{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9996633158578231,
  "eval_steps": 500,
  "global_step": 2598,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 2.362,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 2.1768,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 2.2477,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 2.2123,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.0,
      "learning_rate": 0.0,
      "loss": 2.4623,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 33.83708572387695,
      "learning_rate": 2.564102564102564e-07,
      "loss": 2.2016,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 40.06373977661133,
      "learning_rate": 5.128205128205128e-07,
      "loss": 2.3974,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 40.06584930419922,
      "learning_rate": 7.692307692307694e-07,
      "loss": 2.3746,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 24.79712677001953,
      "learning_rate": 1.0256410256410257e-06,
      "loss": 1.96,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 39.713478088378906,
      "learning_rate": 1.282051282051282e-06,
      "loss": 2.2416,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 27.37666893005371,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 1.9558,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 29.040416717529297,
      "learning_rate": 1.794871794871795e-06,
      "loss": 2.0595,
      "step": 12
    },
    {
      "epoch": 0.01,
      "grad_norm": 24.32331085205078,
      "learning_rate": 2.0512820512820513e-06,
      "loss": 1.8376,
      "step": 13
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.513137817382812,
      "learning_rate": 2.307692307692308e-06,
      "loss": 1.7737,
      "step": 14
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.9006876945495605,
      "learning_rate": 2.564102564102564e-06,
      "loss": 1.6473,
      "step": 15
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.523036003112793,
      "learning_rate": 2.8205128205128207e-06,
      "loss": 1.6891,
      "step": 16
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.945356845855713,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 1.5713,
      "step": 17
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.158085823059082,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.5933,
      "step": 18
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.588412761688232,
      "learning_rate": 3.58974358974359e-06,
      "loss": 1.4806,
      "step": 19
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.631389617919922,
      "learning_rate": 3.846153846153847e-06,
      "loss": 1.5684,
      "step": 20
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.449581146240234,
      "learning_rate": 4.102564102564103e-06,
      "loss": 1.502,
      "step": 21
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7876343727111816,
      "learning_rate": 4.358974358974359e-06,
      "loss": 1.5208,
      "step": 22
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.148611307144165,
      "learning_rate": 4.615384615384616e-06,
      "loss": 1.4228,
      "step": 23
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.127854108810425,
      "learning_rate": 4.871794871794872e-06,
      "loss": 1.5041,
      "step": 24
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1349451541900635,
      "learning_rate": 5.128205128205128e-06,
      "loss": 1.4222,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1183223724365234,
      "learning_rate": 5.384615384615385e-06,
      "loss": 1.4783,
      "step": 26
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9035189151763916,
      "learning_rate": 5.641025641025641e-06,
      "loss": 1.5234,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0864882469177246,
      "learning_rate": 5.897435897435898e-06,
      "loss": 1.486,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.10664963722229,
      "learning_rate": 6.153846153846155e-06,
      "loss": 1.4704,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.530993700027466,
      "learning_rate": 6.410256410256412e-06,
      "loss": 1.4019,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.72135853767395,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.4286,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.577258825302124,
      "learning_rate": 6.923076923076923e-06,
      "loss": 1.4464,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.830336332321167,
      "learning_rate": 7.17948717948718e-06,
      "loss": 1.3781,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5885684490203857,
      "learning_rate": 7.435897435897437e-06,
      "loss": 1.4381,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.764780044555664,
      "learning_rate": 7.692307692307694e-06,
      "loss": 1.4851,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3809397220611572,
      "learning_rate": 7.948717948717949e-06,
      "loss": 1.4027,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4772183895111084,
      "learning_rate": 8.205128205128205e-06,
      "loss": 1.3921,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4958701133728027,
      "learning_rate": 8.461538461538462e-06,
      "loss": 1.4236,
      "step": 38
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.839393138885498,
      "learning_rate": 8.717948717948719e-06,
      "loss": 1.3929,
      "step": 39
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6671385765075684,
      "learning_rate": 8.974358974358976e-06,
      "loss": 1.4167,
      "step": 40
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0103869438171387,
      "learning_rate": 9.230769230769232e-06,
      "loss": 1.3822,
      "step": 41
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3385279178619385,
      "learning_rate": 9.487179487179487e-06,
      "loss": 1.4018,
      "step": 42
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.652867317199707,
      "learning_rate": 9.743589743589744e-06,
      "loss": 1.4804,
      "step": 43
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.36625337600708,
      "learning_rate": 1e-05,
      "loss": 1.4103,
      "step": 44
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6790246963500977,
      "learning_rate": 1.0256410256410256e-05,
      "loss": 1.3868,
      "step": 45
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.194364309310913,
      "learning_rate": 1.0512820512820514e-05,
      "loss": 1.3357,
      "step": 46
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4601776599884033,
      "learning_rate": 1.076923076923077e-05,
      "loss": 1.4171,
      "step": 47
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4877281188964844,
      "learning_rate": 1.1025641025641028e-05,
      "loss": 1.4049,
      "step": 48
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.689100503921509,
      "learning_rate": 1.1282051282051283e-05,
      "loss": 1.3143,
      "step": 49
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4859206676483154,
      "learning_rate": 1.1538461538461538e-05,
      "loss": 1.3807,
      "step": 50
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.498077154159546,
      "learning_rate": 1.1794871794871796e-05,
      "loss": 1.3092,
      "step": 51
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2707862854003906,
      "learning_rate": 1.2051282051282051e-05,
      "loss": 1.3852,
      "step": 52
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.363126516342163,
      "learning_rate": 1.230769230769231e-05,
      "loss": 1.3685,
      "step": 53
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6271862983703613,
      "learning_rate": 1.2564102564102565e-05,
      "loss": 1.3243,
      "step": 54
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6401655673980713,
      "learning_rate": 1.2820512820512823e-05,
      "loss": 1.4403,
      "step": 55
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3446950912475586,
      "learning_rate": 1.3076923076923078e-05,
      "loss": 1.3215,
      "step": 56
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4611451625823975,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 1.2968,
      "step": 57
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5799314975738525,
      "learning_rate": 1.3589743589743592e-05,
      "loss": 1.3113,
      "step": 58
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2068967819213867,
      "learning_rate": 1.3846153846153847e-05,
      "loss": 1.3427,
      "step": 59
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7793164253234863,
      "learning_rate": 1.4102564102564105e-05,
      "loss": 1.4534,
      "step": 60
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.318920373916626,
      "learning_rate": 1.435897435897436e-05,
      "loss": 1.3046,
      "step": 61
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6745667457580566,
      "learning_rate": 1.4615384615384615e-05,
      "loss": 1.3696,
      "step": 62
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.247668504714966,
      "learning_rate": 1.4871794871794874e-05,
      "loss": 1.4079,
      "step": 63
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6244540214538574,
      "learning_rate": 1.5128205128205129e-05,
      "loss": 1.2919,
      "step": 64
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3556137084960938,
      "learning_rate": 1.5384615384615387e-05,
      "loss": 1.3003,
      "step": 65
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2927372455596924,
      "learning_rate": 1.5641025641025644e-05,
      "loss": 1.3145,
      "step": 66
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0944035053253174,
      "learning_rate": 1.5897435897435897e-05,
      "loss": 1.286,
      "step": 67
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2618911266326904,
      "learning_rate": 1.6153846153846154e-05,
      "loss": 1.2792,
      "step": 68
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.104365110397339,
      "learning_rate": 1.641025641025641e-05,
      "loss": 1.3332,
      "step": 69
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1352286338806152,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.3686,
      "step": 70
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3988118171691895,
      "learning_rate": 1.6923076923076924e-05,
      "loss": 1.2754,
      "step": 71
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4062960147857666,
      "learning_rate": 1.717948717948718e-05,
      "loss": 1.2616,
      "step": 72
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.561047315597534,
      "learning_rate": 1.7435897435897438e-05,
      "loss": 1.2901,
      "step": 73
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.542564868927002,
      "learning_rate": 1.7692307692307694e-05,
      "loss": 1.3244,
      "step": 74
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1995208263397217,
      "learning_rate": 1.794871794871795e-05,
      "loss": 1.3989,
      "step": 75
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4986228942871094,
      "learning_rate": 1.8205128205128208e-05,
      "loss": 1.3504,
      "step": 76
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.329167366027832,
      "learning_rate": 1.8461538461538465e-05,
      "loss": 1.2915,
      "step": 77
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.371638774871826,
      "learning_rate": 1.8717948717948718e-05,
      "loss": 1.2933,
      "step": 78
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3216817378997803,
      "learning_rate": 1.8974358974358975e-05,
      "loss": 1.3679,
      "step": 79
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3725318908691406,
      "learning_rate": 1.923076923076923e-05,
      "loss": 1.3317,
      "step": 80
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2525856494903564,
      "learning_rate": 1.9487179487179488e-05,
      "loss": 1.3539,
      "step": 81
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.055257558822632,
      "learning_rate": 1.9743589743589745e-05,
      "loss": 1.3181,
      "step": 82
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.290231466293335,
      "learning_rate": 2e-05,
      "loss": 1.2593,
      "step": 83
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1911017894744873,
      "learning_rate": 1.9999992229148463e-05,
      "loss": 1.3189,
      "step": 84
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1511759757995605,
      "learning_rate": 1.999996891660592e-05,
      "loss": 1.2489,
      "step": 85
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.243377685546875,
      "learning_rate": 1.9999930062408612e-05,
      "loss": 1.2889,
      "step": 86
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.359112024307251,
      "learning_rate": 1.9999875666616918e-05,
      "loss": 1.3467,
      "step": 87
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1951794624328613,
      "learning_rate": 1.9999805729315383e-05,
      "loss": 1.2621,
      "step": 88
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.17930006980896,
      "learning_rate": 1.99997202506127e-05,
      "loss": 1.3218,
      "step": 89
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.5065624713897705,
      "learning_rate": 1.9999619230641714e-05,
      "loss": 1.2296,
      "step": 90
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3997745513916016,
      "learning_rate": 1.9999502669559432e-05,
      "loss": 1.2499,
      "step": 91
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2092480659484863,
      "learning_rate": 1.9999370567547008e-05,
      "loss": 1.3256,
      "step": 92
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0892322063446045,
      "learning_rate": 1.999922292480975e-05,
      "loss": 1.2495,
      "step": 93
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.5424158573150635,
      "learning_rate": 1.999905974157712e-05,
      "loss": 1.227,
      "step": 94
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.323364496231079,
      "learning_rate": 1.9998881018102735e-05,
      "loss": 1.3462,
      "step": 95
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4222540855407715,
      "learning_rate": 1.999868675466436e-05,
      "loss": 1.2474,
      "step": 96
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.257307291030884,
      "learning_rate": 1.9998476951563914e-05,
      "loss": 1.3381,
      "step": 97
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2174766063690186,
      "learning_rate": 1.9998251609127465e-05,
      "loss": 1.3458,
      "step": 98
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2595882415771484,
      "learning_rate": 1.9998010727705237e-05,
      "loss": 1.323,
      "step": 99
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3427112102508545,
      "learning_rate": 1.9997754307671597e-05,
      "loss": 1.3268,
      "step": 100
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.398885488510132,
      "learning_rate": 1.999748234942507e-05,
      "loss": 1.3528,
      "step": 101
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1256563663482666,
      "learning_rate": 1.9997194853388316e-05,
      "loss": 1.1954,
      "step": 102
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1678531169891357,
      "learning_rate": 1.9996891820008165e-05,
      "loss": 1.288,
      "step": 103
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1746480464935303,
      "learning_rate": 1.9996573249755573e-05,
      "loss": 1.247,
      "step": 104
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2657346725463867,
      "learning_rate": 1.999623914312566e-05,
      "loss": 1.3036,
      "step": 105
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1539113521575928,
      "learning_rate": 1.999588950063768e-05,
      "loss": 1.2513,
      "step": 106
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2365169525146484,
      "learning_rate": 1.9995524322835035e-05,
      "loss": 1.2297,
      "step": 107
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4092674255371094,
      "learning_rate": 1.9995143610285275e-05,
      "loss": 1.2839,
      "step": 108
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.46215558052063,
      "learning_rate": 1.99947473635801e-05,
      "loss": 1.3638,
      "step": 109
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2486085891723633,
      "learning_rate": 1.9994335583335336e-05,
      "loss": 1.2936,
      "step": 110
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4556329250335693,
      "learning_rate": 1.999390827019096e-05,
      "loss": 1.3097,
      "step": 111
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.502081871032715,
      "learning_rate": 1.999346542481109e-05,
      "loss": 1.2712,
      "step": 112
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0530834197998047,
      "learning_rate": 1.9993007047883988e-05,
      "loss": 1.1967,
      "step": 113
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1497225761413574,
      "learning_rate": 1.9992533140122044e-05,
      "loss": 1.2817,
      "step": 114
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9765923023223877,
      "learning_rate": 1.9992043702261795e-05,
      "loss": 1.2355,
      "step": 115
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2849507331848145,
      "learning_rate": 1.9991538735063906e-05,
      "loss": 1.2567,
      "step": 116
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.325603723526001,
      "learning_rate": 1.999101823931319e-05,
      "loss": 1.2685,
      "step": 117
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3810677528381348,
      "learning_rate": 1.999048221581858e-05,
      "loss": 1.2415,
      "step": 118
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.360539436340332,
      "learning_rate": 1.9989930665413148e-05,
      "loss": 1.2892,
      "step": 119
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4336113929748535,
      "learning_rate": 1.99893635889541e-05,
      "loss": 1.4046,
      "step": 120
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.715312957763672,
      "learning_rate": 1.998878098732277e-05,
      "loss": 1.271,
      "step": 121
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4826815128326416,
      "learning_rate": 1.9988182861424613e-05,
      "loss": 1.2092,
      "step": 122
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3484036922454834,
      "learning_rate": 1.9987569212189224e-05,
      "loss": 1.2229,
      "step": 123
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.512068033218384,
      "learning_rate": 1.9986940040570323e-05,
      "loss": 1.2934,
      "step": 124
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.507819414138794,
      "learning_rate": 1.9986295347545738e-05,
      "loss": 1.2745,
      "step": 125
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.254659652709961,
      "learning_rate": 1.9985635134117445e-05,
      "loss": 1.2502,
      "step": 126
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.196779727935791,
      "learning_rate": 1.998495940131152e-05,
      "loss": 1.2991,
      "step": 127
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1526870727539062,
      "learning_rate": 1.998426815017817e-05,
      "loss": 1.3108,
      "step": 128
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3455443382263184,
      "learning_rate": 1.998356138179171e-05,
      "loss": 1.2744,
      "step": 129
