747089e-07,
      "loss": 0.9289,
      "step": 4615
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1652648691764713,
      "learning_rate": 6.503475517668168e-07,
      "loss": 0.9329,
      "step": 4616
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0025664428997758,
      "learning_rate": 6.481390406435417e-07,
      "loss": 0.9341,
      "step": 4617
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1272690464899515,
      "learning_rate": 6.459341602346858e-07,
      "loss": 0.8717,
      "step": 4618
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1184285767204067,
      "learning_rate": 6.437329113962576e-07,
      "loss": 0.9528,
      "step": 4619
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.201905438261635,
      "learning_rate": 6.415352949828601e-07,
      "loss": 0.855,
      "step": 4620
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9259509350346813,
      "learning_rate": 6.393413118476821e-07,
      "loss": 0.816,
      "step": 4621
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0719718033292938,
      "learning_rate": 6.371509628425021e-07,
      "loss": 0.8979,
      "step": 4622
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0979910624978282,
      "learning_rate": 6.349642488176943e-07,
      "loss": 0.8813,
      "step": 4623
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8235164405548926,
      "learning_rate": 6.327811706222097e-07,
      "loss": 0.6296,
      "step": 4624
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.224931073294751,
      "learning_rate": 6.306017291035981e-07,
      "loss": 0.9549,
      "step": 4625
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.052958038735949,
      "learning_rate": 6.284259251079939e-07,
      "loss": 0.9113,
      "step": 4626
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8133818117855378,
      "learning_rate": 6.262537594801177e-07,
      "loss": 0.6366,
      "step": 4627
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9432853017823537,
      "learning_rate": 6.240852330632796e-07,
      "loss": 0.9177,
      "step": 4628
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9939599552189116,
      "learning_rate": 6.219203466993762e-07,
      "loss": 0.9088,
      "step": 4629
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0155761256516673,
      "learning_rate": 6.197591012288918e-07,
      "loss": 0.8655,
      "step": 4630
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0806161425823144,
      "learning_rate": 6.17601497490895e-07,
      "loss": 0.8846,
      "step": 4631
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0896719080114703,
      "learning_rate": 6.154475363230417e-07,
      "loss": 0.8765,
      "step": 4632
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9498299494632809,
      "learning_rate": 6.132972185615749e-07,
      "loss": 0.9014,
      "step": 4633
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1924903256452002,
      "learning_rate": 6.111505450413202e-07,
      "loss": 0.9393,
      "step": 4634
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0583047771528022,
      "learning_rate": 6.090075165956943e-07,
      "loss": 0.9108,
      "step": 4635
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1879067087213595,
      "learning_rate": 6.068681340566896e-07,
      "loss": 0.9367,
      "step": 4636
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8302135432802108,
      "learning_rate": 6.047323982548924e-07,
      "loss": 0.6092,
      "step": 4637
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0443876315664369,
      "learning_rate": 6.026003100194633e-07,
      "loss": 0.971,
      "step": 4638
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9968204009958571,
      "learning_rate": 6.004718701781575e-07,
      "loss": 0.8971,
      "step": 4639
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1350718801449897,
      "learning_rate": 5.983470795573088e-07,
      "loss": 0.9836,
      "step": 4640
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9944519504485041,
      "learning_rate": 5.962259389818292e-07,
      "loss": 0.92,
      "step": 4641
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0108491126921617,
      "learning_rate": 5.941084492752236e-07,
      "loss": 0.913,
      "step": 4642
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1132047342474447,
      "learning_rate": 5.91994611259572e-07,
      "loss": 0.9158,
      "step": 4643
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.048840028768079,
      "learning_rate": 5.898844257555392e-07,
      "loss": 0.9141,
      "step": 4644
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9884186772117765,
      "learning_rate": 5.87777893582372e-07,
      "loss": 0.8816,
      "step": 4645
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0788161243321956,
      "learning_rate": 5.856750155578983e-07,
      "loss": 0.9086,
      "step": 4646
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0700088682876716,
      "learning_rate": 5.835757924985286e-07,
      "loss": 0.9287,
      "step": 4647
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0410030605334468,
      "learning_rate": 5.81480225219252e-07,
      "loss": 0.9157,
      "step": 4648
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0231738136434374,
      "learning_rate": 5.793883145336443e-07,
      "loss": 0.9059,
      "step": 4649
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1345282401123253,
      "learning_rate": 5.773000612538505e-07,
      "loss": 0.9497,
      "step": 4650
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8979405792436785,
      "learning_rate": 5.752154661906085e-07,
      "loss": 0.8809,
      "step": 4651
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.18471946970315,
      "learning_rate": 5.731345301532265e-07,
      "loss": 0.8901,
      "step": 4652
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8041858263759337,
      "learning_rate": 5.710572539495962e-07,
      "loss": 0.6391,
      "step": 4653
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9974519886913628,
      "learning_rate": 5.68983638386188e-07,
      "loss": 0.9099,
      "step": 4654
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0015394566861817,
      "learning_rate": 5.669136842680512e-07,
      "loss": 0.8588,
      "step": 4655
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9076246145700576,
      "learning_rate": 5.648473923988129e-07,
      "loss": 0.8844,
      "step": 4656
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1993463231705472,
      "learning_rate": 5.627847635806771e-07,
      "loss": 0.9316,
      "step": 4657
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1049633960357361,
      "learning_rate": 5.607257986144321e-07,
      "loss": 0.9038,
      "step": 4658
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0174346717917104,
      "learning_rate": 5.58670498299434e-07,
      "loss": 0.8631,
      "step": 4659
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9888869020477274,
      "learning_rate": 5.566188634336212e-07,
      "loss": 0.8879,
      "step": 4660
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1850995776567654,
      "learning_rate": 5.545708948135142e-07,
      "loss": 0.9619,
      "step": 4661
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.094808565849101,
      "learning_rate": 5.525265932341984e-07,
      "loss": 0.9023,
      "step": 4662
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2576539664586033,
      "learning_rate": 5.504859594893475e-07,
      "loss": 0.9194,
      "step": 4663
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0009879048798087,
      "learning_rate": 5.484489943712013e-07,
      "loss": 0.919,
      "step": 4664
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0274068625812178,
      "learning_rate": 5.464156986705826e-07,
      "loss": 0.8859,
      "step": 4665
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0217872886434431,
      "learning_rate": 5.443860731768869e-07,
      "loss": 0.8905,
      "step": 4666
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8712277885289486,
      "learning_rate": 5.423601186780836e-07,
      "loss": 0.6403,
      "step": 4667
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0371969031407002,
      "learning_rate": 5.403378359607181e-07,
      "loss": 0.9085,
      "step": 4668
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0532533208320658,
      "learning_rate": 5.383192258099113e-07,
      "loss": 0.8871,
      "step": 4669
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9982293416146287,
      "learning_rate": 5.36304289009355e-07,
      "loss": 0.9189,
      "step": 4670
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.139635977967054,
      "learning_rate": 5.342930263413193e-07,
      "loss": 0.9195,
      "step": 4671
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1418842729616467,
      "learning_rate": 5.322854385866439e-07,
      "loss": 0.9162,
      "step": 4672
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6116244691660901,
      "learning_rate": 5.302815265247452e-07,
      "loss": 0.3211,
      "step": 4673
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9309806072751625,
      "learning_rate": 5.282812909336077e-07,
      "loss": 0.8258,
      "step": 4674
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0193946219752248,
      "learning_rate": 5.262847325897968e-07,
      "loss": 0.9266,
      "step": 4675
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1905833337625742,
      "learning_rate": 5.242918522684392e-07,
      "loss": 0.8897,
      "step": 4676
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0162587176320146,
      "learning_rate": 5.22302650743245e-07,
      "loss": 0.8868,
      "step": 4677
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0632123912409166,
      "learning_rate": 5.203171287864872e-07,
      "loss": 0.929,
      "step": 4678
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9122951817869697,
      "learning_rate": 5.183352871690162e-07,
      "loss": 0.9425,
      "step": 4679
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9554046731311879,
      "learning_rate": 5.163571266602485e-07,
      "loss": 0.8711,
      "step": 4680
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9920199975131706,
      "learning_rate": 5.143826480281778e-07,
      "loss": 0.925,
      "step": 4681
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8505322836208989,
      "learning_rate": 5.124118520393606e-07,
      "loss": 0.6151,
      "step": 4682
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1391260348007586,
      "learning_rate": 5.104447394589295e-07,
      "loss": 0.9343,
      "step": 4683
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8278298949385863,
      "learning_rate": 5.084813110505871e-07,
      "loss": 0.5882,
      "step": 4684
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0147928742286823,
      "learning_rate": 5.065215675766023e-07,
      "loss": 0.8802,
      "step": 4685
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1299892254260078,
      "learning_rate": 5.045655097978131e-07,
      "loss": 0.8766,
      "step": 4686
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9524130603892669,
      "learning_rate": 5.026131384736321e-07,
      "loss": 0.8956,
      "step": 4687
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9329853032886386,
      "learning_rate": 5.006644543620342e-07,
      "loss": 0.8997,
      "step": 4688
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7840529151714933,
      "learning_rate": 4.987194582195687e-07,
      "loss": 0.5739,
      "step": 4689
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0517989638042953,
      "learning_rate": 4.967781508013459e-07,
      "loss": 0.8579,
      "step": 4690
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9913662469329515,
      "learning_rate": 4.948405328610506e-07,
      "loss": 0.8377,
      "step": 4691
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9728522097337106,
      "learning_rate": 4.929066051509346e-07,
      "loss": 0.8784,
      "step": 4692
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9625331278366016,
      "learning_rate": 4.909763684218116e-07,
      "loss": 0.9163,
      "step": 4693
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1332483118205765,
      "learning_rate": 4.890498234230689e-07,
      "loss": 0.9304,
      "step": 4694
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9614881589230326,
      "learning_rate": 4.871269709026561e-07,
      "loss": 0.8769,
      "step": 4695
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.860392391998311,
      "learning_rate": 4.852078116070902e-07,
      "loss": 0.6222,
      "step": 4696
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0677791466699003,
      "learning_rate": 4.832923462814565e-07,
      "loss": 0.9195,
      "step": 4697
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.918911112099261,
      "learning_rate": 4.813805756694035e-07,
      "loss": 0.9031,
      "step": 4698
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9628256914691626,
      "learning_rate": 4.794725005131462e-07,
      "loss": 0.8896,
      "step": 4699
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1438034710947618,
      "learning_rate": 4.775681215534656e-07,
      "loss": 0.9317,
      "step": 4700
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0818884868121386,
      "learning_rate": 4.7566743952970894e-07,
      "loss": 0.8991,
      "step": 4701
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8402860375096703,
      "learning_rate": 4.7377045517978173e-07,
      "loss": 0.6168,
      "step": 4702
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9278103745852696,
      "learning_rate": 4.7187716924016355e-07,
      "loss": 0.9443,
      "step": 4703
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9113869437858824,
      "learning_rate": 4.6998758244588995e-07,
      "loss": 0.8899,
      "step": 4704
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0648923874207294,
      "learning_rate": 4.6810169553056616e-07,
      "loss": 0.871,
      "step": 4705
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.056977192649959,
      "learning_rate": 4.662195092263566e-07,
      "loss": 0.9251,
      "step": 4706
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0754965211706253,
      "learning_rate": 4.643410242639912e-07,
      "loss": 0.9372,
      "step": 4707
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9993339441276426,
      "learning_rate": 4.6246624137276206e-07,
      "loss": 0.8256,
      "step": 4708
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0469580944732695,
      "learning_rate": 4.605951612805237e-07,
      "loss": 0.8668,
      "step": 4709
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0180349115261638,
      "learning_rate": 4.587277847136984e-07,
      "loss": 0.8891,
      "step": 4710
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1463569551216541,
      "learning_rate": 4.568641123972606e-07,
      "loss": 0.9033,
      "step": 4711
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9846999106654859,
      "learning_rate": 4.550041450547549e-07,
      "loss": 0.8987,
      "step": 4712
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0584934416275875,
      "learning_rate": 4.5314788340828365e-07,
      "loss": 0.9373,
      "step": 4713
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2021739207091586,
      "learning_rate": 4.512953281785104e-07,
      "loss": 0.9283,
      "step": 4714
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9025960116527959,
      "learning_rate": 4.494464800846654e-07,
      "loss": 0.622,
      "step": 4715
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1445196769775337,
      "learning_rate": 4.476013398445289e-07,
      "loss": 0.9161,
      "step": 4716
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9932204901438881,
      "learning_rate": 4.4575990817445234e-07,
      "loss": 0.8954,
      "step": 4717
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0008607355419272,
      "learning_rate": 4.4392218578934164e-07,
      "loss": 0.9084,
      "step": 4718
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7852957783031439,
      "learning_rate": 4.4208817340266385e-07,
      "loss": 0.5827,
      "step": 4719
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9950061819008649,
      "learning_rate": 4.4025787172644495e-07,
      "loss": 0.9069,
      "step": 4720
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9665636916911351,
      "learning_rate": 4.384312814712721e-07,
      "loss": 0.9314,
      "step": 4721
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9994054043828866,
      "learning_rate": 4.366084033462914e-07,
      "loss": 0.8843,
      "step": 4722
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2808366490724454,
      "learning_rate": 4.3478923805920335e-07,
      "loss": 0.9579,
      "step": 4723
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0765970079669536,
      "learning_rate": 4.329737863162753e-07,
      "loss": 0.8891,
      "step": 4724
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.090612889532767,
      "learning_rate": 4.311620488223256e-07,
      "loss": 0.8743,
      "step": 4725
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0899146812781384,
      "learning_rate": 4.2935402628073166e-07,
      "loss": 0.9025,
      "step": 4726
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0912993246149953,
      "learning_rate": 4.27549719393433e-07,
      "loss": 0.924,
      "step": 4727
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0408895509700336,
      "learning_rate": 4.2574912886092166e-07,
      "loss": 0.922,
      "step": 4728
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9469705075250122,
      "learning_rate": 4.239522553822495e-07,
      "loss": 0.8236,
      "step": 4729
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9858185026353851,
      "learning_rate": 4.221590996550251e-07,
      "loss": 0.8713,
      "step": 4730
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1083540558857103,
      "learning_rate": 4.203696623754139e-07,
      "loss": 0.8698,
      "step": 4731
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9559926445402893,
      "learning_rate": 4.1858394423813563e-07,
      "loss": 0.8574,
      "step": 4732
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1716072487921267,
      "learning_rate": 4.1680194593646696e-07,
      "loss": 0.9156,
      "step": 4733
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.122203447579292,
      "learning_rate": 4.1502366816224327e-07,
      "loss": 0.9737,
      "step": 4734
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0408572821092155,
      "learning_rate": 4.1324911160585014e-07,
      "loss": 0.88,
      "step": 4735
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.097122550927477,
      "learning_rate": 4.1147827695623643e-07,
      "loss": 0.9495,
      "step": 4736
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0579984484929446,
      "learning_rate": 4.097111649008967e-07,
      "loss": 0.9148,
      "step": 4737
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9944779332274529,
      "learning_rate": 4.0794777612588543e-07,
      "loss": 0.8676,
      "step": 4738
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0007234739548954,
      "learning_rate": 4.061881113158117e-07,
      "loss": 0.8721,
      "step": 4739
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.13652049298781,
      "learning_rate": 4.044321711538368e-07,
      "loss": 0.9205,
      "step": 4740
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1299985679605786,
      "learning_rate": 4.02679956321681e-07,
      "loss": 0.9752,
      "step": 4741
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7921278891093768,
      "learning_rate": 4.00931467499609e-07,
      "loss": 0.6189,
      "step": 4742
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0651668598967392,
      "learning_rate": 3.9918670536644776e-07,
      "loss": 0.9021,
      "step": 4743
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.162068324005122,
      "learning_rate": 3.974456705995733e-07,
      "loss": 0.9392,
      "step": 4744
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0100593623874177,
      "learning_rate": 3.9570836387491487e-07,
      "loss": 0.8723,
      "step": 4745
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2077423390312152,
      "learning_rate": 3.9397478586695513e-07,
      "loss": 0.9176,
      "step": 4746
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9641597540752276,
      "learning_rate": 3.9224493724872915e-07,
      "loss": 0.8649,
      "step": 4747
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9970277406812035,
      "learning_rate": 3.90518818691823e-07,
      "loss": 0.9054,
      "step": 4748
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1186241084314765,
      "learning_rate": 3.8879643086637384e-07,
      "loss": 0.937,
      "step": 4749
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9967206985110283,
      "learning_rate": 3.8707777444107697e-07,
      "loss": 0.9264,
      "step": 4750
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9601167281256149,
      "learning_rate": 3.8536285008316854e-07,
      "loss": 0.8844,
      "step": 4751
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0926586164094014,
      "learning_rate": 3.8365165845844266e-07,
      "loss": 0.909,
      "step": 4752
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0625168792658346,
      "learning_rate": 3.819442002312457e-07,
      "loss": 0.9153,
      "step": 4753
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0933942011731055,
      "learning_rate": 3.8024047606446736e-07,
      "loss": 0.9167,
      "step": 4754
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0519111757470088,
      "learning_rate": 3.785404866195552e-07,
      "loss": 0.8687,
      "step": 4755
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0732007230180354,
      "learning_rate": 3.768442325565036e-07,
      "loss": 0.9319,
      "step": 4756
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1344497544910295,
      "learning_rate": 3.751517145338546e-07,
      "loss": 0.9178,
      "step": 4757
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.093358550739216,
      "learning_rate": 3.7346293320870363e-07,
      "loss": 0.8854,
      "step": 4758
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0241467676739464,
      "learning_rate": 3.717778892366941e-07,
      "loss": 0.8761,
      "step": 4759
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0901814810377668,
      "learning_rate": 3.700965832720171e-07,
      "loss": 0.9088,
      "step": 4760
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0644651007634804,
      "learning_rate": 3.684190159674117e-07,
      "loss": 0.8933,
      "step": 4761
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9127673182235638,