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2342965602874756,
      "learning_rate": 1.9982839097250588e-05,
      "loss": 1.285,
      "step": 130
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.273498058319092,
      "learning_rate": 1.998210129767735e-05,
      "loss": 1.2896,
      "step": 131
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4580445289611816,
      "learning_rate": 1.998134798421867e-05,
      "loss": 1.2745,
      "step": 132
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2461912631988525,
      "learning_rate": 1.9980579158045322e-05,
      "loss": 1.2021,
      "step": 133
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.318605899810791,
      "learning_rate": 1.9979794820352187e-05,
      "loss": 1.3205,
      "step": 134
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.191802978515625,
      "learning_rate": 1.9978994972358265e-05,
      "loss": 1.2273,
      "step": 135
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.36952805519104,
      "learning_rate": 1.9978179615306658e-05,
      "loss": 1.306,
      "step": 136
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.374260187149048,
      "learning_rate": 1.997734875046456e-05,
      "loss": 1.3062,
      "step": 137
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.306464195251465,
      "learning_rate": 1.997650237912329e-05,
      "loss": 1.2961,
      "step": 138
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.217409133911133,
      "learning_rate": 1.9975640502598243e-05,
      "loss": 1.2859,
      "step": 139
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1801583766937256,
      "learning_rate": 1.9974763122228927e-05,
      "loss": 1.2145,
      "step": 140
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.338252305984497,
      "learning_rate": 1.9973870239378938e-05,
      "loss": 1.2138,
      "step": 141
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3280789852142334,
      "learning_rate": 1.997296185543597e-05,
      "loss": 1.2364,
      "step": 142
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2481040954589844,
      "learning_rate": 1.9972037971811802e-05,
      "loss": 1.2289,
      "step": 143
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.36460542678833,
      "learning_rate": 1.9971098589942314e-05,
      "loss": 1.2423,
      "step": 144
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3949170112609863,
      "learning_rate": 1.997014371128746e-05,
      "loss": 1.2954,
      "step": 145
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2784717082977295,
      "learning_rate": 1.9969173337331283e-05,
      "loss": 1.2482,
      "step": 146
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.916337490081787,
      "learning_rate": 1.996818746958191e-05,
      "loss": 1.3185,
      "step": 147
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.183739185333252,
      "learning_rate": 1.996718610957155e-05,
      "loss": 1.2814,
      "step": 148
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.285522222518921,
      "learning_rate": 1.9966169258856488e-05,
      "loss": 1.167,
      "step": 149
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2251217365264893,
      "learning_rate": 1.996513691901708e-05,
      "loss": 1.3493,
      "step": 150
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0339388847351074,
      "learning_rate": 1.996408909165776e-05,
      "loss": 1.2464,
      "step": 151
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.497110605239868,
      "learning_rate": 1.996302577840703e-05,
      "loss": 1.2408,
      "step": 152
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.312018632888794,
      "learning_rate": 1.9961946980917457e-05,
      "loss": 1.1667,
      "step": 153
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.241643190383911,
      "learning_rate": 1.996085270086568e-05,
      "loss": 1.257,
      "step": 154
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2494215965270996,
      "learning_rate": 1.9959742939952393e-05,
      "loss": 1.3016,
      "step": 155
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.354376792907715,
      "learning_rate": 1.9958617699902354e-05,
      "loss": 1.2556,
      "step": 156
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2058186531066895,
      "learning_rate": 1.9957476982464382e-05,
      "loss": 1.2694,
      "step": 157
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1562187671661377,
      "learning_rate": 1.9956320789411338e-05,
      "loss": 1.2799,
      "step": 158
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4779293537139893,
      "learning_rate": 1.995514912254015e-05,
      "loss": 1.2443,
      "step": 159
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3432157039642334,
      "learning_rate": 1.9953961983671792e-05,
      "loss": 1.2247,
      "step": 160
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.3095321655273438,
      "learning_rate": 1.9952759374651266e-05,
      "loss": 1.2866,
      "step": 161
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.384321928024292,
      "learning_rate": 1.9951541297347642e-05,
      "loss": 1.2672,
      "step": 162
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.2591514587402344,
      "learning_rate": 1.9950307753654016e-05,
      "loss": 1.1568,
      "step": 163
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.288320779800415,
      "learning_rate": 1.9949058745487524e-05,
      "loss": 1.1908,
      "step": 164
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.6283698081970215,
      "learning_rate": 1.994779427478934e-05,
      "loss": 1.2433,
      "step": 165
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1756513118743896,
      "learning_rate": 1.994651434352467e-05,
      "loss": 1.2508,
      "step": 166
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.263308048248291,
      "learning_rate": 1.9945218953682736e-05,
      "loss": 1.2545,
      "step": 167
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.318333387374878,
      "learning_rate": 1.99439081072768e-05,
      "loss": 1.2935,
      "step": 168
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9670860767364502,
      "learning_rate": 1.994258180634414e-05,
      "loss": 1.259,
      "step": 169
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0780692100524902,
      "learning_rate": 1.9941240052946055e-05,
      "loss": 1.1961,
      "step": 170
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.311269998550415,
      "learning_rate": 1.9939882849167853e-05,
      "loss": 1.251,
      "step": 171
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.397747039794922,
      "learning_rate": 1.9938510197118868e-05,
      "loss": 1.2958,
      "step": 172
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2074201107025146,
      "learning_rate": 1.9937122098932428e-05,
      "loss": 1.2904,
      "step": 173
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.085521936416626,
      "learning_rate": 1.9935718556765878e-05,
      "loss": 1.2912,
      "step": 174
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0005040168762207,
      "learning_rate": 1.9934299572800556e-05,
      "loss": 1.2388,
      "step": 175
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.340747356414795,
      "learning_rate": 1.9932865149241813e-05,
      "loss": 1.3024,
      "step": 176
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1607589721679688,
      "learning_rate": 1.9931415288318985e-05,
      "loss": 1.2402,
      "step": 177
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0412747859954834,
      "learning_rate": 1.9929949992285397e-05,
      "loss": 1.241,
      "step": 178
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1975045204162598,
      "learning_rate": 1.9928469263418376e-05,
      "loss": 1.2201,
      "step": 179
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1799354553222656,
      "learning_rate": 1.992697310401922e-05,
      "loss": 1.2414,
      "step": 180
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2950644493103027,
      "learning_rate": 1.9925461516413224e-05,
      "loss": 1.2466,
      "step": 181
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2479352951049805,
      "learning_rate": 1.9923934502949645e-05,
      "loss": 1.2193,
      "step": 182
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2703311443328857,
      "learning_rate": 1.9922392066001724e-05,
      "loss": 1.2733,
      "step": 183
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3158016204833984,
      "learning_rate": 1.992083420796667e-05,
      "loss": 1.2406,
      "step": 184
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2654054164886475,
      "learning_rate": 1.9919260931265666e-05,
      "loss": 1.2635,
      "step": 185
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3429250717163086,
      "learning_rate": 1.991767223834384e-05,
      "loss": 1.2728,
      "step": 186
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.323758363723755,
      "learning_rate": 1.9916068131670302e-05,
      "loss": 1.1494,
      "step": 187
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9939885139465332,
      "learning_rate": 1.9914448613738107e-05,
      "loss": 1.2021,
      "step": 188
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2598817348480225,
      "learning_rate": 1.9912813687064255e-05,
      "loss": 1.2651,
      "step": 189
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1655707359313965,
      "learning_rate": 1.9911163354189702e-05,
      "loss": 1.2218,
      "step": 190
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1087121963500977,
      "learning_rate": 1.990949761767935e-05,
      "loss": 1.2379,
      "step": 191
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9608244895935059,
      "learning_rate": 1.9907816480122032e-05,
      "loss": 1.2366,
      "step": 192
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3410162925720215,
      "learning_rate": 1.9906119944130527e-05,
      "loss": 1.1352,
      "step": 193
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.3157193660736084,
      "learning_rate": 1.9904408012341542e-05,
      "loss": 1.2195,
      "step": 194
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2978687286376953,
      "learning_rate": 1.9902680687415704e-05,
      "loss": 1.2246,
      "step": 195
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.179948568344116,
      "learning_rate": 1.9900937972037577e-05,
      "loss": 1.2228,
      "step": 196
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.154505491256714,
      "learning_rate": 1.989917986891563e-05,
      "loss": 1.2372,
      "step": 197
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1358137130737305,
      "learning_rate": 1.9897406380782262e-05,
      "loss": 1.2666,
      "step": 198
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.251882553100586,
      "learning_rate": 1.9895617510393773e-05,
      "loss": 1.1672,
      "step": 199
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.10372257232666,
      "learning_rate": 1.9893813260530368e-05,
      "loss": 1.2662,
      "step": 200
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0817527770996094,
      "learning_rate": 1.9891993633996164e-05,
      "loss": 1.2076,
      "step": 201
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1246836185455322,
      "learning_rate": 1.989015863361917e-05,
      "loss": 1.2138,
      "step": 202
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2330894470214844,
      "learning_rate": 1.9888308262251286e-05,
      "loss": 1.2241,
      "step": 203
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5334792137145996,
      "learning_rate": 1.988644252276831e-05,
      "loss": 1.2449,
      "step": 204
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2931575775146484,
      "learning_rate": 1.988456141806991e-05,
      "loss": 1.2872,
      "step": 205
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6030750274658203,
      "learning_rate": 1.9882664951079652e-05,
      "loss": 1.2259,
      "step": 206
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.965177297592163,
      "learning_rate": 1.9880753124744964e-05,
      "loss": 1.287,
      "step": 207
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1109654903411865,
      "learning_rate": 1.9878825942037147e-05,
      "loss": 1.2879,
      "step": 208
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9684789180755615,
      "learning_rate": 1.9876883405951378e-05,
      "loss": 1.1398,
      "step": 209
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3436923027038574,
      "learning_rate": 1.9874925519506686e-05,
      "loss": 1.2086,
      "step": 210
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2408816814422607,
      "learning_rate": 1.9872952285745958e-05,
      "loss": 1.2647,
      "step": 211
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9891098737716675,
      "learning_rate": 1.987096370773594e-05,
      "loss": 1.1984,
      "step": 212
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.28682541847229,
      "learning_rate": 1.9868959788567213e-05,
      "loss": 1.2638,
      "step": 213
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3960673809051514,
      "learning_rate": 1.9866940531354214e-05,
      "loss": 1.2261,
      "step": 214
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1569385528564453,
      "learning_rate": 1.9864905939235215e-05,
      "loss": 1.1749,
      "step": 215
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0887861251831055,
      "learning_rate": 1.9862856015372315e-05,
      "loss": 1.236,
      "step": 216
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.269155979156494,
      "learning_rate": 1.9860790762951447e-05,
      "loss": 1.1923,
      "step": 217
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.212758779525757,
      "learning_rate": 1.985871018518236e-05,
      "loss": 1.1972,
      "step": 218
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.152022123336792,
      "learning_rate": 1.985661428529863e-05,
      "loss": 1.2075,
      "step": 219
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.161104440689087,
      "learning_rate": 1.9854503066557646e-05,
      "loss": 1.2152,
      "step": 220
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.125103235244751,
      "learning_rate": 1.9852376532240594e-05,
      "loss": 1.1759,
      "step": 221
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0022904872894287,
      "learning_rate": 1.9850234685652478e-05,
      "loss": 1.1913,
      "step": 222
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4505491256713867,
      "learning_rate": 1.9848077530122083e-05,
      "loss": 1.2517,
      "step": 223
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.287059783935547,
      "learning_rate": 1.9845905069002004e-05,
      "loss": 1.1899,
      "step": 224
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5685973167419434,
      "learning_rate": 1.984371730566861e-05,
      "loss": 1.2253,
      "step": 225
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2151622772216797,
      "learning_rate": 1.9841514243522065e-05,
      "loss": 1.1895,
      "step": 226
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0352623462677,
      "learning_rate": 1.98392958859863e-05,
      "loss": 1.2181,
      "step": 227
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.413649559020996,
      "learning_rate": 1.9837062236509013e-05,
      "loss": 1.2631,
      "step": 228
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1189515590667725,
      "learning_rate": 1.9834813298561686e-05,
      "loss": 1.1933,
      "step": 229
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1793789863586426,
      "learning_rate": 1.983254907563955e-05,
      "loss": 1.2518,
      "step": 230
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.728706121444702,
      "learning_rate": 1.9830269571261585e-05,
      "loss": 1.2629,
      "step": 231
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0654258728027344,
      "learning_rate": 1.9827974788970536e-05,
      "loss": 1.2287,
      "step": 232
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.315378189086914,
      "learning_rate": 1.9825664732332886e-05,
      "loss": 1.1878,
      "step": 233
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.201420307159424,
      "learning_rate": 1.982333940493885e-05,
      "loss": 1.2017,
      "step": 234
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3849496841430664,
      "learning_rate": 1.982099881040239e-05,
      "loss": 1.3055,
      "step": 235
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.224586248397827,
      "learning_rate": 1.9818642952361188e-05,
      "loss": 1.2343,
      "step": 236
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.275792360305786,
      "learning_rate": 1.9816271834476642e-05,
      "loss": 1.2985,
      "step": 237
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3047266006469727,
      "learning_rate": 1.981388546043388e-05,
      "loss": 1.2798,
      "step": 238
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.278667688369751,
      "learning_rate": 1.9811483833941726e-05,
      "loss": 1.2083,
      "step": 239
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5371344089508057,
      "learning_rate": 1.9809066958732727e-05,
      "loss": 1.2113,
      "step": 240
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2228798866271973,
      "learning_rate": 1.9806634838563113e-05,
      "loss": 1.2165,
      "step": 241
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.30216908454895,
      "learning_rate": 1.980418747721281e-05,
      "loss": 1.1544,
      "step": 242
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.480717420578003,
      "learning_rate": 1.9801724878485438e-05,
      "loss": 1.243,
      "step": 243
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.152071952819824,
      "learning_rate": 1.9799247046208297e-05,
      "loss": 1.2545,
      "step": 244
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.374875783920288,
      "learning_rate": 1.9796753984232357e-05,
      "loss": 1.2347,
      "step": 245
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.216991424560547,
      "learning_rate": 1.9794245696432262e-05,
      "loss": 1.2553,
      "step": 246
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2298648357391357,
      "learning_rate": 1.9791722186706318e-05,
      "loss": 1.139,
      "step": 247
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.096233606338501,
      "learning_rate": 1.9789183458976485e-05,
      "loss": 1.2227,
      "step": 248
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.278942346572876,
      "learning_rate": 1.9786629517188384e-05,
      "loss": 1.283,
      "step": 249
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.279949426651001,
      "learning_rate": 1.9784060365311276e-05,
      "loss": 1.1881,
      "step": 250
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3361737728118896,
      "learning_rate": 1.9781476007338058e-05,
      "loss": 1.2122,
      "step": 251
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3072116374969482,
      "learning_rate": 1.9778876447285264e-05,
      "loss": 1.2259,
      "step": 252
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3757102489471436,
      "learning_rate": 1.977626168919305e-05,
      "loss": 1.317,
      "step": 253