      "learning_rate": 3.6674518797417236e-07,
      "loss": 0.8666,
      "step": 4762
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1368889537850688,
      "learning_rate": 3.6507509994213155e-07,
      "loss": 0.9149,
      "step": 4763
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.781302269438765,
      "learning_rate": 3.6340875251967946e-07,
      "loss": 0.6184,
      "step": 4764
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0896315075297065,
      "learning_rate": 3.617461463537464e-07,
      "loss": 0.8746,
      "step": 4765
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8029621996255628,
      "learning_rate": 3.6008728208981157e-07,
      "loss": 0.6166,
      "step": 4766
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0191046095695095,
      "learning_rate": 3.5843216037190873e-07,
      "loss": 0.9046,
      "step": 4767
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1723719250583986,
      "learning_rate": 3.5678078184260834e-07,
      "loss": 0.9274,
      "step": 4768
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0241354808085996,
      "learning_rate": 3.5513314714303524e-07,
      "loss": 0.9093,
      "step": 4769
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.134463455287337,
      "learning_rate": 3.5348925691285675e-07,
      "loss": 0.9264,
      "step": 4770
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1389169132846513,
      "learning_rate": 3.518491117902878e-07,
      "loss": 0.9047,
      "step": 4771
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9604763105728492,
      "learning_rate": 3.502127124120891e-07,
      "loss": 0.9105,
      "step": 4772
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0863725207700148,
      "learning_rate": 3.48580059413568e-07,
      "loss": 0.9051,
      "step": 4773
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1249155358058343,
      "learning_rate": 3.4695115342857524e-07,
      "loss": 0.9066,
      "step": 4774
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0451591227936141,
      "learning_rate": 3.4532599508950826e-07,
      "loss": 0.9185,
      "step": 4775
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.981047193713904,
      "learning_rate": 3.437045850273113e-07,
      "loss": 0.9326,
      "step": 4776
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1205397746624408,
      "learning_rate": 3.420869238714708e-07,
      "loss": 0.5951,
      "step": 4777
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8562264799016732,
      "learning_rate": 3.404730122500155e-07,
      "loss": 0.5907,
      "step": 4778
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.878743715953627,
      "learning_rate": 3.3886285078952753e-07,
      "loss": 0.8711,
      "step": 4779
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.120720299590495,
      "learning_rate": 3.3725644011512125e-07,
      "loss": 0.9053,
      "step": 4780
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9566910237488425,
      "learning_rate": 3.356537808504634e-07,
      "loss": 0.9487,
      "step": 4781
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.083379242360614,
      "learning_rate": 3.3405487361776177e-07,
      "loss": 0.8902,
      "step": 4782
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.959134869653779,
      "learning_rate": 3.3245971903776654e-07,
      "loss": 0.8734,
      "step": 4783
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1205597234460223,
      "learning_rate": 3.308683177297711e-07,
      "loss": 0.9212,
      "step": 4784
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7789077692653145,
      "learning_rate": 3.292806703116125e-07,
      "loss": 0.6061,
      "step": 4785
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9587040763954989,
      "learning_rate": 3.2769677739966975e-07,
      "loss": 0.9209,
      "step": 4786
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.00778235392247,
      "learning_rate": 3.2611663960886665e-07,
      "loss": 0.9182,
      "step": 4787
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9362211448087369,
      "learning_rate": 3.245402575526646e-07,
      "loss": 0.8968,
      "step": 4788
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1053049131240362,
      "learning_rate": 3.2296763184306965e-07,
      "loss": 0.8807,
      "step": 4789
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8752951267985747,
      "learning_rate": 3.2139876309063233e-07,
      "loss": 0.6148,
      "step": 4790
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0666398303049847,
      "learning_rate": 3.198336519044376e-07,
      "loss": 0.9377,
      "step": 4791
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0661371780452256,
      "learning_rate": 3.182722988921161e-07,
      "loss": 0.8951,
      "step": 4792
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9688908763904488,
      "learning_rate": 3.167147046598418e-07,
      "loss": 0.9249,
      "step": 4793
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9624009318658313,
      "learning_rate": 3.151608698123232e-07,
      "loss": 0.9214,
      "step": 4794
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1246309885917436,
      "learning_rate": 3.1361079495281443e-07,
      "loss": 0.8808,
      "step": 4795
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.149235251976471,
      "learning_rate": 3.1206448068310635e-07,
      "loss": 0.9115,
      "step": 4796
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0009104552187864,
      "learning_rate": 3.1052192760353316e-07,
      "loss": 0.8489,
      "step": 4797
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1209954829712403,
      "learning_rate": 3.0898313631296586e-07,
      "loss": 0.8954,
      "step": 4798
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1669392054992371,
      "learning_rate": 3.0744810740881646e-07,
      "loss": 0.8785,
      "step": 4799
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9588693767847719,
      "learning_rate": 3.0591684148703617e-07,
      "loss": 0.9238,
      "step": 4800
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0186534531512037,
      "learning_rate": 3.043893391421149e-07,
      "loss": 0.9259,
      "step": 4801
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7915813936430822,
      "learning_rate": 3.0286560096708275e-07,
      "loss": 0.5774,
      "step": 4802
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0283678054936747,
      "learning_rate": 3.013456275535054e-07,
      "loss": 0.871,
      "step": 4803
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1009686874093114,
      "learning_rate": 2.998294194914897e-07,
      "loss": 0.8471,
      "step": 4804
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1295733582391962,
      "learning_rate": 2.983169773696815e-07,
      "loss": 0.9167,
      "step": 4805
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.029447191801493,
      "learning_rate": 2.968083017752599e-07,
      "loss": 0.9081,
      "step": 4806
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0793679955981859,
      "learning_rate": 2.953033932939464e-07,
      "loss": 0.9539,
      "step": 4807
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0450931390894513,
      "learning_rate": 2.938022525099982e-07,
      "loss": 0.8963,
      "step": 4808
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1140767818835584,
      "learning_rate": 2.9230488000621003e-07,
      "loss": 0.8846,
      "step": 4809
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0946655605242777,
      "learning_rate": 2.908112763639137e-07,
      "loss": 0.8721,
      "step": 4810
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.076247006141582,
      "learning_rate": 2.8932144216297643e-07,
      "loss": 0.9264,
      "step": 4811
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0989706930237433,
      "learning_rate": 2.878353779818044e-07,
      "loss": 0.9191,
      "step": 4812
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0503720147799098,
      "learning_rate": 2.863530843973372e-07,
      "loss": 0.9014,
      "step": 4813
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0691748393218086,
      "learning_rate": 2.848745619850546e-07,
      "loss": 0.9151,
      "step": 4814
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9192169750191784,
      "learning_rate": 2.833998113189662e-07,
      "loss": 0.6033,
      "step": 4815
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0180595708750224,
      "learning_rate": 2.8192883297162634e-07,
      "loss": 0.8889,
      "step": 4816
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0914717607746758,
      "learning_rate": 2.804616275141148e-07,
      "loss": 0.8863,
      "step": 4817
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.059811979632335,
      "learning_rate": 2.7899819551605256e-07,
      "loss": 0.9262,
      "step": 4818
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1610335389009772,
      "learning_rate": 2.7753853754559634e-07,
      "loss": 0.8883,
      "step": 4819
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8748969446848729,
      "learning_rate": 2.760826541694328e-07,
      "loss": 0.6112,
      "step": 4820
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8388749557741606,
      "learning_rate": 2.746305459527876e-07,
      "loss": 0.6221,
      "step": 4821
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.992923960268331,
      "learning_rate": 2.7318221345941865e-07,
      "loss": 0.9182,
      "step": 4822
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8234911537368306,
      "learning_rate": 2.717376572516184e-07,
      "loss": 0.6298,
      "step": 4823
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9671990976366243,
      "learning_rate": 2.7029687789021377e-07,
      "loss": 0.8641,
      "step": 4824
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9472855797489622,
      "learning_rate": 2.688598759345651e-07,
      "loss": 0.846,
      "step": 4825
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9156500247139084,
      "learning_rate": 2.67426651942565e-07,
      "loss": 0.853,
      "step": 4826
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0171800638401682,
      "learning_rate": 2.659972064706406e-07,
      "loss": 0.9138,
      "step": 4827
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9656740968434485,
      "learning_rate": 2.645715400737536e-07,
      "loss": 0.8864,
      "step": 4828
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0646478399400323,
      "learning_rate": 2.631496533053934e-07,
      "loss": 0.9112,
      "step": 4829
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0115655281101044,
      "learning_rate": 2.6173154671758847e-07,
      "loss": 0.9145,
      "step": 4830
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2674028517152471,
      "learning_rate": 2.603172208608962e-07,
      "loss": 0.8767,
      "step": 4831
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0148229585443844,
      "learning_rate": 2.589066762844039e-07,
      "loss": 0.9442,
      "step": 4832
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7460683328670543,
      "learning_rate": 2.57499913535737e-07,
      "loss": 0.6152,
      "step": 4833
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1291227630122125,
      "learning_rate": 2.5609693316104745e-07,
      "loss": 0.9553,
      "step": 4834
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0477925694614005,
      "learning_rate": 2.5469773570502063e-07,
      "loss": 0.9111,
      "step": 4835
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.12015586219211,
      "learning_rate": 2.5330232171087433e-07,
      "loss": 0.9489,
      "step": 4836
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.000249481955064,
      "learning_rate": 2.51910691720354e-07,
      "loss": 0.8774,
      "step": 4837
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0977700465495113,
      "learning_rate": 2.5052284627374077e-07,
      "loss": 0.926,
      "step": 4838
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0438594661267266,
      "learning_rate": 2.491387859098426e-07,
      "loss": 0.9348,
      "step": 4839
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0912297668406299,
      "learning_rate": 2.477585111659997e-07,
      "loss": 0.924,
      "step": 4840
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.028182507063602,
      "learning_rate": 2.463820225780811e-07,
      "loss": 0.9291,
      "step": 4841
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9256525836786541,
      "learning_rate": 2.4500932068049046e-07,
      "loss": 0.8353,
      "step": 4842
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1122446284687317,
      "learning_rate": 2.4364040600615477e-07,
      "loss": 0.9287,
      "step": 4843
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9980547193370121,
      "learning_rate": 2.422752790865346e-07,
      "loss": 0.8829,
      "step": 4844
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9266490896372006,
      "learning_rate": 2.409139404516203e-07,
      "loss": 0.885,
      "step": 4845
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1112209747764419,
      "learning_rate": 2.3955639062992696e-07,
      "loss": 0.9098,
      "step": 4846
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0403334762808853,
      "learning_rate": 2.3820263014850741e-07,
      "loss": 0.9049,
      "step": 4847
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9491311939746919,
      "learning_rate": 2.3685265953293345e-07,
      "loss": 0.8683,
      "step": 4848
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0911702343580238,
      "learning_rate": 2.3550647930731362e-07,
      "loss": 0.886,
      "step": 4849
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.028949071656366,
      "learning_rate": 2.3416408999427876e-07,
      "loss": 0.8818,
      "step": 4850
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8963198968018733,
      "learning_rate": 2.3282549211499307e-07,
      "loss": 0.876,
      "step": 4851
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.064426420471769,
      "learning_rate": 2.3149068618914417e-07,
      "loss": 0.8721,
      "step": 4852
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1974477298856248,
      "learning_rate": 2.3015967273494867e-07,
      "loss": 0.8949,
      "step": 4853
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1551989660742863,
      "learning_rate": 2.2883245226915652e-07,
      "loss": 0.9727,
      "step": 4854
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8832365786846434,
      "learning_rate": 2.2750902530703667e-07,
      "loss": 0.8784,
      "step": 4855
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0341895975830413,
      "learning_rate": 2.2618939236238924e-07,
      "loss": 0.9004,
      "step": 4856
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.941556262814989,
      "learning_rate": 2.2487355394754328e-07,
      "loss": 0.8978,
      "step": 4857
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.039428941880098,
      "learning_rate": 2.2356151057334908e-07,
      "loss": 0.9026,
      "step": 4858
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9400862167326456,
      "learning_rate": 2.2225326274919135e-07,
      "loss": 0.8871,
      "step": 4859
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9354821768361865,
      "learning_rate": 2.209488109829727e-07,
      "loss": 0.8627,
      "step": 4860
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1772359988405923,
      "learning_rate": 2.196481557811303e-07,
      "loss": 0.9367,
      "step": 4861
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0607160743694326,
      "learning_rate": 2.1835129764861907e-07,
      "loss": 0.8812,
      "step": 4862
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0096610140060545,
      "learning_rate": 2.1705823708892737e-07,
      "loss": 0.892,
      "step": 4863
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1518632826675508,
      "learning_rate": 2.1576897460406477e-07,
      "loss": 0.8875,
      "step": 4864
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.177523858280195,
      "learning_rate": 2.144835106945664e-07,
      "loss": 0.9812,
      "step": 4865
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9656804513142365,
      "learning_rate": 2.1320184585949532e-07,
      "loss": 0.86,
      "step": 4866
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0005847517636102,
      "learning_rate": 2.119239805964357e-07,
      "loss": 0.8887,
      "step": 4867
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0668013728430041,
      "learning_rate": 2.106499154015018e-07,
      "loss": 0.94,
      "step": 4868
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0896395896689839,
      "learning_rate": 2.0937965076932576e-07,
      "loss": 0.8958,
      "step": 4869
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.976628796472867,
      "learning_rate": 2.0811318719307194e-07,
      "loss": 0.9301,
      "step": 4870
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0123581312316572,
      "learning_rate": 2.0685052516442373e-07,
      "loss": 0.8814,
      "step": 4871
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0169453042057026,
      "learning_rate": 2.0559166517358787e-07,
      "loss": 0.9147,
      "step": 4872
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.074687326494141,
      "learning_rate": 2.0433660770930009e-07,
      "loss": 0.8737,
      "step": 4873
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0804619128817017,
      "learning_rate": 2.0308535325881616e-07,
      "loss": 0.8685,
      "step": 4874
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0915848819620566,
      "learning_rate": 2.0183790230791532e-07,
      "loss": 0.8958,
      "step": 4875
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0972059904962386,
      "learning_rate": 2.0059425534090128e-07,
      "loss": 0.9475,
      "step": 4876
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1317932420779455,
      "learning_rate": 1.9935441284059998e-07,
      "loss": 0.9327,
      "step": 4877
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1528221588002874,
      "learning_rate": 1.981183752883631e-07,
      "loss": 0.9183,
      "step": 4878
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9599402601046657,
      "learning_rate": 1.9688614316406006e-07,
      "loss": 0.8812,
      "step": 4879
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.219463510528957,
      "learning_rate": 1.9565771694608937e-07,
      "loss": 0.8875,
      "step": 4880
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0119564269012429,
      "learning_rate": 1.9443309711136393e-07,
      "loss": 0.91,
      "step": 4881
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3692831386656454,
      "learning_rate": 1.9321228413532788e-07,
      "loss": 0.8467,
      "step": 4882
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.022529705626104,
      "learning_rate": 1.9199527849194098e-07,
      "loss": 0.9076,
      "step": 4883
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7791696954984282,
      "learning_rate": 1.907820806536842e-07,
      "loss": 0.5885,
      "step": 4884
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.139656188631217,
      "learning_rate": 1.895726910915663e-07,
      "loss": 0.9371,
      "step": 4885
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.026958468119672,
      "learning_rate": 1.883671102751128e-07,
      "loss": 0.8896,
      "step": 4886
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.190022413762465,
      "learning_rate": 1.8716533867237153e-07,
      "loss": 0.927,
      "step": 4887
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1219897345020813,
      "learning_rate": 1.859673767499115e-07,
      "loss": 0.8371,
      "step": 4888
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0559455842914935,
      "learning_rate": 1.847732249728218e-07,
      "loss": 0.8315,
      "step": 4889
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9953339973900454,
      "learning_rate": 1.83582883804716e-07,
      "loss": 0.9521,
      "step": 4890
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9249116990251961,
      "learning_rate": 1.8239635370772223e-07,
      "loss": 0.8518,
      "step": 4891
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.129216925851611,
      "learning_rate": 1.8121363514249534e-07,
      "loss": 0.868,
      "step": 4892
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.20525360220915,
      "learning_rate": 1.8003472856820469e-07,
      "loss": 0.8786,
      "step": 4893
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9917239188889285,
      "learning_rate": 1.7885963444254528e-07,
      "loss": 0.9412,
      "step": 4894
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9633630242756944,
      "learning_rate": 1.7768835322172552e-07,
      "loss": 0.8623,
      "step": 4895
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1826666487722843,
      "learning_rate": 1.7652088536048052e-07,
      "loss": 0.8954,
      "step": 4896
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0092100491025384,
      "learning_rate": 1.7535723131206106e-07,
      "loss": 0.9137,
      "step": 4897
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9731696786421001,
      "learning_rate": 1.7419739152823468e-07,
      "loss": 0.8692,
      "step": 4898
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.103094963265573,
      "learning_rate": 1.7304136645929448e-07,
      "loss": 0.896,
      "step": 4899
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8014055275560514,
      "learning_rate": 1.7188915655404814e-07,
      "loss": 0.5842,
      "step": 4900
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9740855478192633,
      "learning_rate": 1.707407622598223e-07,
      "loss": 0.8703,
      "step": 4901
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.978261950203399,
      "learning_rate": 1.695961840224636e-07,
      "loss": 0.9119,
      "step": 4902
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0730795624270635,
      "learning_rate": 1.6845542228633772e-07,
      "loss": 0.9429,
      "step": 4903
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0260458773580634,
      "learning_rate": 1.6731847749432705e-07,
      "loss": 0.8253,
      "step": 4904
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0368063438188093,
      "learning_rate": 1.6618535008783075e-07,
      "loss": 0.9227,
      "step": 4905
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0024063784299566,
      "learning_rate": 1.6505604050677249e-07,
      "loss": 0.9167,
      "step": 4906
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1164357061476005,
      "learning_rate": 1.6393054918958373e-07,
      "loss": 0.9411,
      "step": 4907
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0118024252474291,
      "learning_rate": 1.6280887657322276e-07,