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.119673252105713,
      "learning_rate": 1.9773631737125192e-05,
      "loss": 1.23,
      "step": 254
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3259479999542236,
      "learning_rate": 1.9770986595169096e-05,
      "loss": 1.1484,
      "step": 255
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.043926239013672,
      "learning_rate": 1.976832626743575e-05,
      "loss": 1.1826,
      "step": 256
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3053832054138184,
      "learning_rate": 1.9765650758059766e-05,
      "loss": 1.2374,
      "step": 257
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.45096492767334,
      "learning_rate": 1.9762960071199334e-05,
      "loss": 1.171,
      "step": 258
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2242844104766846,
      "learning_rate": 1.9760254211036245e-05,
      "loss": 1.2457,
      "step": 259
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.744565725326538,
      "learning_rate": 1.9757533181775865e-05,
      "loss": 1.2168,
      "step": 260
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.15044903755188,
      "learning_rate": 1.9754796987647134e-05,
      "loss": 1.2934,
      "step": 261
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1242144107818604,
      "learning_rate": 1.9752045632902567e-05,
      "loss": 1.2223,
      "step": 262
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2642345428466797,
      "learning_rate": 1.9749279121818235e-05,
      "loss": 1.1263,
      "step": 263
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1664299964904785,
      "learning_rate": 1.9746497458693776e-05,
      "loss": 1.2241,
      "step": 264
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.16222882270813,
      "learning_rate": 1.9743700647852356e-05,
      "loss": 1.2461,
      "step": 265
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.329148769378662,
      "learning_rate": 1.9740888693640698e-05,
      "loss": 1.2002,
      "step": 266
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1356124877929688,
      "learning_rate": 1.9738061600429062e-05,
      "loss": 1.2486,
      "step": 267
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0238022804260254,
      "learning_rate": 1.9735219372611232e-05,
      "loss": 1.2479,
      "step": 268
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2120707035064697,
      "learning_rate": 1.9732362014604515e-05,
      "loss": 1.2628,
      "step": 269
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.148346185684204,
      "learning_rate": 1.972948953084973e-05,
      "loss": 1.2392,
      "step": 270
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3298606872558594,
      "learning_rate": 1.9726601925811204e-05,
      "loss": 1.2278,
      "step": 271
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2186288833618164,
      "learning_rate": 1.9723699203976768e-05,
      "loss": 1.2307,
      "step": 272
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.947746992111206,
      "learning_rate": 1.9720781369857747e-05,
      "loss": 1.2305,
      "step": 273
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2245407104492188,
      "learning_rate": 1.9717848427988954e-05,
      "loss": 1.2319,
      "step": 274
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.018190860748291,
      "learning_rate": 1.9714900382928674e-05,
      "loss": 1.3151,
      "step": 275
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.210710048675537,
      "learning_rate": 1.971193723925868e-05,
      "loss": 1.23,
      "step": 276
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.122095823287964,
      "learning_rate": 1.9708959001584197e-05,
      "loss": 1.2069,
      "step": 277
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2687864303588867,
      "learning_rate": 1.970596567453391e-05,
      "loss": 1.308,
      "step": 278
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.095733404159546,
      "learning_rate": 1.9702957262759964e-05,
      "loss": 1.2258,
      "step": 279
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1374547481536865,
      "learning_rate": 1.9699933770937948e-05,
      "loss": 1.1522,
      "step": 280
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.171401023864746,
      "learning_rate": 1.969689520376687e-05,
      "loss": 1.2302,
      "step": 281
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.229557752609253,
      "learning_rate": 1.969384156596919e-05,
      "loss": 1.2671,
      "step": 282
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0715456008911133,
      "learning_rate": 1.969077286229078e-05,
      "loss": 1.2566,
      "step": 283
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.9227606058120728,
      "learning_rate": 1.9687689097500927e-05,
      "loss": 1.2056,
      "step": 284
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.136815309524536,
      "learning_rate": 1.968459027639233e-05,
      "loss": 1.1646,
      "step": 285
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4012396335601807,
      "learning_rate": 1.968147640378108e-05,
      "loss": 1.1822,
      "step": 286
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.187089681625366,
      "learning_rate": 1.9678347484506667e-05,
      "loss": 1.2382,
      "step": 287
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1604082584381104,
      "learning_rate": 1.9675203523431964e-05,
      "loss": 1.2207,
      "step": 288
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2646968364715576,
      "learning_rate": 1.9672044525443222e-05,
      "loss": 1.233,
      "step": 289
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3587231636047363,
      "learning_rate": 1.9668870495450064e-05,
      "loss": 1.2298,
      "step": 290
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.193830966949463,
      "learning_rate": 1.9665681438385475e-05,
      "loss": 1.1364,
      "step": 291
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2050600051879883,
      "learning_rate": 1.9662477359205786e-05,
      "loss": 1.3197,
      "step": 292
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4618115425109863,
      "learning_rate": 1.9659258262890683e-05,
      "loss": 1.2359,
      "step": 293
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.539912223815918,
      "learning_rate": 1.9656024154443195e-05,
      "loss": 1.2604,
      "step": 294
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.277086019515991,
      "learning_rate": 1.9652775038889676e-05,
      "loss": 1.2512,
      "step": 295
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1279098987579346,
      "learning_rate": 1.9649510921279803e-05,
      "loss": 1.1781,
      "step": 296
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.500462293624878,
      "learning_rate": 1.9646231806686566e-05,
      "loss": 1.1193,
      "step": 297
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.228179454803467,
      "learning_rate": 1.964293770020628e-05,
      "loss": 1.2479,
      "step": 298
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2313849925994873,
      "learning_rate": 1.9639628606958535e-05,
      "loss": 1.2444,
      "step": 299
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.112659454345703,
      "learning_rate": 1.963630453208623e-05,
      "loss": 1.266,
      "step": 300
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0741355419158936,
      "learning_rate": 1.963296548075555e-05,
      "loss": 1.2137,
      "step": 301
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.360323190689087,
      "learning_rate": 1.9629611458155936e-05,
      "loss": 1.1804,
      "step": 302
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.218899965286255,
      "learning_rate": 1.962624246950012e-05,
      "loss": 1.2136,
      "step": 303
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.358285427093506,
      "learning_rate": 1.9622858520024087e-05,
      "loss": 1.2715,
      "step": 304
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1432554721832275,
      "learning_rate": 1.961945961498706e-05,
      "loss": 1.2454,
      "step": 305
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.229492425918579,
      "learning_rate": 1.9616045759671524e-05,
      "loss": 1.2812,
      "step": 306
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.398219347000122,
      "learning_rate": 1.961261695938319e-05,
      "loss": 1.197,
      "step": 307
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1905367374420166,
      "learning_rate": 1.9609173219450998e-05,
      "loss": 1.2751,
      "step": 308
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.19826340675354,
      "learning_rate": 1.9605714545227105e-05,
      "loss": 1.2393,
      "step": 309
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2849152088165283,
      "learning_rate": 1.960224094208688e-05,
      "loss": 1.1738,
      "step": 310
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0722708702087402,
      "learning_rate": 1.9598752415428893e-05,
      "loss": 1.227,
      "step": 311
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1467223167419434,
      "learning_rate": 1.959524897067491e-05,
      "loss": 1.2558,
      "step": 312
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.069854259490967,
      "learning_rate": 1.9591730613269878e-05,
      "loss": 1.2266,
      "step": 313
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2407312393188477,
      "learning_rate": 1.958819734868193e-05,
      "loss": 1.2149,
      "step": 314
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3539700508117676,
      "learning_rate": 1.9584649182402358e-05,
      "loss": 1.2225,
      "step": 315
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.197462797164917,
      "learning_rate": 1.9581086119945615e-05,
      "loss": 1.2455,
      "step": 316
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6031572818756104,
      "learning_rate": 1.9577508166849308e-05,
      "loss": 1.2335,
      "step": 317
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0838623046875,
      "learning_rate": 1.957391532867418e-05,
      "loss": 1.2339,
      "step": 318
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.196815013885498,
      "learning_rate": 1.9570307611004124e-05,
      "loss": 1.198,
      "step": 319
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1002931594848633,
      "learning_rate": 1.956668501944614e-05,
      "loss": 1.1803,
      "step": 320
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0884618759155273,
      "learning_rate": 1.9563047559630356e-05,
      "loss": 1.2458,
      "step": 321
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.144683837890625,
      "learning_rate": 1.9559395237210002e-05,
      "loss": 1.1453,
      "step": 322
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.249293088912964,
      "learning_rate": 1.955572805786141e-05,
      "loss": 1.2053,
      "step": 323
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3029136657714844,
      "learning_rate": 1.9552046027284e-05,
      "loss": 1.2095,
      "step": 324
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.175297260284424,
      "learning_rate": 1.9548349151200275e-05,
      "loss": 1.193,
      "step": 325
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.131852149963379,
      "learning_rate": 1.954463743535581e-05,
      "loss": 1.1515,
      "step": 326
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.117029905319214,
      "learning_rate": 1.9540910885519243e-05,
      "loss": 1.2115,
      "step": 327
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3540101051330566,
      "learning_rate": 1.953716950748227e-05,
      "loss": 1.1712,
      "step": 328
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.036766529083252,
      "learning_rate": 1.953341330705963e-05,
      "loss": 1.1977,
      "step": 329
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.328587055206299,
      "learning_rate": 1.952964229008909e-05,
      "loss": 1.2532,
      "step": 330
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2466907501220703,
      "learning_rate": 1.9525856462431463e-05,
      "loss": 1.2568,
      "step": 331
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.298018217086792,
      "learning_rate": 1.9522055829970564e-05,
      "loss": 1.2405,
      "step": 332
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1745989322662354,
      "learning_rate": 1.9518240398613226e-05,
      "loss": 1.1996,
      "step": 333
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.332949638366699,
      "learning_rate": 1.951441017428928e-05,
      "loss": 1.2357,
      "step": 334
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1081840991973877,
      "learning_rate": 1.9510565162951538e-05,
      "loss": 1.2056,
      "step": 335
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1658542156219482,
      "learning_rate": 1.950670537057581e-05,
      "loss": 1.2234,
      "step": 336
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.454874038696289,
      "learning_rate": 1.9502830803160873e-05,
      "loss": 1.2228,
      "step": 337
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1150763034820557,
      "learning_rate": 1.9498941466728462e-05,
      "loss": 1.1897,
      "step": 338
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.302992343902588,
      "learning_rate": 1.9495037367323264e-05,
      "loss": 1.2133,
      "step": 339
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2587568759918213,
      "learning_rate": 1.949111851101292e-05,
      "loss": 1.2245,
      "step": 340
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.204352617263794,
      "learning_rate": 1.9487184903888e-05,
      "loss": 1.2415,
      "step": 341
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.967620611190796,
      "learning_rate": 1.9483236552061996e-05,
      "loss": 1.2289,
      "step": 342
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.057037830352783,
      "learning_rate": 1.947927346167132e-05,
      "loss": 1.1596,
      "step": 343
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.324186325073242,
      "learning_rate": 1.947529563887529e-05,
      "loss": 1.2369,
      "step": 344
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1921534538269043,
      "learning_rate": 1.947130308985612e-05,
      "loss": 1.1937,
      "step": 345
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1376328468322754,
      "learning_rate": 1.9467295820818913e-05,
      "loss": 1.1598,
      "step": 346
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.143648862838745,
      "learning_rate": 1.9463273837991643e-05,
      "loss": 1.2168,
      "step": 347
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3809800148010254,
      "learning_rate": 1.945923714762516e-05,
      "loss": 1.1196,
      "step": 348
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.062492609024048,
      "learning_rate": 1.945518575599317e-05,
      "loss": 1.2448,
      "step": 349
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.110219717025757,
      "learning_rate": 1.945111966939222e-05,
      "loss": 1.2097,
      "step": 350
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.167130470275879,
      "learning_rate": 1.9447038894141707e-05,
      "loss": 1.2661,
      "step": 351
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.168332815170288,
      "learning_rate": 1.9442943436583845e-05,
      "loss": 1.2077,
      "step": 352
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.290954828262329,
      "learning_rate": 1.9438833303083677e-05,
      "loss": 1.1761,
      "step": 353
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.265655279159546,
      "learning_rate": 1.943470850002905e-05,
      "loss": 1.1869,
      "step": 354
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0975725650787354,
      "learning_rate": 1.9430569033830606e-05,
      "loss": 1.1805,
      "step": 355
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.124629497528076,
      "learning_rate": 1.9426414910921785e-05,
      "loss": 1.1835,
      "step": 356
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0969207286834717,
      "learning_rate": 1.94222461377588e-05,
      "loss": 1.198,
      "step": 357
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1524336338043213,
      "learning_rate": 1.9418062720820636e-05,
      "loss": 1.1264,
      "step": 358
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.178459882736206,
      "learning_rate": 1.9413864666609036e-05,
      "loss": 1.2028,
      "step": 359
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2509171962738037,
      "learning_rate": 1.9409651981648486e-05,
      "loss": 1.1527,
      "step": 360
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3319737911224365,
      "learning_rate": 1.9405424672486218e-05,
      "loss": 1.1469,
      "step": 361
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.281989097595215,
      "learning_rate": 1.940118274569219e-05,
      "loss": 1.2543,
      "step": 362
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1432223320007324,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 1.1887,
      "step": 363
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1963324546813965,
      "learning_rate": 1.939265506560228e-05,
      "loss": 1.2312,
      "step": 364
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0228586196899414,
      "learning_rate": 1.9388369325559862e-05,
      "loss": 1.1931,
      "step": 365
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.151243209838867,
      "learning_rate": 1.93840689943926e-05,
      "loss": 1.2197,
      "step": 366
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.167109727859497,
      "learning_rate": 1.937975407878394e-05,
      "loss": 1.1948,
      "step": 367
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4130468368530273,
      "learning_rate": 1.9375424585439994e-05,
      "loss": 1.1963,
      "step": 368
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.342984676361084,
      "learning_rate": 1.9371080521089536e-05,
      "loss": 1.148,
      "step": 369
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0029289722442627,
      "learning_rate": 1.9366721892483976e-05,
      "loss": 1.1844,
      "step": 370
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.020735740661621,
      "learning_rate": 1.9362348706397374e-05,
      "loss": 1.2291,
      "step": 371
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4407870769500732,
      "learning_rate": 1.9357960969626398e-05,
      "loss": 1.2176,
      "step": 372
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.211923122406006,
      "learning_rate": 1.935355868899034e-05,
      "loss": 1.1994,
      "step": 373
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.002372980117798,
      "learning_rate": 1.93491418713311e-05,
      "loss": 1.2149,
      "step": 374
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7742695808410645,
      "learning_rate": 1.9344710523513157e-05,
      "loss": 1.1891,
      "step": 375
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8936272859573364,
      "learning_rate": 1.9340264652423584e-05,
      "loss": 1.2333,
      "step": 376
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.483459234237671,
      "learning_rate": 1.9335804264972018e-05,
      "loss": 1.118,
      "step": 377