      "loss": 0.9792,
      "step": 4908
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8426348283811618,
      "learning_rate": 1.616910230931612e-07,
      "loss": 0.6511,
      "step": 4909
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.030393974430501,
      "learning_rate": 1.6057698918338526e-07,
      "loss": 0.9198,
      "step": 4910
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7699624278073619,
      "learning_rate": 1.5946677527640563e-07,
      "loss": 0.5956,
      "step": 4911
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0029845649875173,
      "learning_rate": 1.5836038180324198e-07,
      "loss": 0.9062,
      "step": 4912
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9968218367699789,
      "learning_rate": 1.5725780919343624e-07,
      "loss": 0.9032,
      "step": 4913
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0238403342416669,
      "learning_rate": 1.561590578750438e-07,
      "loss": 0.9002,
      "step": 4914
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0736378750812712,
      "learning_rate": 1.55064128274639e-07,
      "loss": 0.8633,
      "step": 4915
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0232028960737207,
      "learning_rate": 1.5397302081731069e-07,
      "loss": 0.8617,
      "step": 4916
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1506434249471154,
      "learning_rate": 1.5288573592666445e-07,
      "loss": 0.9027,
      "step": 4917
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0147099155052193,
      "learning_rate": 1.518022740248215e-07,
      "loss": 0.8902,
      "step": 4918
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8007409033264212,
      "learning_rate": 1.5072263553241872e-07,
      "loss": 0.6307,
      "step": 4919
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.896455900928522,
      "learning_rate": 1.4964682086861082e-07,
      "loss": 0.6595,
      "step": 4920
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1263040139249747,
      "learning_rate": 1.4857483045106258e-07,
      "loss": 0.9254,
      "step": 4921
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1154026955226102,
      "learning_rate": 1.475066646959611e-07,
      "loss": 0.9156,
      "step": 4922
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3257381459598312,
      "learning_rate": 1.4644232401800352e-07,
      "loss": 0.8959,
      "step": 4923
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9685183050213257,
      "learning_rate": 1.4538180883040264e-07,
      "loss": 0.8648,
      "step": 4924
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0869798869553828,
      "learning_rate": 1.4432511954488915e-07,
      "loss": 0.9035,
      "step": 4925
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8670677591577467,
      "learning_rate": 1.4327225657170485e-07,
      "loss": 0.5961,
      "step": 4926
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7857693347228528,
      "learning_rate": 1.4222322031960723e-07,
      "loss": 0.588,
      "step": 4927
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.007812085135415,
      "learning_rate": 1.411780111958694e-07,
      "loss": 0.935,
      "step": 4928
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0908163087894318,
      "learning_rate": 1.4013662960627562e-07,
      "loss": 0.8505,
      "step": 4929
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0762076088516468,
      "learning_rate": 1.3909907595512806e-07,
      "loss": 0.888,
      "step": 4930
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.168284787882422,
      "learning_rate": 1.3806535064524006e-07,
      "loss": 0.9629,
      "step": 4931
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2882900833636133,
      "learning_rate": 1.3703545407793951e-07,
      "loss": 0.934,
      "step": 4932
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0719977567127938,
      "learning_rate": 1.360093866530665e-07,
      "loss": 0.9443,
      "step": 4933
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9989342780240861,
      "learning_rate": 1.34987148768978e-07,
      "loss": 0.9328,
      "step": 4934
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9860606160078408,
      "learning_rate": 1.3396874082253986e-07,
      "loss": 0.8838,
      "step": 4935
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0149849355710443,
      "learning_rate": 1.3295416320913357e-07,
      "loss": 0.9088,
      "step": 4936
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0327852899228454,
      "learning_rate": 1.3194341632265518e-07,
      "loss": 0.9375,
      "step": 4937
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8356517128871513,
      "learning_rate": 1.3093650055550855e-07,
      "loss": 0.6065,
      "step": 4938
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0669160407536522,
      "learning_rate": 1.2993341629861432e-07,
      "loss": 0.9154,
      "step": 4939
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1406299637088813,
      "learning_rate": 1.2893416394140323e-07,
      "loss": 0.8893,
      "step": 4940
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.016541195917611,
      "learning_rate": 1.279387438718216e-07,
      "loss": 0.8477,
      "step": 4941
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0328500120384976,
      "learning_rate": 1.269471564763247e-07,
      "loss": 0.8991,
      "step": 4942
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8092326184972471,
      "learning_rate": 1.2595940213988024e-07,
      "loss": 0.5867,
      "step": 4943
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.116295594161477,
      "learning_rate": 1.2497548124597026e-07,
      "loss": 0.9243,
      "step": 4944
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9787561095890949,
      "learning_rate": 1.2399539417658368e-07,
      "loss": 0.8792,
      "step": 4945
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0407417028230057,
      "learning_rate": 1.2301914131222726e-07,
      "loss": 0.9153,
      "step": 4946
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0017874519859467,
      "learning_rate": 1.2204672303191335e-07,
      "loss": 0.9634,
      "step": 4947
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1076038851642958,
      "learning_rate": 1.2107813971317106e-07,
      "loss": 0.9412,
      "step": 4948
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0318542714054337,
      "learning_rate": 1.201133917320363e-07,
      "loss": 0.9144,
      "step": 4949
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0970027845499295,
      "learning_rate": 1.1915247946305498e-07,
      "loss": 0.9554,
      "step": 4950
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.905739271266224,
      "learning_rate": 1.1819540327929092e-07,
      "loss": 0.6194,
      "step": 4951
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9555809222616328,
      "learning_rate": 1.1724216355231022e-07,
      "loss": 0.8568,
      "step": 4952
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1369591267076822,
      "learning_rate": 1.1629276065219575e-07,
      "loss": 0.9075,
      "step": 4953
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2310416764871932,
      "learning_rate": 1.1534719494753821e-07,
      "loss": 0.9006,
      "step": 4954
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2078257533786039,
      "learning_rate": 1.144054668054373e-07,
      "loss": 0.9073,
      "step": 4955
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9719413281666469,
      "learning_rate": 1.1346757659150498e-07,
      "loss": 0.8947,
      "step": 4956
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9881962266778558,
      "learning_rate": 1.1253352466986334e-07,
      "loss": 0.906,
      "step": 4957
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1696067141368602,
      "learning_rate": 1.116033114031434e-07,
      "loss": 0.907,
      "step": 4958
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9657127744304307,
      "learning_rate": 1.1067693715248406e-07,
      "loss": 0.8716,
      "step": 4959
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8581266062299341,
      "learning_rate": 1.0975440227753764e-07,
      "loss": 0.6025,
      "step": 4960
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0900237446885885,
      "learning_rate": 1.0883570713646318e-07,
      "loss": 0.9031,
      "step": 4961
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9851674013183563,
      "learning_rate": 1.0792085208593095e-07,
      "loss": 0.873,
      "step": 4962
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.106621517786655,
      "learning_rate": 1.0700983748111792e-07,
      "loss": 0.9113,
      "step": 4963
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9658170480402981,
      "learning_rate": 1.061026636757101e-07,
      "loss": 0.8437,
      "step": 4964
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0296397187860855,
      "learning_rate": 1.0519933102190682e-07,
      "loss": 0.9745,
      "step": 4965
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8692211684400436,
      "learning_rate": 1.0429983987041092e-07,
      "loss": 0.592,
      "step": 4966
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0944419726152037,
      "learning_rate": 1.0340419057043527e-07,
      "loss": 0.9008,
      "step": 4967
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2137863446058739,
      "learning_rate": 1.0251238346970393e-07,
      "loss": 0.8824,
      "step": 4968
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0426015377095699,
      "learning_rate": 1.0162441891444441e-07,
      "loss": 0.8734,
      "step": 4969
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0030825276998163,
      "learning_rate": 1.007402972493976e-07,
      "loss": 0.8114,
      "step": 4970
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.108518891154298,
      "learning_rate": 9.986001881780783e-08,
      "loss": 0.9603,
      "step": 4971
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1589798232774418,
      "learning_rate": 9.898358396143171e-08,
      "loss": 0.9361,
      "step": 4972
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1847273479963936,
      "learning_rate": 9.811099302052928e-08,
      "loss": 0.9649,
      "step": 4973
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8768630362489067,
      "learning_rate": 9.72422463338718e-08,
      "loss": 0.6492,
      "step": 4974
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9071081043613481,
      "learning_rate": 9.637734423873612e-08,
      "loss": 0.5851,
      "step": 4975
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9329998842674732,
      "learning_rate": 9.55162870709081e-08,
      "loss": 0.8762,
      "step": 4976
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0320397591889825,
      "learning_rate": 9.465907516467698e-08,
      "loss": 0.9206,
      "step": 4977
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0308925803325169,
      "learning_rate": 9.380570885284546e-08,
      "loss": 0.878,
      "step": 4978
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6191451771234866,
      "learning_rate": 9.295618846671739e-08,
      "loss": 0.2792,
      "step": 4979
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1598750218713136,
      "learning_rate": 9.211051433610674e-08,
      "loss": 0.9576,
      "step": 4980
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0145632120142551,
      "learning_rate": 9.126868678933198e-08,
      "loss": 0.9345,