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1463980674743652,
      "learning_rate": 1.9331329368090664e-05,
      "loss": 1.2092,
      "step": 378
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.129972457885742,
      "learning_rate": 1.9326839968734278e-05,
      "loss": 1.2001,
      "step": 379
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0306105613708496,
      "learning_rate": 1.9322336073880143e-05,
      "loss": 1.2054,
      "step": 380
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.209451198577881,
      "learning_rate": 1.9317817690528086e-05,
      "loss": 1.1804,
      "step": 381
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.313292980194092,
      "learning_rate": 1.931328482570044e-05,
      "loss": 1.1688,
      "step": 382
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9034878015518188,
      "learning_rate": 1.9308737486442045e-05,
      "loss": 1.1648,
      "step": 383
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9300029277801514,
      "learning_rate": 1.9304175679820247e-05,
      "loss": 1.1176,
      "step": 384
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.001166582107544,
      "learning_rate": 1.929959941292487e-05,
      "loss": 1.2538,
      "step": 385
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2398672103881836,
      "learning_rate": 1.929500869286821e-05,
      "loss": 1.203,
      "step": 386
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9895315170288086,
      "learning_rate": 1.9290403526785025e-05,
      "loss": 1.182,
      "step": 387
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0807528495788574,
      "learning_rate": 1.9285783921832537e-05,
      "loss": 1.1824,
      "step": 388
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.141988754272461,
      "learning_rate": 1.928114988519039e-05,
      "loss": 1.2153,
      "step": 389
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1146364212036133,
      "learning_rate": 1.927650142406067e-05,
      "loss": 1.251,
      "step": 390
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2481977939605713,
      "learning_rate": 1.9271838545667876e-05,
      "loss": 1.1892,
      "step": 391
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2106428146362305,
      "learning_rate": 1.9267161257258916e-05,
      "loss": 1.189,
      "step": 392
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.216952085494995,
      "learning_rate": 1.926246956610309e-05,
      "loss": 1.3035,
      "step": 393
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.748043417930603,
      "learning_rate": 1.9257763479492086e-05,
      "loss": 1.234,
      "step": 394
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1788971424102783,
      "learning_rate": 1.9253043004739967e-05,
      "loss": 1.1617,
      "step": 395
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.082805871963501,
      "learning_rate": 1.9248308149183155e-05,
      "loss": 1.2181,
      "step": 396
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.009429693222046,
      "learning_rate": 1.9243558920180417e-05,
      "loss": 1.1619,
      "step": 397
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9463576078414917,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 1.1971,
      "step": 398
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.130729913711548,
      "learning_rate": 1.9234017371383946e-05,
      "loss": 1.1438,
      "step": 399
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.225952386856079,
      "learning_rate": 1.92292250664194e-05,
      "loss": 1.2351,
      "step": 400
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2027974128723145,
      "learning_rate": 1.9224418417667295e-05,
      "loss": 1.2408,
      "step": 401
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4277420043945312,
      "learning_rate": 1.9219597432597975e-05,
      "loss": 1.0969,
      "step": 402
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.020223617553711,
      "learning_rate": 1.921476211870408e-05,
      "loss": 1.1856,
      "step": 403
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8073242902755737,
      "learning_rate": 1.92099124835005e-05,
      "loss": 1.1656,
      "step": 404
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.35741925239563,
      "learning_rate": 1.9205048534524405e-05,
      "loss": 1.1596,
      "step": 405
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.139561414718628,
      "learning_rate": 1.9200170279335194e-05,
      "loss": 1.1599,
      "step": 406
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1246283054351807,
      "learning_rate": 1.919527772551451e-05,
      "loss": 1.213,
      "step": 407
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1687419414520264,
      "learning_rate": 1.9190370880666206e-05,
      "loss": 1.17,
      "step": 408
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0663907527923584,
      "learning_rate": 1.9185449752416367e-05,
      "loss": 1.1951,
      "step": 409
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.338210105895996,
      "learning_rate": 1.918051434841326e-05,
      "loss": 1.1592,
      "step": 410
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1316592693328857,
      "learning_rate": 1.917556467632734e-05,
      "loss": 1.1646,
      "step": 411
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.4319498538970947,
      "learning_rate": 1.917060074385124e-05,
      "loss": 1.1876,
      "step": 412
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9116313457489014,
      "learning_rate": 1.9165622558699763e-05,
      "loss": 1.116,
      "step": 413
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.097460985183716,
      "learning_rate": 1.916063012860985e-05,
      "loss": 1.2626,
      "step": 414
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.314021587371826,
      "learning_rate": 1.9155623461340594e-05,
      "loss": 1.1703,
      "step": 415
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0281221866607666,
      "learning_rate": 1.91506025646732e-05,
      "loss": 1.1763,
      "step": 416
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0726029872894287,
      "learning_rate": 1.9145567446411003e-05,
      "loss": 1.1615,
      "step": 417
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.372267007827759,
      "learning_rate": 1.9140518114379433e-05,
      "loss": 1.1769,
      "step": 418
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.332857131958008,
      "learning_rate": 1.913545457642601e-05,
      "loss": 1.1253,
      "step": 419
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.057389736175537,
      "learning_rate": 1.9130376840420337e-05,
      "loss": 1.1593,
      "step": 420
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2536749839782715,
      "learning_rate": 1.912528491425408e-05,
      "loss": 1.1695,
      "step": 421
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.974597454071045,
      "learning_rate": 1.9120178805840956e-05,
      "loss": 1.1664,
      "step": 422
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1424496173858643,
      "learning_rate": 1.9115058523116734e-05,
      "loss": 1.1847,
      "step": 423
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2070212364196777,
      "learning_rate": 1.9109924074039198e-05,
      "loss": 1.1625,
      "step": 424
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3570125102996826,
      "learning_rate": 1.9104775466588162e-05,
      "loss": 1.2026,
      "step": 425
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.307504177093506,
      "learning_rate": 1.9099612708765432e-05,
      "loss": 1.1913,
      "step": 426
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2014806270599365,
      "learning_rate": 1.9094435808594823e-05,
      "loss": 1.1244,
      "step": 427
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1234195232391357,
      "learning_rate": 1.908924477412211e-05,
      "loss": 1.1198,
      "step": 428
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.15922474861145,
      "learning_rate": 1.9084039613415052e-05,
      "loss": 1.1628,
      "step": 429
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.173921823501587,
      "learning_rate": 1.9078820334563346e-05,
      "loss": 1.225,
      "step": 430
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0838921070098877,
      "learning_rate": 1.907358694567865e-05,
      "loss": 1.2123,
      "step": 431
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2293474674224854,
      "learning_rate": 1.9068339454894536e-05,
      "loss": 1.2544,
      "step": 432
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1708827018737793,
      "learning_rate": 1.9063077870366504e-05,
      "loss": 1.1958,
      "step": 433
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2230634689331055,
      "learning_rate": 1.9057802200271943e-05,
      "loss": 1.197,
      "step": 434
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.5192413330078125,
      "learning_rate": 1.905251245281015e-05,
      "loss": 1.1966,
      "step": 435
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0417089462280273,
      "learning_rate": 1.904720863620229e-05,
      "loss": 1.2558,
      "step": 436
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2727530002593994,
      "learning_rate": 1.9041890758691403e-05,
      "loss": 1.2317,
      "step": 437
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0783374309539795,
      "learning_rate": 1.903655882854237e-05,
      "loss": 1.1583,
      "step": 438
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.233299493789673,
      "learning_rate": 1.903121285404192e-05,
      "loss": 1.1272,
      "step": 439
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3763625621795654,
      "learning_rate": 1.902585284349861e-05,
      "loss": 1.1895,
      "step": 440
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2493138313293457,
      "learning_rate": 1.9020478805242803e-05,
      "loss": 1.1472,
      "step": 441
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.203070878982544,
      "learning_rate": 1.901509074762668e-05,
      "loss": 1.1619,
      "step": 442
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.8983192443847656,
      "learning_rate": 1.900968867902419e-05,
      "loss": 1.1177,
      "step": 443
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0650458335876465,
      "learning_rate": 1.9004272607831077e-05,
      "loss": 1.2061,
      "step": 444
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1242988109588623,
      "learning_rate": 1.8998842542464833e-05,
      "loss": 1.2129,
      "step": 445
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.975136399269104,
      "learning_rate": 1.8993398491364702e-05,
      "loss": 1.1429,
      "step": 446
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.244490623474121,
      "learning_rate": 1.8987940462991673e-05,
      "loss": 1.1717,
      "step": 447
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.107891798019409,
      "learning_rate": 1.898246846582844e-05,
      "loss": 1.1628,
      "step": 448
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.146310329437256,
      "learning_rate": 1.8976982508379436e-05,
      "loss": 1.248,
      "step": 449
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.118587017059326,
      "learning_rate": 1.897148259917076e-05,
      "loss": 1.1543,
      "step": 450
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.118992805480957,
      "learning_rate": 1.896596874675021e-05,
      "loss": 1.1552,
      "step": 451
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1333742141723633,
      "learning_rate": 1.8960440959687254e-05,
      "loss": 1.2132,
      "step": 452
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4819490909576416,
      "learning_rate": 1.895489924657301e-05,
      "loss": 1.1703,
      "step": 453
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2328195571899414,
      "learning_rate": 1.894934361602025e-05,
      "loss": 1.1419,
      "step": 454
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.12747859954834,
      "learning_rate": 1.8943774076663372e-05,
      "loss": 1.2211,
      "step": 455
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1832423210144043,
      "learning_rate": 1.8938190637158377e-05,
      "loss": 1.1935,
      "step": 456
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.307791233062744,
      "learning_rate": 1.893259330618289e-05,
      "loss": 1.2078,
      "step": 457
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.339052438735962,
      "learning_rate": 1.8926982092436117e-05,
      "loss": 1.2189,
      "step": 458
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2158572673797607,
      "learning_rate": 1.8921357004638837e-05,
      "loss": 1.1776,
      "step": 459
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8780187368392944,
      "learning_rate": 1.891571805153339e-05,
      "loss": 1.1881,
      "step": 460
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.283869981765747,
      "learning_rate": 1.891006524188368e-05,
      "loss": 1.161,
      "step": 461
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4903321266174316,
      "learning_rate": 1.890439858447513e-05,
      "loss": 1.1895,
      "step": 462
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1354801654815674,
      "learning_rate": 1.8898718088114688e-05,
      "loss": 1.1896,
      "step": 463
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.088341474533081,
      "learning_rate": 1.8893023761630816e-05,
      "loss": 1.1158,
      "step": 464
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2244045734405518,
      "learning_rate": 1.888731561387347e-05,
      "loss": 1.1492,
      "step": 465
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.978247880935669,
      "learning_rate": 1.8881593653714075e-05,
      "loss": 1.1926,
      "step": 466
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1522152423858643,
      "learning_rate": 1.8875857890045544e-05,
      "loss": 1.1375,
      "step": 467
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3932409286499023,
      "learning_rate": 1.887010833178222e-05,
      "loss": 1.1875,
      "step": 468
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.348376512527466,
      "learning_rate": 1.8864344987859898e-05,
      "loss": 1.2022,
      "step": 469
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2742807865142822,
      "learning_rate": 1.88585678672358e-05,
      "loss": 1.1972,
      "step": 470
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.216028928756714,
      "learning_rate": 1.885277697888855e-05,
      "loss": 1.2013,
      "step": 471
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.155895471572876,
      "learning_rate": 1.8846972331818182e-05,
      "loss": 1.1414,
      "step": 472
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.039586305618286,
      "learning_rate": 1.8841153935046098e-05,
      "loss": 1.2522,
      "step": 473
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2482640743255615,
      "learning_rate": 1.8835321797615083e-05,
      "loss": 1.1231,
      "step": 474
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.185901641845703,
      "learning_rate": 1.8829475928589272e-05,
      "loss": 1.2209,
      "step": 475
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.172161340713501,
      "learning_rate": 1.8823616337054136e-05,
      "loss": 1.2124,
      "step": 476
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0638039112091064,
      "learning_rate": 1.8817743032116483e-05,
      "loss": 1.1969,
      "step": 477
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.145916223526001,
      "learning_rate": 1.8811856022904423e-05,
      "loss": 1.1816,
      "step": 478
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9360535144805908,
      "learning_rate": 1.880595531856738e-05,
      "loss": 1.201,
      "step": 479
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2227909564971924,
      "learning_rate": 1.880004092827605e-05,
      "loss": 1.1866,
      "step": 480
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.019174575805664,
      "learning_rate": 1.87941128612224e-05,
      "loss": 1.171,
      "step": 481
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0187439918518066,
      "learning_rate": 1.8788171126619653e-05,
      "loss": 1.0769,
      "step": 482
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0424869060516357,
      "learning_rate": 1.8782215733702286e-05,
      "loss": 1.1048,
      "step": 483
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.129167079925537,
      "learning_rate": 1.8776246691725988e-05,
      "loss": 1.1616,
      "step": 484
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.180488348007202,
      "learning_rate": 1.8770264009967667e-05,
      "loss": 1.1631,
      "step": 485
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.039923906326294,
      "learning_rate": 1.8764267697725432e-05,
      "loss": 1.1683,
      "step": 486
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.135549545288086,
      "learning_rate": 1.8758257764318566e-05,
      "loss": 1.1861,
      "step": 487
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.897406816482544,
      "learning_rate": 1.8752234219087538e-05,
      "loss": 1.2141,
      "step": 488
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1337647438049316,
      "learning_rate": 1.874619707139396e-05,
      "loss": 1.1466,
      "step": 489
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1091883182525635,
      "learning_rate": 1.8740146330620584e-05,
      "loss": 1.1737,
      "step": 490
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.191256523132324,
      "learning_rate": 1.87340820061713e-05,
      "loss": 1.1986,
      "step": 491
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9010835886001587,
      "learning_rate": 1.872800410747109e-05,
      "loss": 1.1426,
      "step": 492
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.130600690841675,
      "learning_rate": 1.8721912643966055e-05,
      "loss": 1.1395,
      "step": 493
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.271958112716675,
      "learning_rate": 1.871580762512336e-05,
      "loss": 1.2176,
      "step": 494
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2112390995025635,
      "learning_rate": 1.8709689060431242e-05,
      "loss": 1.1028,
      "step": 495
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.13985013961792,
      "learning_rate": 1.8703556959398998e-05,
      "loss": 1.1216,
      "step": 496
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2039289474487305,
      "learning_rate": 1.8697411331556958e-05,
      "loss": 1.2213,
      "step": 497
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0783586502075195,
      "learning_rate": 1.8691252186456465e-05,
      "loss": 1.1831,
      "step": 498
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.063455820083618,
      "learning_rate": 1.868507953366989e-05,
      "loss": 1.2251,
      "step": 499
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2001898288726807,
      "learning_rate": 1.8678893382790583e-05,
      "loss": 1.1164,
      "step": 500
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2533223628997803,
      "learning_rate": 1.8672693743432875e-05,
      "loss": 1.25,
      "step": 501
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.149364471435547,
      "learning_rate": 1.866648062523206e-05,
      "loss": 1.1276,
      "step": 502
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.198911428451538,
      "learning_rate": 1.866025403784439e-05,
      "loss": 1.1598,
      "step": 503
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1691691875457764,
      "learning_rate": 1.865401399094703e-05,
      "loss": 1.1085,
      "step": 504
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2854654788970947,
      "learning_rate": 1.8647760494238082e-05,
      "loss": 1.2454,
      "step": 505
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.079948663711548,
      "learning_rate": 1.864149355743655e-05,
      "loss": 1.2304,
      "step": 506
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2750673294067383,
      "learning_rate": 1.8635213190282312e-05,
      "loss": 1.1122,
      "step": 507
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2954394817352295,
      "learning_rate": 1.862891940253613e-05,
      "loss": 1.2072,
      "step": 508
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.005375623703003,
      "learning_rate": 1.8622612203979628e-05,
      "loss": 1.0773,
      "step": 509
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.958028793334961,
      "learning_rate": 1.861629160441526e-05,
      "loss": 1.1563,
      "step": 510
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9793323278427124,
      "learning_rate": 1.8609957613666316e-05,
      "loss": 1.2058,
      "step": 511
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0789783000946045,
      "learning_rate": 1.86036102415769e-05,
      "loss": 1.1741,
      "step": 512
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.003485679626465,
      "learning_rate": 1.8597249498011906e-05,
      "loss": 1.1529,
      "step": 513
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.223801374435425,
      "learning_rate": 1.8590875392857008e-05,
      "loss": 1.1701,
      "step": 514
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1722559928894043,
      "learning_rate": 1.8584487936018663e-05,
      "loss": 1.1412,
      "step": 515
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2359793186187744,
      "learning_rate": 1.8578087137424057e-05,
      "loss": 1.2157,
      "step": 516
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2716424465179443,
      "learning_rate": 1.8571673007021124e-05,
      "loss": 1.2146,
      "step": 517
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3044779300689697,
      "learning_rate": 1.8565245554778516e-05,
      "loss": 1.1053,
      "step": 518
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1560781002044678,
      "learning_rate": 1.855880479068559e-05,
      "loss": 1.1961,
      "step": 519
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1356141567230225,
      "learning_rate": 1.8552350724752383e-05,
      "loss": 1.1697,
      "step": 520
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1666057109832764,
      "learning_rate": 1.854588336700962e-05,
      "loss": 1.1513,
      "step": 521
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.29411244392395,
      "learning_rate": 1.853940272750867e-05,
      "loss": 1.1244,
      "step": 522
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9961618185043335,
      "learning_rate": 1.8532908816321557e-05,
      "loss": 1.1448,
      "step": 523
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.220252513885498,
      "learning_rate": 1.8526401643540924e-05,
      "loss": 1.2123,
      "step": 524
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1020164489746094,
      "learning_rate": 1.851988121928002e-05,
      "loss": 1.1975,
      "step": 525
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3075919151306152,
      "learning_rate": 1.85133475536727e-05,
      "loss": 1.1974,
      "step": 526
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.099823236465454,
      "learning_rate": 1.8506800656873397e-05,
      "loss": 1.1915,
      "step": 527
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.242279052734375,
      "learning_rate": 1.8500240539057093e-05,
      "loss": 1.1641,
      "step": 528
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.275453805923462,
      "learning_rate": 1.8493667210419337e-05,
      "loss": 1.1594,
      "step": 529
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3849680423736572,
      "learning_rate": 1.8487080681176195e-05,
      "loss": 1.1557,
      "step": 530
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2680253982543945,
      "learning_rate": 1.848048096156426e-05,
      "loss": 1.1753,
      "step": 531
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0878379344940186,
      "learning_rate": 1.8473868061840617e-05,
      "loss": 1.2025,
      "step": 532
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.068784236907959,
      "learning_rate": 1.8467241992282842e-05,
      "loss": 1.1268,
      "step": 533
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0175490379333496,
      "learning_rate": 1.8460602763188974e-05,
      "loss": 1.1566,
      "step": 534
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.350396156311035,
      "learning_rate": 1.8453950384877504e-05,
      "loss": 1.1803,
      "step": 535
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2505195140838623,
      "learning_rate": 1.844728486768736e-05,
      "loss": 1.1998,
      "step": 536
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.3242056369781494,
      "learning_rate": 1.8440606221977893e-05,
      "loss": 1.1301,
      "step": 537
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.415846347808838,
      "learning_rate": 1.843391445812886e-05,
      "loss": 1.1896,
      "step": 538
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.082308292388916,
      "learning_rate": 1.8427209586540392e-05,
      "loss": 1.1961,
      "step": 539
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1382784843444824,
      "learning_rate": 1.842049161763301e-05,
      "loss": 1.1835,
      "step": 540
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.074077844619751,
      "learning_rate": 1.841376056184758e-05,
      "loss": 1.2248,
      "step": 541
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9486840963363647,
      "learning_rate": 1.8407016429645305e-05,
      "loss": 1.1622,
      "step": 542
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.072636604309082,
      "learning_rate": 1.8400259231507716e-05,
      "loss": 1.2034,
      "step": 543
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0383877754211426,
      "learning_rate": 1.8393488977936654e-05,
      "loss": 1.1263,
      "step": 544
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1238882541656494,
      "learning_rate": 1.8386705679454243e-05,
      "loss": 1.1772,
      "step": 545
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2432312965393066,
      "learning_rate": 1.837990934660288e-05,
      "loss": 1.1211,
      "step": 546
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9180233478546143,
      "learning_rate": 1.8373099989945236e-05,
      "loss": 1.1764,
      "step": 547
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0255978107452393,
      "learning_rate": 1.83662776200642e-05,
      "loss": 1.1941,
      "step": 548
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.411198854446411,
      "learning_rate": 1.8359442247562896e-05,
      "loss": 1.078,
      "step": 549
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.229306936264038,
      "learning_rate": 1.8352593883064664e-05,
      "loss": 1.1524,
      "step": 550
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.142324686050415,
      "learning_rate": 1.834573253721303e-05,
      "loss": 1.1638,
      "step": 551
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2250072956085205,
      "learning_rate": 1.8338858220671683e-05,
      "loss": 1.2288,
      "step": 552
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1798598766326904,
      "learning_rate": 1.833197094412449e-05,
      "loss": 1.2524,
      "step": 553
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0497348308563232,
      "learning_rate": 1.8325070718275452e-05,
      "loss": 1.1675,
      "step": 554
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.291598320007324,
      "learning_rate": 1.8318157553848694e-05,
      "loss": 1.1924,
      "step": 555
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6750928163528442,
      "learning_rate": 1.8311231461588446e-05,
      "loss": 1.2015,
      "step": 556
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0865728855133057,
      "learning_rate": 1.8304292452259037e-05,
      "loss": 1.1349,
      "step": 557
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.015840530395508,
      "learning_rate": 1.8297340536644877e-05,
      "loss": 1.1444,
      "step": 558
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0729453563690186,
      "learning_rate": 1.8290375725550417e-05,
      "loss": 1.1294,
      "step": 559
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.928125023841858,
      "learning_rate": 1.8283398029800167e-05,
      "loss": 1.1408,
      "step": 560
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9908708333969116,
      "learning_rate": 1.827640746023865e-05,
      "loss": 1.1829,
      "step": 561
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1214394569396973,
      "learning_rate": 1.82694040277304e-05,
      "loss": 1.1245,
      "step": 562
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0230727195739746,
      "learning_rate": 1.826238774315995e-05,
      "loss": 1.1605,
      "step": 563
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.286651372909546,
      "learning_rate": 1.82553586174318e-05,
      "loss": 1.1619,
      "step": 564
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9647496938705444,
      "learning_rate": 1.82483166614704e-05,
      "loss": 1.1971,
      "step": 565
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9685417413711548,
      "learning_rate": 1.8241261886220155e-05,
      "loss": 1.2139,
      "step": 566
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.018923759460449,
      "learning_rate": 1.8234194302645393e-05,
      "loss": 1.193,
      "step": 567
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.986161708831787,
      "learning_rate": 1.8227113921730336e-05,
      "loss": 1.1273,
      "step": 568
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.161348819732666,
      "learning_rate": 1.8220020754479104e-05,
      "loss": 1.1814,
      "step": 569
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3438127040863037,
      "learning_rate": 1.821291481191568e-05,
      "loss": 1.1959,
      "step": 570
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2213973999023438,
      "learning_rate": 1.8205796105083917e-05,
      "loss": 1.213,
      "step": 571
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1624553203582764,
      "learning_rate": 1.8198664645047496e-05,
      "loss": 1.2213,
      "step": 572
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.156085968017578,
      "learning_rate": 1.819152044288992e-05,
      "loss": 1.2127,
      "step": 573
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.305025577545166,
      "learning_rate": 1.8184363509714494e-05,
      "loss": 1.2085,
      "step": 574
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9706476926803589,
      "learning_rate": 1.8177193856644315e-05,
      "loss": 1.1821,
      "step": 575
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1848785877227783,
      "learning_rate": 1.8170011494822242e-05,
      "loss": 1.1039,
      "step": 576
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9938225746154785,
      "learning_rate": 1.8162816435410892e-05,
      "loss": 1.0698,
      "step": 577
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.041311740875244,
      "learning_rate": 1.8155608689592604e-05,
      "loss": 1.1711,
      "step": 578
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2170591354370117,
      "learning_rate": 1.8148388268569453e-05,
      "loss": 1.2636,
      "step": 579
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.386672258377075,
      "learning_rate": 1.8141155183563195e-05,
      "loss": 1.1228,
      "step": 580
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2750444412231445,
      "learning_rate": 1.8133909445815277e-05,
      "loss": 1.1939,
      "step": 581
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2243592739105225,
      "learning_rate": 1.8126651066586818e-05,
      "loss": 1.1913,
      "step": 582
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9920014142990112,
      "learning_rate": 1.811938005715857e-05,
      "loss": 1.233,
      "step": 583
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0380051136016846,
      "learning_rate": 1.8112096428830912e-05,
      "loss": 1.1357,
      "step": 584
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2395403385162354,
      "learning_rate": 1.8104800192923856e-05,
      "loss": 1.1514,
      "step": 585
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.029069662094116,
      "learning_rate": 1.8097491360776987e-05,
      "loss": 1.144,
      "step": 586
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.656147003173828,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 1.167,
      "step": 587
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0938100814819336,
      "learning_rate": 1.8082835953220055e-05,
      "loss": 1.1741,
      "step": 588
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2160096168518066,
      "learning_rate": 1.8075489400586993e-05,
      "loss": 1.1589,
      "step": 589
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.17441987991333,
      "learning_rate": 1.806813029726808e-05,
      "loss": 1.1674,
      "step": 590
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2803690433502197,
      "learning_rate": 1.8060758654700622e-05,
      "loss": 1.1852,
      "step": 591
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.022815465927124,
      "learning_rate": 1.8053374484341403e-05,
      "loss": 1.1163,
      "step": 592
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3145856857299805,
      "learning_rate": 1.8045977797666685e-05,
      "loss": 1.1411,
      "step": 593
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3268139362335205,
      "learning_rate": 1.8038568606172172e-05,
      "loss": 1.1137,
      "step": 594
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.225128173828125,
      "learning_rate": 1.803114692137302e-05,
      "loss": 1.1411,
      "step": 595
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0605666637420654,
      "learning_rate": 1.8023712754803783e-05,
      "loss": 1.1914,
      "step": 596
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0572869777679443,
      "learning_rate": 1.801626611801842e-05,
      "loss": 1.185,
      "step": 597
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.601340889930725,
      "learning_rate": 1.8008807022590283e-05,
      "loss": 1.0972,
      "step": 598
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9462857246398926,
      "learning_rate": 1.8001335480112067e-05,
      "loss": 1.1978,
      "step": 599
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.079775094985962,
      "learning_rate": 1.7993851502195822e-05,
      "loss": 1.1856,
      "step": 600
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1858863830566406,
      "learning_rate": 1.798635510047293e-05,
      "loss": 1.1835,
      "step": 601
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.075789213180542,
      "learning_rate": 1.7978846286594073e-05,
      "loss": 1.1427,
      "step": 602
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.070838689804077,
      "learning_rate": 1.7971325072229227e-05,
      "loss": 1.158,
      "step": 603
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4548771381378174,
      "learning_rate": 1.7963791469067637e-05,
      "loss": 1.1648,
      "step": 604
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.247897148132324,
      "learning_rate": 1.795624548881781e-05,
      "loss": 1.1541,
      "step": 605
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.069981336593628,
      "learning_rate": 1.7948687143207485e-05,
      "loss": 1.1113,
      "step": 606
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.271416187286377,
      "learning_rate": 1.7941116443983612e-05,
      "loss": 1.1263,
      "step": 607
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.961619257926941,
      "learning_rate": 1.7933533402912354e-05,
      "loss": 1.1909,
      "step": 608
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3557465076446533,
      "learning_rate": 1.7925938031779044e-05,
      "loss": 1.1194,
      "step": 609
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.950345516204834,
      "learning_rate": 1.7918330342388183e-05,
      "loss": 1.1954,
      "step": 610
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.3135554790496826,
      "learning_rate": 1.7910710346563417e-05,
      "loss": 1.1986,
      "step": 611
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.398646593093872,
      "learning_rate": 1.7903078056147515e-05,
      "loss": 1.2256,
      "step": 612
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.052675485610962,
      "learning_rate": 1.7895433483002356e-05,
      "loss": 1.1716,
      "step": 613
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.07888126373291,
      "learning_rate": 1.7887776639008912e-05,
      "loss": 1.1009,
      "step": 614
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9149290323257446,
      "learning_rate": 1.788010753606722e-05,
      "loss": 1.1452,
      "step": 615
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0604777336120605,
      "learning_rate": 1.7872426186096374e-05,
      "loss": 1.2065,
      "step": 616
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4113054275512695,
      "learning_rate": 1.78647326010345e-05,
      "loss": 1.2169,
      "step": 617
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.307846784591675,
      "learning_rate": 1.785702679283874e-05,
      "loss": 1.2023,
      "step": 618
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.039536714553833,
      "learning_rate": 1.7849308773485226e-05,
      "loss": 1.1712,
      "step": 619
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.947080373764038,
      "learning_rate": 1.784157855496908e-05,
      "loss": 1.1891,
      "step": 620
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.001661539077759,
      "learning_rate": 1.783383614930438e-05,
      "loss": 1.0914,
      "step": 621
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2831873893737793,
      "learning_rate": 1.782608156852414e-05,
      "loss": 1.2025,
      "step": 622
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1427505016326904,
      "learning_rate": 1.78183148246803e-05,
      "loss": 1.083,
      "step": 623
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2457664012908936,
      "learning_rate": 1.7810535929843703e-05,
      "loss": 1.1776,
      "step": 624
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.366323947906494,
      "learning_rate": 1.780274489610407e-05,
      "loss": 1.1641,