      "step": 4981
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.062774728308112,
      "learning_rate": 9.04307061532217e-08,
      "loss": 0.8394,
      "step": 4982
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9834976336851877,
      "learning_rate": 8.959657275310674e-08,
      "loss": 0.8995,
      "step": 4983
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3094880640413729,
      "learning_rate": 8.876628691282918e-08,
      "loss": 0.8738,
      "step": 4984
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2074698522030163,
      "learning_rate": 8.793984895473117e-08,
      "loss": 0.9595,
      "step": 4985
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9762211730025552,
      "learning_rate": 8.711725919966718e-08,
      "loss": 0.8878,
      "step": 4986
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9041854306195712,
      "learning_rate": 8.629851796699284e-08,
      "loss": 0.6381,
      "step": 4987
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9810421123748636,
      "learning_rate": 8.54836255745728e-08,
      "loss": 0.9008,
      "step": 4988
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9215649883149193,
      "learning_rate": 8.467258233877728e-08,
      "loss": 0.8601,
      "step": 4989
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8854703471592358,
      "learning_rate": 8.386538857447779e-08,
      "loss": 0.6008,
      "step": 4990
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0201472265935883,
      "learning_rate": 8.306204459505807e-08,
      "loss": 0.8682,
      "step": 4991
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.088359642628459,
      "learning_rate": 8.226255071240308e-08,
      "loss": 0.9678,
      "step": 4992
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8491041943797174,
      "learning_rate": 8.146690723690342e-08,
      "loss": 0.8422,
      "step": 4993
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1205385532013954,
      "learning_rate": 8.067511447745535e-08,
      "loss": 0.8959,
      "step": 4994
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9726105040350055,
      "learning_rate": 7.988717274146074e-08,
      "loss": 0.8899,
      "step": 4995
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0781234535175095,
      "learning_rate": 7.910308233482488e-08,
      "loss": 0.8895,
      "step": 4996
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1897179210176292,
      "learning_rate": 7.832284356195764e-08,
      "loss": 0.8821,
      "step": 4997
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9446356005999813,
      "learning_rate": 7.754645672577776e-08,
      "loss": 0.5889,
      "step": 4998
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8532753410852505,
      "learning_rate": 7.677392212770196e-08,
      "loss": 0.6316,
      "step": 4999
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0128240731082603,
      "learning_rate": 7.600524006765808e-08,
      "loss": 0.9202,
      "step": 5000
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0561174552497006,
      "learning_rate": 7.524041084407185e-08,
      "loss": 0.9287,
      "step": 5001
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.077418492772927,
      "learning_rate": 7.447943475387797e-08,
      "loss": 0.869,
      "step": 5002
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.973932921129932,
      "learning_rate": 7.372231209251346e-08,
      "loss": 0.8814,
      "step": 5003
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0134833928729168,
      "learning_rate": 7.296904315391873e-08,
      "loss": 0.8941,
      "step": 5004
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.93274241026181,
      "learning_rate": 7.221962823053874e-08,
      "loss": 0.9067,
      "step": 5005
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1817692410800316,
      "learning_rate": 7.147406761332298e-08,
      "loss": 0.8956,
      "step": 5006
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.128031702919094,
      "learning_rate": 7.073236159172325e-08,
      "loss": 0.9139,
      "step": 5007
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9997051946962805,
      "learning_rate": 6.999451045369587e-08,
      "loss": 0.8837,
      "step": 5008
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9718269871480854,
      "learning_rate": 6.926051448569948e-08,
      "loss": 0.9289,
      "step": 5009
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0123300943354818,
      "learning_rate": 6.853037397269724e-08,
      "loss": 0.6155,
      "step": 5010
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.924139739209956,
      "learning_rate": 6.78040891981524e-08,
      "loss": 0.8906,
      "step": 5011
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1188173567817956,
      "learning_rate": 6.70816604440383e-08,
      "loss": 0.8252,
      "step": 5012
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9994370216425945,
      "learning_rate": 6.63630879908217e-08,
      "loss": 0.8825,
      "step": 5013
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.132809501265799,
      "learning_rate": 6.564837211748054e-08,
      "loss": 0.9127,
      "step": 5014
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8585278362795054,
      "learning_rate": 6.493751310149177e-08,
      "loss": 0.6479,
      "step": 5015
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8115088839379314,
      "learning_rate": 6.42305112188335e-08,
      "loss": 0.6222,
      "step": 5016
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9469398147149385,
      "learning_rate": 6.352736674398951e-08,
      "loss": 0.8803,
      "step": 5017
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1756052273774158,
      "learning_rate": 6.282807994994477e-08,
      "loss": 0.9393,
      "step": 5018
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0270670678687368,
      "learning_rate": 6.213265110818656e-08,
      "loss": 0.9065,
      "step": 5019
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8247361106048381,
      "learning_rate": 6.144108048870335e-08,
      "loss": 0.5576,
      "step": 5020
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7686240905291675,
      "learning_rate": 6.075336835998813e-08,
      "loss": 0.5832,
      "step": 5021
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1372638406072617,
      "learning_rate": 6.00695149890329e-08,
      "loss": 0.8837,
      "step": 5022
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0278670741062859,
      "learning_rate": 5.938952064133419e-08,
      "loss": 0.8956,
      "step": 5023
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9933848588065776,
      "learning_rate": 5.871338558088857e-08,
      "loss": 0.8729,
      "step": 5024
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0347794310266893,
      "learning_rate": 5.8041110070194976e-08,
      "loss": 0.8899,
      "step": 5025
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1553489605151774,
      "learning_rate": 5.7372694370254614e-08,
      "loss": 0.9636,
      "step": 5026
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0028937072391548,
      "learning_rate": 5.67081387405688e-08,
      "loss": 0.9085,
      "step": 5027
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0080926166479376,
      "learning_rate": 5.6047443439141146e-08,
      "loss": 0.9254,
      "step": 5028
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1570361688980697,
      "learning_rate": 5.539060872247537e-08,
      "loss": 0.9506,
      "step": 5029
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0304903282325877,
      "learning_rate": 5.47376348455797e-08,
      "loss": 0.9086,
      "step": 5030
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0476671149414427,
      "learning_rate": 5.408852206195914e-08,
      "loss": 0.8831,
      "step": 5031
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.021535647383742,
      "learning_rate": 5.344327062362098e-08,
      "loss": 0.8719,
      "step": 5032
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9620517411918302,
      "learning_rate": 5.2801880781075954e-08,
      "loss": 0.8986,
      "step": 5033
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0892592709948627,
      "learning_rate": 5.216435278333376e-08,
      "loss": 0.8714,
      "step": 5034
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9715109462961893,
      "learning_rate": 5.153068687790197e-08,
      "loss": 0.8832,
      "step": 5035
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0459113665635569,
      "learning_rate": 5.0900883310794903e-08,
      "loss": 0.9327,
      "step": 5036
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9443689787290565,
      "learning_rate": 5.0274942326521414e-08,
      "loss": 0.902,
      "step": 5037
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9770050762568331,
      "learning_rate": 4.9652864168096e-08,
      "loss": 0.904,
      "step": 5038
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0922643030049282,
      "learning_rate": 4.9034649077027706e-08,
      "loss": 0.9176,
      "step": 5039
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1251684230822159,
      "learning_rate": 4.84202972933312e-08,
      "loss": 0.8766,
      "step": 5040
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.010958211383922,
      "learning_rate": 4.7809809055517906e-08,
      "loss": 0.9015,
      "step": 5041
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0591686208105682,
      "learning_rate": 4.720318460060047e-08,
      "loss": 0.6715,
      "step": 5042
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1535169399911507,
      "learning_rate": 4.6600424164091606e-08,
      "loss": 0.8858,
      "step": 5043
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0716886659662874,
      "learning_rate": 4.6001527980004125e-08,
      "loss": 0.8831,
      "step": 5044
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9226533051119974,
      "learning_rate": 4.54064962808487e-08,
      "loss": 0.8834,
      "step": 5045
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1534401268542858,
      "learning_rate": 4.4815329297639434e-08,
      "loss": 0.9059,
      "step": 5046
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9686288866063603,
      "learning_rate": 4.422802725988606e-08,
      "loss": 0.8189,
      "step": 5047
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1731978599216561,
      "learning_rate": 4.364459039559843e-08,
      "loss": 0.9608,
      "step": 5048
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8604114755981862,
      "learning_rate": 4.3065018931289784e-08,
      "loss": 0.8837,
      "step": 5049
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9975976124105185,
      "learning_rate": 4.248931309196791e-08,
      "loss": 0.921,
      "step": 5050
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9779719024195391,
      "learning_rate": 4.1917473101140696e-08,
      "loss": 0.9403,
      "step": 5051
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.018212390628429,
      "learning_rate": 4.134949918081832e-08,
      "loss": 0.9076,
      "step": 5052
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.967739158388003,
      "learning_rate": 4.0785391551506626e-08,
      "loss": 0.962,
      "step": 5053
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9478654847359663,
      "learning_rate": 4.022515043221154e-08,
      "loss": 0.8941,
      "step": 5054
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.016973940535608,
      "learning_rate": 3.966877604043795e-08,
      "loss": 0.896,
      "step": 5055
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.012155561028714,
      "learning_rate": 3.9116268592189755e-08,
      "loss": 0.9179,
      "step": 5056
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1088402476750356,
      "learning_rate": 3.8567628301969806e-08,
      "loss": 0.9462,
      "step": 5057
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0773904987542482,
      "learning_rate": 3.802285538277772e-08,
      "loss": 0.9287,
      "step": 5058
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0143163482237465,
      "learning_rate": 3.748195004611543e-08,
      "loss": 0.8898,
      "step": 5059
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1696058809396888,
      "learning_rate": 3.69449125019794e-08,
      "loss": 0.9207,
      "step": 5060
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9548666728253901,
      "learning_rate": 3.6411742958866184e-08,
      "loss": 0.641,
      "step": 5061
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9959553222678244,
      "learning_rate": 3.588244162377019e-08,
      "loss": 0.8869,
      "step": 5062
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1292528610128525,
      "learning_rate": 3.5357008702185945e-08,
      "loss": 0.8872,
      "step": 5063
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9805864587831892,
      "learning_rate": 3.483544439810249e-08,
      "loss": 0.8961,
      "step": 5064
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1435690322272722,
      "learning_rate": 3.4317748914011187e-08,
      "loss": 0.9255,
      "step": 5065
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0135501454741098,
      "learning_rate": 3.3803922450897917e-08,
      "loss": 0.909,
      "step": 5066
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1850277979605508,
      "learning_rate": 3.329396520824757e-08,
      "loss": 0.8452,
      "step": 5067
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.138588012572821,
      "learning_rate": 3.2787877384045095e-08,
      "loss": 0.9461,
      "step": 5068
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9854825416473177,
      "learning_rate": 3.228565917476889e-08,
      "loss": 0.9473,
      "step": 5069
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9972708191775599,
      "learning_rate": 3.178731077539743e-08,
      "loss": 0.8395,
      "step": 5070
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1912562807468945,
      "learning_rate": 3.129283237940928e-08,
      "loss": 0.9119,
      "step": 5071
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0283334768345647,
      "learning_rate": 3.080222417877421e-08,
      "loss": 0.9371,
      "step": 5072
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9259378673216474,
      "learning_rate": 3.031548636396764e-08,
      "loss": 0.8876,
      "step": 5073
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0842218188392752,
      "learning_rate": 2.983261912395397e-08,
      "loss": 0.8657,
      "step": 5074
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.102663758185133,
      "learning_rate": 2.9353622646199898e-08,
      "loss": 0.8961,
      "step": 5075
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0529324054682938,
      "learning_rate": 2.8878497116671124e-08,
      "loss": 0.8999,
      "step": 5076
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0298186740331317,
      "learning_rate": 2.8407242719823424e-08,
      "loss": 0.8881,
      "step": 5077
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1024779039040846,
      "learning_rate": 2.7939859638617118e-08,
      "loss": 0.9163,
      "step": 5078
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0757676506013991,
      "learning_rate": 2.7476348054504832e-08,
      "loss": 0.9632,
      "step": 5079
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9766913283726577,
      "learning_rate": 2.7016708147439285e-08,
      "loss": 0.9038,
      "step": 5080
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0464657178723589,
      "learning_rate": 2.6560940095866626e-08,
      "loss": 0.9202,
      "step": 5081
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1055274676509936,
      "learning_rate": 2.6109044076733092e-08,
      "loss": 0.9322,
      "step": 5082
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0871189722006935,
      "learning_rate": 2.5661020265479452e-08,
      "loss": 0.8555,
      "step": 5083
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9041997667391919,
      "learning_rate": 2.5216868836043242e-08,
      "loss": 0.8348,
      "step": 5084
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0119594318528653,
      "learning_rate": 2.4776589960862074e-08,
      "loss": 0.9046,
      "step": 5085
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8006269006726108,
      "learning_rate": 2.434018381086589e-08,
      "loss": 0.6263,
      "step": 5086
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.056524238482987,
      "learning_rate": 2.3907650555481387e-08,
      "loss": 0.9065,
      "step": 5087
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.972079261495623,
      "learning_rate": 2.3478990362634235e-08,
      "loss": 0.907,
      "step": 5088
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0661384223052004,
      "learning_rate": 2.3054203398743537e-08,
      "loss": 0.9179,
      "step": 5089
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0108487460039517,
      "learning_rate": 2.263328982872959e-08,
      "loss": 0.9015,
      "step": 5090
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.912276016916757,
      "learning_rate": 2.221624981600168e-08,
      "loss": 0.8548,
      "step": 5091
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.151437570193738,
      "learning_rate": 2.1803083522471402e-08,
      "loss": 0.9606,
      "step": 5092
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9939805963664111,
      "learning_rate": 2.1393791108542672e-08,
      "loss": 0.8914,
      "step": 5093
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9979761945692214,
      "learning_rate": 2.098837273311838e-08,
      "loss": 0.9101,
      "step": 5094
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0925930440408862,
      "learning_rate": 2.058682855359595e-08,
      "loss": 0.8924,
      "step": 5095
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1284430850579295,
      "learning_rate": 2.0189158725867353e-08,
      "loss": 0.922,
      "step": 5096
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2014472441286137,
      "learning_rate": 1.979536340432131e-08,
      "loss": 0.8972,
      "step": 5097
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0220705806513863,
      "learning_rate": 1.9405442741844415e-08,
      "loss": 0.9358,
      "step": 5098
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0460156522868091,
      "learning_rate": 1.9019396889816688e-08,
      "loss": 0.9235,
      "step": 5099
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9512382617318933,
      "learning_rate": 1.8637225998114904e-08,
      "loss": 0.8765,
      "step": 5100
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9515477036240275,
      "learning_rate": 1.825893021510927e-08,
      "loss": 0.8707,
      "step": 5101
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9779947159267324,
      "learning_rate": 1.7884509687668972e-08,
      "loss": 0.8893,
      "step": 5102
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9501238111898623,
      "learning_rate": 1.7513964561156617e-08,
      "loss": 0.8663,
      "step": 5103
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0480751867083438,
      "learning_rate": 1.714729497942935e-08,
      "loss": 0.8558,
      "step": 5104
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0101874118971106,
      "learning_rate": 1.6784501084843307e-08,
      "loss": 0.9047,
      "step": 5105
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0593787496969513,
      "learning_rate": 1.6425583018244706e-08,
      "loss": 0.9456,
      "step": 5106
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9612874987007172,
      "learning_rate": 1.607054091897986e-08,
      "loss": 0.8998,
      "step": 5107
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0772403915993713,
      "learning_rate": 1.57193749248874e-08,
      "loss": 0.9381,
      "step": 5108
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0940396017032818,
      "learning_rate": 1.537208517230271e-08,
      "loss": 0.8956,
      "step": 5109
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9801622741470157,
      "learning_rate": 1.5028671796055715e-08,
      "loss": 0.8882,
      "step": 5110
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0824722594597955,
      "learning_rate": 1.4689134929470884e-08,
      "loss": 0.8583,
      "step": 5111
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1424866996361758,
      "learning_rate": 1.435347470436832e-08,
      "loss": 0.8318,
      "step": 5112
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7973858404676697,
      "learning_rate": 1.4021691251062675e-08,
      "loss": 0.6279,
      "step": 5113
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7902784518269048,
      "learning_rate": 1.3693784698363133e-08,
      "loss": 0.6312,
      "step": 5114
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9229995704722658,
      "learning_rate": 1.3369755173575639e-08,
      "loss": 0.644,
      "step": 5115
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0686265854919546,
      "learning_rate": 1.3049602802498451e-08,
      "loss": 0.8575,
      "step": 5116
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.179516562809919,
      "learning_rate": 1.273332770942659e-08,
      "loss": 0.8712,
      "step": 5117
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9823121749992133,
      "learning_rate": 1.2420930017148503e-08,
      "loss": 0.8914,
      "step": 5118
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.078413700498442,
      "learning_rate": 1.2112409846947171e-08,
      "loss": 0.9406,
      "step": 5119
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.098728550024428,
      "learning_rate": 1.1807767318602337e-08,
      "loss": 0.8899,
      "step": 5120
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9438945939410808,
      "learning_rate": 1.150700255038606e-08,
      "loss": 0.9388,
      "step": 5121
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9959589792220979,
      "learning_rate": 1.1210115659063825e-08,
      "loss": 0.8929,
      "step": 5122
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9971905951239338,
      "learning_rate": 1.0917106759900097e-08,
      "loss": 0.8605,
      "step": 5123
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0933619858293888,
      "learning_rate": 1.0627975966649439e-08,
      "loss": 0.9198,
      "step": 5124
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0132567526360452,
      "learning_rate": 1.034272339156206e-08,
      "loss": 0.8577,
      "step": 5125
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9636496848751225,
      "learning_rate": 1.0061349145383814e-08,
      "loss": 0.8827,
      "step": 5126
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0485702205105456,
      "learning_rate": 9.783853337353987e-09,
      "loss": 0.952,
      "step": 5127
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9590709294203988,