      "step": 625
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.300736904144287,
      "learning_rate": 1.7794941735570005e-05,
      "loss": 1.1688,
      "step": 626
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.879165530204773,
      "learning_rate": 1.778712646036894e-05,
      "loss": 1.1896,
      "step": 627
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.146275043487549,
      "learning_rate": 1.777929908264715e-05,
      "loss": 1.1362,
      "step": 628
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.239445447921753,
      "learning_rate": 1.777145961456971e-05,
      "loss": 1.1505,
      "step": 629
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1150336265563965,
      "learning_rate": 1.7763608068320487e-05,
      "loss": 1.1324,
      "step": 630
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.136029005050659,
      "learning_rate": 1.7755744456102123e-05,
      "loss": 1.1711,
      "step": 631
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.187053680419922,
      "learning_rate": 1.7747868790136012e-05,
      "loss": 1.144,
      "step": 632
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.246870756149292,
      "learning_rate": 1.7739981082662275e-05,
      "loss": 1.1939,
      "step": 633
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.088726043701172,
      "learning_rate": 1.773208134593976e-05,
      "loss": 1.1463,
      "step": 634
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.26412296295166,
      "learning_rate": 1.7724169592245996e-05,
      "loss": 1.1222,
      "step": 635
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0142581462860107,
      "learning_rate": 1.7716245833877202e-05,
      "loss": 1.1667,
      "step": 636
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.025657892227173,
      "learning_rate": 1.7708310083148242e-05,
      "loss": 1.214,
      "step": 637
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0525450706481934,
      "learning_rate": 1.7700362352392632e-05,
      "loss": 1.0629,
      "step": 638
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2509586811065674,
      "learning_rate": 1.769240265396249e-05,
      "loss": 1.2436,
      "step": 639
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.227137327194214,
      "learning_rate": 1.7684431000228548e-05,
      "loss": 1.1507,
      "step": 640
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1595985889434814,
      "learning_rate": 1.7676447403580114e-05,
      "loss": 1.2298,
      "step": 641
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4204514026641846,
      "learning_rate": 1.7668451876425056e-05,
      "loss": 1.2093,
      "step": 642
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4298224449157715,
      "learning_rate": 1.766044443118978e-05,
      "loss": 1.1767,
      "step": 643
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1682794094085693,
      "learning_rate": 1.765242508031923e-05,
      "loss": 1.116,
      "step": 644
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9856442213058472,
      "learning_rate": 1.7644393836276832e-05,
      "loss": 1.1177,
      "step": 645
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.202268600463867,
      "learning_rate": 1.7636350711544518e-05,
      "loss": 1.1481,
      "step": 646
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.19736647605896,
      "learning_rate": 1.7628295718622666e-05,
      "loss": 1.0957,
      "step": 647
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1993982791900635,
      "learning_rate": 1.762022887003011e-05,
      "loss": 1.2132,
      "step": 648
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.133937120437622,
      "learning_rate": 1.7612150178304102e-05,
      "loss": 1.1629,
      "step": 649
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0732550621032715,
      "learning_rate": 1.7604059656000313e-05,
      "loss": 1.1535,
      "step": 650
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8752312660217285,
      "learning_rate": 1.7595957315692782e-05,
      "loss": 1.1758,
      "step": 651
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1404361724853516,
      "learning_rate": 1.7587843169973934e-05,
      "loss": 1.1827,
      "step": 652
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.3998889923095703,
      "learning_rate": 1.757971723145453e-05,
      "loss": 1.1293,
      "step": 653
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.098823308944702,
      "learning_rate": 1.7571579512763665e-05,
      "loss": 1.1539,
      "step": 654
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1867806911468506,
      "learning_rate": 1.7563430026548737e-05,
      "loss": 1.1718,
      "step": 655
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.082909107208252,
      "learning_rate": 1.7555268785475438e-05,
      "loss": 1.1722,
      "step": 656
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.053598165512085,
      "learning_rate": 1.7547095802227723e-05,
      "loss": 1.2109,
      "step": 657
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9640586376190186,
      "learning_rate": 1.75389110895078e-05,
      "loss": 1.1451,
      "step": 658
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9443564414978027,
      "learning_rate": 1.7530714660036112e-05,
      "loss": 1.1925,
      "step": 659
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0698344707489014,
      "learning_rate": 1.7522506526551303e-05,
      "loss": 1.1967,
      "step": 660
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.556574821472168,
      "learning_rate": 1.7514286701810203e-05,
      "loss": 1.0821,
      "step": 661
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1912436485290527,
      "learning_rate": 1.7506055198587835e-05,
      "loss": 1.1969,
      "step": 662
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3437693119049072,
      "learning_rate": 1.7497812029677344e-05,
      "loss": 1.1144,
      "step": 663
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1868386268615723,
      "learning_rate": 1.7489557207890025e-05,
      "loss": 1.151,
      "step": 664
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.4069347381591797,
      "learning_rate": 1.748129074605527e-05,
      "loss": 1.1508,
      "step": 665
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0951108932495117,
      "learning_rate": 1.7473012657020577e-05,
      "loss": 1.2108,
      "step": 666
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2041478157043457,
      "learning_rate": 1.7464722953651504e-05,
      "loss": 1.0997,
      "step": 667
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.270289182662964,
      "learning_rate": 1.7456421648831658e-05,
      "loss": 1.156,
      "step": 668
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.126605272293091,
      "learning_rate": 1.7448108755462684e-05,
      "loss": 1.204,
      "step": 669
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1846349239349365,
      "learning_rate": 1.7439784286464234e-05,
      "loss": 1.1449,
      "step": 670
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.4118194580078125,
      "learning_rate": 1.7431448254773943e-05,
      "loss": 1.1971,
      "step": 671
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1178319454193115,
      "learning_rate": 1.7423100673347435e-05,
      "loss": 1.121,
      "step": 672
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.218557834625244,
      "learning_rate": 1.741474155515827e-05,
      "loss": 1.1807,
      "step": 673
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.028425693511963,
      "learning_rate": 1.7406370913197938e-05,
      "loss": 1.1747,
      "step": 674
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.427220582962036,
      "learning_rate": 1.7397988760475842e-05,
      "loss": 1.1665,
      "step": 675
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1961441040039062,
      "learning_rate": 1.738959511001928e-05,
      "loss": 1.1221,
      "step": 676
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.139732599258423,
      "learning_rate": 1.738118997487341e-05,
      "loss": 1.1311,
      "step": 677
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1087067127227783,
      "learning_rate": 1.737277336810124e-05,
      "loss": 1.1268,
      "step": 678
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.931382656097412,
      "learning_rate": 1.736434530278362e-05,
      "loss": 1.1601,
      "step": 679
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0525057315826416,
      "learning_rate": 1.7355905792019195e-05,
      "loss": 1.1196,
      "step": 680
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.227074146270752,
      "learning_rate": 1.73474548489244e-05,
      "loss": 1.1166,
      "step": 681
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8980870246887207,
      "learning_rate": 1.7338992486633438e-05,
      "loss": 1.1529,
      "step": 682
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1788265705108643,
      "learning_rate": 1.7330518718298263e-05,
      "loss": 1.0753,
      "step": 683
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0133700370788574,
      "learning_rate": 1.732203355708856e-05,
      "loss": 1.1459,
      "step": 684
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0878899097442627,
      "learning_rate": 1.7313537016191706e-05,
      "loss": 1.1556,
      "step": 685
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.703956127166748,
      "learning_rate": 1.7305029108812777e-05,
      "loss": 1.0564,
      "step": 686
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1182374954223633,
      "learning_rate": 1.729650984817451e-05,
      "loss": 1.0801,
      "step": 687
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9487473964691162,
      "learning_rate": 1.7287979247517285e-05,
      "loss": 1.165,
      "step": 688
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1141397953033447,
      "learning_rate": 1.727943732009911e-05,
      "loss": 1.1555,
      "step": 689
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2833237648010254,
      "learning_rate": 1.7270884079195596e-05,
      "loss": 1.1031,
      "step": 690
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.036330223083496,
      "learning_rate": 1.726231953809993e-05,
      "loss": 1.1775,
      "step": 691
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2103805541992188,
      "learning_rate": 1.7253743710122877e-05,
      "loss": 1.1232,
      "step": 692
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.073535919189453,
      "learning_rate": 1.7245156608592727e-05,
      "loss": 1.1352,
      "step": 693
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0641415119171143,
      "learning_rate": 1.7236558246855303e-05,
      "loss": 1.2011,
      "step": 694
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.3134775161743164,
      "learning_rate": 1.7227948638273918e-05,
      "loss": 1.1321,
      "step": 695
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.4093737602233887,
      "learning_rate": 1.721932779622937e-05,
      "loss": 1.221,
      "step": 696
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2581515312194824,
      "learning_rate": 1.7210695734119926e-05,
      "loss": 1.1173,
      "step": 697
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9901444911956787,
      "learning_rate": 1.7202052465361268e-05,
      "loss": 1.1732,
      "step": 698
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.408879041671753,
      "learning_rate": 1.7193398003386514e-05,
      "loss": 1.2302,
      "step": 699
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.330909252166748,
      "learning_rate": 1.7184732361646168e-05,
      "loss": 1.199,
      "step": 700
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1203629970550537,
      "learning_rate": 1.717605555360812e-05,
      "loss": 1.2316,
      "step": 701
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0897858142852783,
      "learning_rate": 1.71673675927576e-05,
      "loss": 1.1191,
      "step": 702
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.11863112449646,
      "learning_rate": 1.7158668492597186e-05,
      "loss": 1.1344,
      "step": 703
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2705633640289307,
      "learning_rate": 1.7149958266646756e-05,
      "loss": 1.1263,
      "step": 704
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1114447116851807,
      "learning_rate": 1.7141236928443482e-05,
      "loss": 1.0987,
      "step": 705
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1904983520507812,
      "learning_rate": 1.713250449154182e-05,
      "loss": 1.1961,
      "step": 706
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.040287494659424,
      "learning_rate": 1.712376096951345e-05,
      "loss": 1.1187,
      "step": 707
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2336745262145996,
      "learning_rate": 1.7115006375947304e-05,
      "loss": 1.1351,
      "step": 708
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2484261989593506,
      "learning_rate": 1.7106240724449507e-05,
      "loss": 1.1791,
      "step": 709
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2028791904449463,
      "learning_rate": 1.7097464028643376e-05,
      "loss": 1.1408,
      "step": 710
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1362013816833496,
      "learning_rate": 1.7088676302169394e-05,
      "loss": 1.0836,
      "step": 711
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0745632648468018,
      "learning_rate": 1.7079877558685178e-05,
      "loss": 1.1753,
      "step": 712
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0526390075683594,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 1.2456,
      "step": 713
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.028933048248291,
      "learning_rate": 1.7062247075402137e-05,
      "loss": 1.1213,
      "step": 714
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.281759023666382,
      "learning_rate": 1.705341536300409e-05,
      "loss": 1.127,
      "step": 715
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.204625129699707,
      "learning_rate": 1.7044572688397316e-05,
      "loss": 1.2269,
      "step": 716
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0828142166137695,
      "learning_rate": 1.7035719065324837e-05,
      "loss": 1.1954,
      "step": 717
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0966272354125977,
      "learning_rate": 1.7026854507546694e-05,
      "loss": 1.1656,
      "step": 718
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.214087724685669,
      "learning_rate": 1.7017979028839918e-05,
      "loss": 1.1096,
      "step": 719
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2793142795562744,
      "learning_rate": 1.700909264299851e-05,
      "loss": 1.1627,
      "step": 720
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.10190486907959,
      "learning_rate": 1.7000195363833434e-05,
      "loss": 1.1274,
      "step": 721
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.239226818084717,
      "learning_rate": 1.6991287205172575e-05,
      "loss": 1.2142,
      "step": 722
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.946247935295105,
      "learning_rate": 1.698236818086073e-05,
      "loss": 1.1964,
      "step": 723
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.3132076263427734,
      "learning_rate": 1.697343830475958e-05,
      "loss": 1.1892,
      "step": 724
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0867323875427246,
      "learning_rate": 1.696449759074767e-05,
      "loss": 1.1795,
      "step": 725
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.090459108352661,
      "learning_rate": 1.6955546052720398e-05,
      "loss": 1.0905,
      "step": 726
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.031367778778076,
      "learning_rate": 1.6946583704589973e-05,
      "loss": 1.2218,
      "step": 727
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0468785762786865,
      "learning_rate": 1.693761056028542e-05,
      "loss": 1.2094,
      "step": 728
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9751163721084595,
      "learning_rate": 1.692862663375252e-05,
      "loss": 1.1313,
      "step": 729
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2197937965393066,
      "learning_rate": 1.6919631938953835e-05,
      "loss": 1.1389,
      "step": 730
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0486156940460205,
      "learning_rate": 1.691062648986865e-05,
      "loss": 1.1651,
      "step": 731
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1628472805023193,
      "learning_rate": 1.6901610300492963e-05,
      "loss": 1.1026,
      "step": 732
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.039085626602173,
      "learning_rate": 1.689258338483947e-05,
      "loss": 1.1639,
      "step": 733
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.093853712081909,
      "learning_rate": 1.688354575693754e-05,
      "loss": 1.2254,
      "step": 734
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.021364212036133,
      "learning_rate": 1.6874497430833182e-05,
      "loss": 1.1859,
      "step": 735
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2229232788085938,
      "learning_rate": 1.6865438420589036e-05,
      "loss": 1.1465,
      "step": 736
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1148059368133545,
      "learning_rate": 1.6856368740284342e-05,
      "loss": 1.1167,
      "step": 737
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2569079399108887,
      "learning_rate": 1.6847288404014937e-05,
      "loss": 1.141,
      "step": 738
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.3591811656951904,
      "learning_rate": 1.68381974258932e-05,
      "loss": 1.1725,
      "step": 739
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1351094245910645,
      "learning_rate": 1.682909582004807e-05,
      "loss": 1.1446,
      "step": 740
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9958903789520264,
      "learning_rate": 1.6819983600624986e-05,
      "loss": 1.1324,
      "step": 741
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0584397315979004,
      "learning_rate": 1.681086078178589e-05,
      "loss": 1.1953,
      "step": 742
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.199082136154175,
      "learning_rate": 1.6801727377709195e-05,
      "loss": 1.1448,
      "step": 743
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.151099443435669,
      "learning_rate": 1.679258340258977e-05,
      "loss": 1.2052,
      "step": 744
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0236334800720215,
      "learning_rate": 1.6783428870638904e-05,
      "loss": 1.1508,
      "step": 745
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0364227294921875,
      "learning_rate": 1.6774263796084304e-05,
      "loss": 1.2201,
      "step": 746
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.159165382385254,
      "learning_rate": 1.6765088193170055e-05,
      "loss": 1.1104,
      "step": 747
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.054800033569336,
      "learning_rate": 1.6755902076156606e-05,
      "loss": 1.1166,
      "step": 748
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9964845180511475,
      "learning_rate": 1.6746705459320746e-05,