      "learning_rate": 9.510236075205292e-09,
      "loss": 0.883,
      "step": 5128
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0069613500024697,
      "learning_rate": 9.240497465164978e-09,
      "loss": 0.8597,
      "step": 5129
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1470615780611668,
      "learning_rate": 8.974637611955939e-09,
      "loss": 0.8988,
      "step": 5130
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1124572088351279,
      "learning_rate": 8.712656618793391e-09,
      "loss": 0.9531,
      "step": 5131
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0258752929709203,
      "learning_rate": 8.454554587388198e-09,
      "loss": 0.8977,
      "step": 5132
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9772874014462872,
      "learning_rate": 8.200331617943535e-09,
      "loss": 0.8848,
      "step": 5133
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0751741971983748,
      "learning_rate": 7.949987809158232e-09,
      "loss": 0.9086,
      "step": 5134
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9576849826106739,
      "learning_rate": 7.703523258223433e-09,
      "loss": 0.8757,
      "step": 5135
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0823253522820695,
      "learning_rate": 7.460938060825929e-09,
      "loss": 0.8723,
      "step": 5136
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1454133131361153,
      "learning_rate": 7.222232311145938e-09,
      "loss": 0.8942,
      "step": 5137
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8934074408211088,
      "learning_rate": 6.987406101855998e-09,
      "loss": 0.8745,
      "step": 5138
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0531341907308176,
      "learning_rate": 6.756459524125403e-09,
      "loss": 0.8894,
      "step": 5139
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0655414385453783,
      "learning_rate": 6.5293926676135434e-09,
      "loss": 0.9241,
      "step": 5140
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0672331595584326,
      "learning_rate": 6.306205620477679e-09,
      "loss": 0.9079,
      "step": 5141
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0226053458837852,
      "learning_rate": 6.086898469365166e-09,
      "loss": 0.9443,
      "step": 5142
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8417612911047423,
      "learning_rate": 5.871471299419007e-09,
      "loss": 0.594,
      "step": 5143
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0513571207931216,
      "learning_rate": 5.6599241942767445e-09,
      "loss": 0.9041,
      "step": 5144
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0518372856777078,
      "learning_rate": 5.452257236066017e-09,
      "loss": 0.9419,
      "step": 5145
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.989658642256631,
      "learning_rate": 5.248470505412328e-09,
      "loss": 0.6779,
      "step": 5146
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0952811078669273,
      "learning_rate": 5.0485640814312844e-09,
      "loss": 0.9388,
      "step": 5147
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1570398250441736,
      "learning_rate": 4.8525380417330234e-09,
      "loss": 0.9091,
      "step": 5148
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9518561003694546,
      "learning_rate": 4.660392462424446e-09,
      "loss": 0.9117,
      "step": 5149
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1177871580071683,
      "learning_rate": 4.472127418099215e-09,
      "loss": 0.902,
      "step": 5150
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0984461175387572,
      "learning_rate": 4.287742981851084e-09,
      "loss": 0.877,
      "step": 5151
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7945782134735652,
      "learning_rate": 4.1072392252639034e-09,
      "loss": 0.5907,
      "step": 5152
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9139258415512043,
      "learning_rate": 3.930616218414951e-09,
      "loss": 0.8853,
      "step": 5153
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.026528909212039,
      "learning_rate": 3.757874029874931e-09,
      "loss": 0.9141,
      "step": 5154
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1148194035382475,
      "learning_rate": 3.5890127267090844e-09,
      "loss": 0.8943,
      "step": 5155
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9962450384738305,
      "learning_rate": 3.424032374476083e-09,
      "loss": 0.8782,
      "step": 5156
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1148351421956182,
      "learning_rate": 3.2629330372246915e-09,
      "loss": 0.8448,
      "step": 5157
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1237931158478731,
      "learning_rate": 3.105714777501545e-09,
      "loss": 0.8657,
      "step": 5158
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8717029944395336,
      "learning_rate": 2.9523776563422644e-09,
      "loss": 0.597,
      "step": 5159
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.076684279290529,
      "learning_rate": 2.802921733278119e-09,
      "loss": 0.9359,
      "step": 5160
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8259691265524233,
      "learning_rate": 2.657347066333804e-09,
      "loss": 0.6566,
      "step": 5161
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8592137372367029,
      "learning_rate": 2.5156537120263335e-09,
      "loss": 0.6153,
      "step": 5162
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0270981005418856,
      "learning_rate": 2.3778417253650376e-09,
      "loss": 0.8965,
      "step": 5163
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1928366465987765,
      "learning_rate": 2.2439111598537844e-09,
      "loss": 0.9616,
      "step": 5164
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7656941216939974,
      "learning_rate": 2.113862067488759e-09,
      "loss": 0.6214,
      "step": 5165
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0805852760592782,
      "learning_rate": 1.987694498760684e-09,
      "loss": 0.8726,
      "step": 5166
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0265381595332932,
      "learning_rate": 1.865408502650379e-09,
      "loss": 0.8927,
      "step": 5167
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0518270585315115,
      "learning_rate": 1.747004126635421e-09,
      "loss": 0.8247,
      "step": 5168
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0166263482253974,
      "learning_rate": 1.6324814166823744e-09,
      "loss": 0.902,
      "step": 5169
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1286849614329544,
      "learning_rate": 1.5218404172545609e-09,
      "loss": 0.8883,
      "step": 5170
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1233337699316706,
      "learning_rate": 1.415081171305399e-09,
      "loss": 0.8339,
      "step": 5171
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.149540717569594,
      "learning_rate": 1.3122037202828452e-09,
      "loss": 0.8829,
      "step": 5172
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9730780847261752,
      "learning_rate": 1.2132081041282829e-09,
      "loss": 0.8694,
      "step": 5173
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1952620907435705,
      "learning_rate": 1.1180943612754124e-09,
      "loss": 0.8997,
      "step": 5174
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0884084676131094,
      "learning_rate": 1.026862528649142e-09,
      "loss": 0.9435,
      "step": 5175
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9857364046916567,
      "learning_rate": 9.39512641668916e-10,
      "loss": 0.8828,
      "step": 5176
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0126701382199834,
      "learning_rate": 8.560447342487177e-10,
      "loss": 0.9097,
      "step": 5177
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0112055375264493,
      "learning_rate": 7.764588387915161e-10,
      "loss": 0.8795,
      "step": 5178
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9695023873491678,
      "learning_rate": 7.007549861970387e-10,
      "loss": 0.8666,
      "step": 5179
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.106484565749829,
      "learning_rate": 6.289332058551089e-10,
      "loss": 0.9588,
      "step": 5180
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0272883073456105,
      "learning_rate": 5.609935256500887e-10,
      "loss": 0.9251,
      "step": 5181
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9729458308886018,
      "learning_rate": 4.969359719586563e-10,
      "loss": 0.8789,
      "step": 5182
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9357269891202515,
      "learning_rate": 4.3676056964869764e-10,
      "loss": 0.882,
      "step": 5183
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7914525408965669,
      "learning_rate": 3.804673420837457e-10,
      "loss": 0.5835,
      "step": 5184
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0293467626525035,
      "learning_rate": 3.2805631111743064e-10,
      "loss": 0.9296,
      "step": 5185
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.990586238764238,
      "learning_rate": 2.795274971001405e-10,
      "loss": 0.9256,
      "step": 5186
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.145394016550248,
      "learning_rate": 2.3488091886902933e-10,
      "loss": 0.9455,
      "step": 5187
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.053949498552156,
      "learning_rate": 1.941165937602296e-10,
      "loss": 0.8785,
      "step": 5188
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9899092158974745,
      "learning_rate": 1.5723453759886042e-10,
      "loss": 0.9303,
      "step": 5189
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1435231339640153,
      "learning_rate": 1.2423476470346808e-10,
      "loss": 0.8897,
      "step": 5190
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9056693514539293,
      "learning_rate": 9.511728788602625e-11,
      "loss": 0.8852,
      "step": 5191
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9013111783167348,
      "learning_rate": 6.988211845082582e-11,
      "loss": 0.8483,
      "step": 5192
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0202326000810378,
      "learning_rate": 4.852926619447473e-11,
      "loss": 0.8728,
      "step": 5193
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.982722936367773,
      "learning_rate": 3.105873940811854e-11,
      "loss": 0.8691,
      "step": 5194
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0581710561414677,
      "learning_rate": 1.7470544874109706e-11,
      "loss": 0.8763,
      "step": 5195
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9892732177977154,
      "learning_rate": 7.764687866007592e-12,
      "loss": 0.908,
      "step": 5196
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.962770800601801,
      "learning_rate": 1.9411721552398123e-12,
      "loss": 0.9077,
      "step": 5197
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0126609397515678,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 5198
    },
    {
      "epoch": 1.0,
      "step": 5198,
      "total_flos": 1.7329815663937585e+19,
      "train_loss": 0.9119912797657515,
      "train_runtime": 43263.1026,
      "train_samples_per_second": 15.378,
      "train_steps_per_second": 0.12
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 5198,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 1.7329815663937585e+19,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}