      "loss": 1.1366,
      "step": 749
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.965458869934082,
      "learning_rate": 1.6737498356955586e-05,
      "loss": 1.1378,
      "step": 750
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1688618659973145,
      "learning_rate": 1.672828078337053e-05,
      "loss": 1.177,
      "step": 751
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9804893732070923,
      "learning_rate": 1.6719052752891262e-05,
      "loss": 1.1398,
      "step": 752
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.179184913635254,
      "learning_rate": 1.67098142798597e-05,
      "loss": 1.1667,
      "step": 753
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7952988147735596,
      "learning_rate": 1.6700565378634022e-05,
      "loss": 1.1012,
      "step": 754
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0727527141571045,
      "learning_rate": 1.6691306063588583e-05,
      "loss": 1.1078,
      "step": 755
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9869143962860107,
      "learning_rate": 1.668203634911394e-05,
      "loss": 1.084,
      "step": 756
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.100505828857422,
      "learning_rate": 1.667275624961681e-05,
      "loss": 1.2108,
      "step": 757
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9542704820632935,
      "learning_rate": 1.6663465779520042e-05,
      "loss": 1.1498,
      "step": 758
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.039180040359497,
      "learning_rate": 1.6654164953262614e-05,
      "loss": 1.1408,
      "step": 759
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1117851734161377,
      "learning_rate": 1.6644853785299592e-05,
      "loss": 1.2212,
      "step": 760
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1890995502471924,
      "learning_rate": 1.6635532290102114e-05,
      "loss": 1.1109,
      "step": 761
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7709732055664062,
      "learning_rate": 1.6626200482157378e-05,
      "loss": 1.1881,
      "step": 762
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8674650192260742,
      "learning_rate": 1.6616858375968596e-05,
      "loss": 1.1293,
      "step": 763
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0869269371032715,
      "learning_rate": 1.6607505986054998e-05,
      "loss": 1.1495,
      "step": 764
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1247758865356445,
      "learning_rate": 1.6598143326951784e-05,
      "loss": 1.0665,
      "step": 765
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.134139060974121,
      "learning_rate": 1.6588770413210126e-05,
      "loss": 1.0951,
      "step": 766
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.197136402130127,
      "learning_rate": 1.657938725939713e-05,
      "loss": 1.2169,
      "step": 767
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1024904251098633,
      "learning_rate": 1.6569993880095807e-05,
      "loss": 1.1397,
      "step": 768
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1600143909454346,
      "learning_rate": 1.6560590289905074e-05,
      "loss": 1.1836,
      "step": 769
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0708024501800537,
      "learning_rate": 1.6551176503439713e-05,
      "loss": 1.2304,
      "step": 770
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.087913990020752,
      "learning_rate": 1.6541752535330345e-05,
      "loss": 1.1246,
      "step": 771
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9836457967758179,
      "learning_rate": 1.6532318400223427e-05,
      "loss": 1.1276,
      "step": 772
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.071894645690918,
      "learning_rate": 1.6522874112781213e-05,
      "loss": 1.1228,
      "step": 773
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8972511291503906,
      "learning_rate": 1.651341968768173e-05,
      "loss": 1.1931,
      "step": 774
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.127654552459717,
      "learning_rate": 1.6503955139618765e-05,
      "loss": 1.1551,
      "step": 775
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.083272695541382,
      "learning_rate": 1.6494480483301836e-05,
      "loss": 1.1338,
      "step": 776
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1229493618011475,
      "learning_rate": 1.6484995733456178e-05,
      "loss": 1.1048,
      "step": 777
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0119144916534424,
      "learning_rate": 1.6475500904822707e-05,
      "loss": 1.1666,
      "step": 778
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1689565181732178,
      "learning_rate": 1.6465996012157996e-05,
      "loss": 1.1725,
      "step": 779
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1973772048950195,
      "learning_rate": 1.6456481070234277e-05,
      "loss": 1.0835,
      "step": 780
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2673871517181396,
      "learning_rate": 1.6446956093839385e-05,
      "loss": 1.1408,
      "step": 781
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9054604768753052,
      "learning_rate": 1.6437421097776754e-05,
      "loss": 1.1873,
      "step": 782
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.193080425262451,
      "learning_rate": 1.6427876096865394e-05,
      "loss": 1.1305,
      "step": 783
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.322634696960449,
      "learning_rate": 1.6418321105939863e-05,
      "loss": 1.2132,
      "step": 784
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.330106496810913,
      "learning_rate": 1.6408756139850243e-05,
      "loss": 1.1827,
      "step": 785
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9974806308746338,
      "learning_rate": 1.639918121346212e-05,
      "loss": 1.1297,
      "step": 786
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8104221820831299,
      "learning_rate": 1.638959634165656e-05,
      "loss": 1.1327,
      "step": 787
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1339352130889893,
      "learning_rate": 1.6380001539330088e-05,
      "loss": 1.1574,
      "step": 788
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2013633251190186,
      "learning_rate": 1.637039682139466e-05,
      "loss": 1.134,
      "step": 789
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2388668060302734,
      "learning_rate": 1.636078220277764e-05,
      "loss": 1.1431,
      "step": 790
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.1685380935668945,
      "learning_rate": 1.635115769842179e-05,
      "loss": 1.1631,
      "step": 791
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.215772867202759,
      "learning_rate": 1.634152332328522e-05,
      "loss": 1.1616,
      "step": 792
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9519283771514893,
      "learning_rate": 1.6331879092341402e-05,
      "loss": 1.138,
      "step": 793
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1373023986816406,
      "learning_rate": 1.63222250205791e-05,
      "loss": 1.1667,
      "step": 794
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0924360752105713,
      "learning_rate": 1.631256112300239e-05,
      "loss": 1.0863,
      "step": 795
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0675132274627686,
      "learning_rate": 1.630288741463062e-05,
      "loss": 1.1274,
      "step": 796
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.349987506866455,
      "learning_rate": 1.6293203910498375e-05,
      "loss": 1.1585,
      "step": 797
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8733530044555664,
      "learning_rate": 1.6283510625655474e-05,
      "loss": 1.0967,
      "step": 798
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.133479356765747,
      "learning_rate": 1.6273807575166927e-05,
      "loss": 1.2104,
      "step": 799
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1441917419433594,
      "learning_rate": 1.626409477411293e-05,
      "loss": 1.1678,
      "step": 800
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0580782890319824,
      "learning_rate": 1.625437223758883e-05,
      "loss": 1.1165,
      "step": 801
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0730810165405273,
      "learning_rate": 1.6244639980705104e-05,
      "loss": 1.1501,
      "step": 802
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0370750427246094,
      "learning_rate": 1.6234898018587336e-05,
      "loss": 1.1763,
      "step": 803
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.206151008605957,
      "learning_rate": 1.6225146366376198e-05,
      "loss": 1.2406,
      "step": 804
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.094416618347168,
      "learning_rate": 1.621538503922741e-05,
      "loss": 1.1851,
      "step": 805
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9711549282073975,
      "learning_rate": 1.620561405231175e-05,
      "loss": 1.1747,
      "step": 806
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.113659381866455,
      "learning_rate": 1.6195833420814983e-05,
      "loss": 1.1771,
      "step": 807
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1694793701171875,
      "learning_rate": 1.6186043159937884e-05,
      "loss": 1.1985,
      "step": 808
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2843434810638428,
      "learning_rate": 1.617624328489618e-05,
      "loss": 1.1653,
      "step": 809
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1957004070281982,
      "learning_rate": 1.616643381092055e-05,
      "loss": 1.1137,
      "step": 810
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.260986566543579,
      "learning_rate": 1.6156614753256583e-05,
      "loss": 1.1346,
      "step": 811
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.043829917907715,
      "learning_rate": 1.6146786127164773e-05,
      "loss": 1.1787,
      "step": 812
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7588573694229126,
      "learning_rate": 1.6136947947920477e-05,
      "loss": 1.1207,
      "step": 813
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.14469838142395,
      "learning_rate": 1.6127100230813897e-05,
      "loss": 1.1379,
      "step": 814
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1047861576080322,
      "learning_rate": 1.6117242991150064e-05,
      "loss": 1.1225,
      "step": 815
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9796526432037354,
      "learning_rate": 1.610737624424881e-05,
      "loss": 1.0919,
      "step": 816
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.971156120300293,
      "learning_rate": 1.609750000544474e-05,
      "loss": 1.1358,
      "step": 817
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.128633737564087,
      "learning_rate": 1.608761429008721e-05,
      "loss": 1.1649,
      "step": 818
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.117347240447998,
      "learning_rate": 1.6077719113540303e-05,
      "loss": 1.1571,
      "step": 819
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1148552894592285,
      "learning_rate": 1.6067814491182806e-05,
      "loss": 1.1429,
      "step": 820
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0618293285369873,
      "learning_rate": 1.60579004384082e-05,
      "loss": 1.1249,
      "step": 821
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2234442234039307,
      "learning_rate": 1.6047976970624603e-05,
      "loss": 1.0955,
      "step": 822
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9246585369110107,
      "learning_rate": 1.6038044103254775e-05,
      "loss": 1.1801,
      "step": 823
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2353978157043457,
      "learning_rate": 1.6028101851736083e-05,
      "loss": 1.1644,
      "step": 824
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0133609771728516,
      "learning_rate": 1.6018150231520486e-05,
      "loss": 1.1617,
      "step": 825
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.278444766998291,
      "learning_rate": 1.6008189258074483e-05,
      "loss": 1.1011,
      "step": 826
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.057840585708618,
      "learning_rate": 1.599821894687914e-05,
      "loss": 1.1845,
      "step": 827
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.133265733718872,
      "learning_rate": 1.5988239313430004e-05,
      "loss": 1.1036,
      "step": 828
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.023371458053589,
      "learning_rate": 1.5978250373237132e-05,
      "loss": 1.1568,
      "step": 829
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.954972267150879,
      "learning_rate": 1.5968252141825038e-05,
      "loss": 1.1383,
      "step": 830
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.121967077255249,
      "learning_rate": 1.5958244634732673e-05,
      "loss": 1.1382,
      "step": 831
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9449777603149414,
      "learning_rate": 1.5948227867513416e-05,
      "loss": 1.1744,
      "step": 832
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8645703792572021,
      "learning_rate": 1.5938201855735017e-05,
      "loss": 1.1535,
      "step": 833
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.153003454208374,
      "learning_rate": 1.5928166614979613e-05,
      "loss": 1.168,
      "step": 834
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.206799268722534,
      "learning_rate": 1.591812216084368e-05,
      "loss": 1.1329,
      "step": 835
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.241159439086914,
      "learning_rate": 1.5908068508938e-05,
      "loss": 1.103,
      "step": 836
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.193718910217285,
      "learning_rate": 1.5898005674887673e-05,
      "loss": 1.1155,
      "step": 837
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2546944618225098,
      "learning_rate": 1.5887933674332048e-05,
      "loss": 1.1644,
      "step": 838
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1780433654785156,
      "learning_rate": 1.5877852522924733e-05,
      "loss": 1.1723,
      "step": 839
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0463407039642334,
      "learning_rate": 1.5867762236333553e-05,
      "loss": 1.1072,
      "step": 840
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0950937271118164,
      "learning_rate": 1.585766283024053e-05,
      "loss": 1.1356,
      "step": 841
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.013115644454956,
      "learning_rate": 1.5847554320341865e-05,
      "loss": 1.1609,
      "step": 842
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0511467456817627,
      "learning_rate": 1.5837436722347902e-05,
      "loss": 1.1558,
      "step": 843
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9035463333129883,
      "learning_rate": 1.582731005198311e-05,
      "loss": 1.1615,
      "step": 844
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.969377040863037,
      "learning_rate": 1.581717432498606e-05,
      "loss": 1.1826,
      "step": 845
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.134600877761841,
      "learning_rate": 1.5807029557109398e-05,
      "loss": 1.1223,
      "step": 846
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1350183486938477,
      "learning_rate": 1.5796875764119826e-05,
      "loss": 1.1053,
      "step": 847
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0572690963745117,
      "learning_rate": 1.578671296179806e-05,
      "loss": 1.1608,
      "step": 848
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.936173439025879,
      "learning_rate": 1.577654116593883e-05,
      "loss": 1.1374,
      "step": 849
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2613766193389893,
      "learning_rate": 1.5766360392350837e-05,
      "loss": 1.2149,
      "step": 850
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0538418292999268,
      "learning_rate": 1.575617065685674e-05,
      "loss": 1.129,
      "step": 851
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0415163040161133,
      "learning_rate": 1.5745971975293114e-05,
      "loss": 1.1239,
      "step": 852
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.498513698577881,
      "learning_rate": 1.573576436351046e-05,
      "loss": 1.1144,
      "step": 853
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.042130470275879,
      "learning_rate": 1.5725547837373142e-05,
      "loss": 1.1141,
      "step": 854
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.00840163230896,
      "learning_rate": 1.5715322412759374e-05,
      "loss": 1.1094,
      "step": 855
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.053180456161499,
      "learning_rate": 1.5705088105561217e-05,
      "loss": 1.0236,
      "step": 856
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1435821056365967,
      "learning_rate": 1.569484493168452e-05,
      "loss": 1.1271,
      "step": 857
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.064344882965088,
      "learning_rate": 1.5684592907048925e-05,
      "loss": 1.1532,
      "step": 858
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0792386531829834,
      "learning_rate": 1.567433204758782e-05,
      "loss": 1.1325,
      "step": 859
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.210310697555542,
      "learning_rate": 1.566406236924833e-05,
      "loss": 1.1135,
      "step": 860
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3632009029388428,
      "learning_rate": 1.5653783887991282e-05,
      "loss": 1.112,
      "step": 861
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1095528602600098,
      "learning_rate": 1.564349661979119e-05,
      "loss": 1.1545,
      "step": 862
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.080655813217163,
      "learning_rate": 1.563320058063622e-05,
      "loss": 1.1047,
      "step": 863
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.093625783920288,
      "learning_rate": 1.5622895786528173e-05,
      "loss": 1.1247,
      "step": 864
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.934838056564331,
      "learning_rate": 1.5612582253482444e-05,
      "loss": 1.1148,
      "step": 865
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0245628356933594,
      "learning_rate": 1.5602259997528028e-05,
      "loss": 1.1895,
      "step": 866
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.249128818511963,
      "learning_rate": 1.5591929034707468e-05,
      "loss": 1.1128,
      "step": 867
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.5355708599090576,
      "learning_rate": 1.5581589381076843e-05,
      "loss": 1.1866,
      "step": 868
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.170046329498291,
      "learning_rate": 1.5571241052705724e-05,
      "loss": 1.1531,
      "step": 869
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.2702717781066895,
      "learning_rate": 1.5560884065677188e-05,
      "loss": 1.1574,
      "step": 870
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.154869556427002,
      "learning_rate": 1.5550518436087753e-05,
      "loss": 1.1182,
      "step": 871
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.127877950668335,
      "learning_rate": 1.5540144180047367e-05,
      "loss": 1.1307,
      "step": 872
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0092933177948,
      "learning_rate": 1.5529761313679396e-05,
      "loss": 1.1544,
      "step": 873
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1401944160461426,
      "learning_rate": 1.5519369853120584e-05,
      "loss": 1.1688,
      "step": 874
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0076067447662354,
      "learning_rate": 1.5508969814521026e-05,
      "loss": 1.1658,
      "step": 875
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1093618869781494,
      "learning_rate": 1.549856121404416e-05,
      "loss": 1.1394,
      "step": 876
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.043623685836792,
      "learning_rate": 1.548814406786671e-05,
      "loss": 1.1376,
      "step": 877
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9477869272232056,
      "learning_rate": 1.5477718392178716e-05,
      "loss": 1.1288,
      "step": 878
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2440407276153564,
      "learning_rate": 1.5467284203183437e-05,
      "loss": 1.1644,
      "step": 879
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.01481032371521,
      "learning_rate": 1.5456841517097385e-05,
      "loss": 1.1938,
      "step": 880
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.4178969860076904,
      "learning_rate": 1.5446390350150272e-05,
      "loss": 1.1307,
      "step": 881
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.020395278930664,
      "learning_rate": 1.5435930718584993e-05,
      "loss": 1.062,
      "step": 882
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.171915292739868,
      "learning_rate": 1.5425462638657597e-05,
      "loss": 1.171,
      "step": 883
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8823093175888062,
      "learning_rate": 1.541498612663726e-05,
      "loss": 1.1573,
      "step": 884
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.008413553237915,
      "learning_rate": 1.5404501198806267e-05,
      "loss": 1.1316,
      "step": 885
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.188957929611206,
      "learning_rate": 1.5394007871459983e-05,
      "loss": 1.1189,
      "step": 886
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9325833320617676,
      "learning_rate": 1.5383506160906826e-05,
      "loss": 1.1379,
      "step": 887
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2164764404296875,
      "learning_rate": 1.5372996083468242e-05,
      "loss": 1.1378,
      "step": 888
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0638344287872314,
      "learning_rate": 1.5362477655478677e-05,
      "loss": 1.1864,
      "step": 889
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0723531246185303,
      "learning_rate": 1.535195089328557e-05,
      "loss": 1.1466,
      "step": 890
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1423823833465576,
      "learning_rate": 1.534141581324929e-05,
      "loss": 1.1394,
      "step": 891
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8938361406326294,
      "learning_rate": 1.533087243174315e-05,
      "loss": 1.129,
      "step": 892
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.089263916015625,
      "learning_rate": 1.5320320765153367e-05,
      "loss": 1.2059,
      "step": 893
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.958177089691162,
      "learning_rate": 1.530976082987902e-05,
      "loss": 1.1073,
      "step": 894
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1485514640808105,
      "learning_rate": 1.529919264233205e-05,
      "loss": 1.1873,
      "step": 895
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.102787494659424,
      "learning_rate": 1.528861621893722e-05,
      "loss": 1.1478,
      "step": 896
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.157888889312744,
      "learning_rate": 1.527803157613209e-05,
      "loss": 1.1182,
      "step": 897
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.99153470993042,
      "learning_rate": 1.526743873036701e-05,
      "loss": 1.2166,
      "step": 898
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.270622730255127,
      "learning_rate": 1.5256837698105047e-05,
      "loss": 1.114,
      "step": 899
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.954586386680603,
      "learning_rate": 1.5246228495822025e-05,
      "loss": 1.1746,
      "step": 900
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0639493465423584,
      "learning_rate": 1.5235611140006446e-05,
      "loss": 1.1343,
      "step": 901
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9895256757736206,
      "learning_rate": 1.5224985647159489e-05,
      "loss": 1.1593,
      "step": 902
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.222998857498169,
      "learning_rate": 1.5214352033794981e-05,
      "loss": 1.2017,
      "step": 903
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9380887746810913,
      "learning_rate": 1.5203710316439368e-05,
      "loss": 1.1002,
      "step": 904
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3405141830444336,
      "learning_rate": 1.5193060511631692e-05,
      "loss": 1.1501,
      "step": 905
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7035748958587646,
      "learning_rate": 1.518240263592356e-05,
      "loss": 1.1144,
      "step": 906
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9584877490997314,
      "learning_rate": 1.5171736705879127e-05,
      "loss": 1.1731,
      "step": 907
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.170832395553589,
      "learning_rate": 1.5161062738075068e-05,
      "loss": 1.1197,
      "step": 908
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2153236865997314,
      "learning_rate": 1.5150380749100545e-05,
      "loss": 1.1433,
      "step": 909
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2203285694122314,
      "learning_rate": 1.5139690755557186e-05,
      "loss": 1.1707,
      "step": 910
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.048740863800049,
      "learning_rate": 1.5128992774059063e-05,
      "loss": 1.1395,
      "step": 911
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0772135257720947,
      "learning_rate": 1.5118286821232661e-05,
      "loss": 1.2392,
      "step": 912
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0804805755615234,
      "learning_rate": 1.5107572913716859e-05,
      "loss": 1.1453,
      "step": 913
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0957820415496826,
      "learning_rate": 1.5096851068162885e-05,
      "loss": 1.1692,
      "step": 914
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.181565761566162,
      "learning_rate": 1.5086121301234318e-05,
      "loss": 1.0533,
      "step": 915
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2094130516052246,
      "learning_rate": 1.5075383629607043e-05,
      "loss": 1.1554,
      "step": 916
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.048105001449585,
      "learning_rate": 1.5064638069969228e-05,
      "loss": 1.1618,
      "step": 917
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.823391079902649,
      "learning_rate": 1.505388463902131e-05,
      "loss": 1.0297,
      "step": 918
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.126972198486328,
      "learning_rate": 1.5043123353475944e-05,
      "loss": 1.0788,
      "step": 919
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.01954984664917,
      "learning_rate": 1.5032354230058004e-05,
      "loss": 1.1037,
      "step": 920
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1041252613067627,
      "learning_rate": 1.5021577285504538e-05,
      "loss": 1.179,
      "step": 921
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0628104209899902,
      "learning_rate": 1.5010792536564762e-05,
      "loss": 1.1522,
      "step": 922
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.147951364517212,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.1951,
      "step": 923
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0865650177001953,
      "learning_rate": 1.4989199692583705e-05,
      "loss": 1.1296,
      "step": 924
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9865614175796509,
      "learning_rate": 1.4978391631101383e-05,
      "loss": 1.1465,
      "step": 925
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9702848196029663,
      "learning_rate": 1.4967575832350614e-05,
      "loss": 1.1819,
      "step": 926
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0604124069213867,
      "learning_rate": 1.4956752313140978e-05,
      "loss": 1.2182,
      "step": 927
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9728455543518066,
      "learning_rate": 1.4945921090294076e-05,
      "loss": 1.108,
      "step": 928
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0167994499206543,
      "learning_rate": 1.493508218064347e-05,
      "loss": 1.1,
      "step": 929
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0446083545684814,
      "learning_rate": 1.4924235601034673e-05,
      "loss": 1.1384,
      "step": 930
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.293241024017334,
      "learning_rate": 1.4913381368325115e-05,
      "loss": 1.1137,
      "step": 931
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1402266025543213,
      "learning_rate": 1.4902519499384122e-05,
      "loss": 1.1661,
      "step": 932
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.11446213722229,
      "learning_rate": 1.4891650011092896e-05,
      "loss": 1.1379,
      "step": 933
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.109114170074463,
      "learning_rate": 1.488077292034446e-05,
      "loss": 1.1509,
      "step": 934
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8482778072357178,
      "learning_rate": 1.4869888244043674e-05,
      "loss": 1.1741,
      "step": 935
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8298062086105347,
      "learning_rate": 1.4858995999107175e-05,
      "loss": 1.1803,
      "step": 936
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.021455764770508,
      "learning_rate": 1.4848096202463373e-05,
      "loss": 1.1559,
      "step": 937
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0921902656555176,
      "learning_rate": 1.4837188871052399e-05,
      "loss": 1.1233,
      "step": 938
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0836234092712402,
      "learning_rate": 1.482627402182611e-05,
      "loss": 1.1029,
      "step": 939
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.170893907546997,
      "learning_rate": 1.4815351671748039e-05,
      "loss": 1.1373,
      "step": 940
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0577144622802734,
      "learning_rate": 1.4804421837793379e-05,
      "loss": 1.0926,
      "step": 941
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8858768939971924,
      "learning_rate": 1.479348453694895e-05,
      "loss": 1.1451,
      "step": 942
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1611320972442627,
      "learning_rate": 1.4782539786213184e-05,
      "loss": 1.1528,
      "step": 943
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0309722423553467,
      "learning_rate": 1.4771587602596085e-05,
      "loss": 1.1216,
      "step": 944
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.096972942352295,
      "learning_rate": 1.4760628003119213e-05,
      "loss": 1.1027,
      "step": 945
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.112083673477173,
      "learning_rate": 1.4749661004815654e-05,
      "loss": 1.1119,
      "step": 946
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.114081859588623,
      "learning_rate": 1.4738686624729987e-05,
      "loss": 1.1746,
      "step": 947
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9607064723968506,
      "learning_rate": 1.4727704879918272e-05,
      "loss": 1.1162,
      "step": 948
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2203288078308105,
      "learning_rate": 1.4716715787448007e-05,
      "loss": 1.1232,
      "step": 949
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.038907051086426,
      "learning_rate": 1.4705719364398117e-05,
      "loss": 1.1114,
      "step": 950
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.2017502784729004,
      "learning_rate": 1.469471562785891e-05,
      "loss": 1.1527,
      "step": 951
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7876474857330322,
      "learning_rate": 1.4683704594932072e-05,
      "loss": 1.1335,
      "step": 952
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0904629230499268,
      "learning_rate": 1.4672686282730622e-05,
      "loss": 1.1031,
      "step": 953
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.062002420425415,
      "learning_rate": 1.4661660708378893e-05,
      "loss": 1.1496,
      "step": 954
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1169936656951904,
      "learning_rate": 1.4650627889012507e-05,
      "loss": 1.1401,
      "step": 955
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.084095001220703,
      "learning_rate": 1.4639587841778342e-05,
      "loss": 1.1835,
      "step": 956
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9750787019729614,
      "learning_rate": 1.4628540583834511e-05,
      "loss": 1.0634,
      "step": 957
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.370492458343506,
      "learning_rate": 1.4617486132350343e-05,
      "loss": 1.1533,
      "step": 958
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9188987016677856,
      "learning_rate": 1.4606424504506325e-05,
      "loss": 1.1452,
      "step": 959
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7508416175842285,
      "learning_rate": 1.4595355717494118e-05,
      "loss": 1.1481,
      "step": 960
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.252556800842285,
      "learning_rate": 1.45842797885165e-05,
      "loss": 1.1727,
      "step": 961
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.260576009750366,
      "learning_rate": 1.4573196734787357e-05,
      "loss": 1.192,
      "step": 962
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.079566478729248,
      "learning_rate": 1.4562106573531632e-05,
      "loss": 1.135,
      "step": 963
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.192981243133545,
      "learning_rate": 1.455100932198533e-05,
      "loss": 1.1761,
      "step": 964
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1805098056793213,
      "learning_rate": 1.4539904997395468e-05,
      "loss": 1.1312,
      "step": 965
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1327884197235107,
      "learning_rate": 1.4528793617020063e-05,
      "loss": 1.1152,
      "step": 966
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9582301378250122,
      "learning_rate": 1.4517675198128086e-05,
      "loss": 1.1807,
      "step": 967
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7847270965576172,
      "learning_rate": 1.4506549757999456e-05,
      "loss": 1.1833,
      "step": 968
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0416650772094727,
      "learning_rate": 1.4495417313924996e-05,
      "loss": 1.1395,
      "step": 969
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1243045330047607,
      "learning_rate": 1.4484277883206433e-05,
      "loss": 1.1244,
      "step": 970
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9312341213226318,
      "learning_rate": 1.4473131483156326e-05,
      "loss": 1.1844,
      "step": 971
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.235759973526001,
      "learning_rate": 1.4461978131098089e-05,
      "loss": 1.1316,
      "step": 972
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.339448928833008,
      "learning_rate": 1.4450817844365924e-05,
      "loss": 1.0929,
      "step": 973
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.93638014793396,
      "learning_rate": 1.4439650640304822e-05,
      "loss": 1.0793,
      "step": 974
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.058004856109619,
      "learning_rate": 1.4428476536270517e-05,
      "loss": 1.1548,
      "step": 975
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0283775329589844,
      "learning_rate": 1.4417295549629469e-05,
      "loss": 1.2049,
      "step": 976
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1237754821777344,
      "learning_rate": 1.4406107697758838e-05,
      "loss": 1.0935,
      "step": 977
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.4134082794189453,
      "learning_rate": 1.4394912998046451e-05,
      "loss": 1.1818,
      "step": 978
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.290454864501953,
      "learning_rate": 1.4383711467890776e-05,
      "loss": 1.1391,
      "step": 979
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.010493278503418,
      "learning_rate": 1.4372503124700899e-05,
      "loss": 1.0706,
      "step": 980
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0310723781585693,
      "learning_rate": 1.4361287985896495e-05,
      "loss": 1.1452,
      "step": 981
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.154677391052246,
      "learning_rate": 1.43500660689078e-05,
      "loss": 1.0837,
      "step": 982
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8113762140274048,
      "learning_rate": 1.4338837391175582e-05,
      "loss": 1.1967,
      "step": 983
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9489281177520752,
      "learning_rate": 1.4327601970151122e-05,
      "loss": 1.1348,
      "step": 984
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7468442916870117,
      "learning_rate": 1.4316359823296174e-05,
      "loss": 1.1268,
      "step": 985
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1347217559814453,
      "learning_rate": 1.4305110968082953e-05,
      "loss": 1.1347,
      "step": 986
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.179237127304077,
      "learning_rate": 1.4293855421994094e-05,
      "loss": 1.1686,
      "step": 987
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8693894147872925,
      "learning_rate": 1.4282593202522627e-05,
      "loss": 1.1571,
      "step": 988
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.069723129272461,
      "learning_rate": 1.4271324327171969e-05,
      "loss": 1.1776,
      "step": 989
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9631353616714478,
      "learning_rate": 1.4260048813455867e-05,
      "loss": 1.1384,
      "step": 990
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1168227195739746,
      "learning_rate": 1.4248766678898386e-05,
      "loss": 1.1551,
      "step": 991
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1345579624176025,
      "learning_rate": 1.4237477941033888e-05,
      "loss": 1.0779,
      "step": 992
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1054611206054688,
      "learning_rate": 1.4226182617406996e-05,
      "loss": 1.1423,
      "step": 993
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.121643304824829,
      "learning_rate": 1.4214880725572563e-05,
      "loss": 1.096,
      "step": 994
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.127092123031616,
      "learning_rate": 1.4203572283095657e-05,
      "loss": 1.0805,
      "step": 995
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1172807216644287,
      "learning_rate": 1.419225730755152e-05,
      "loss": 1.0541,
      "step": 996
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.014103651046753,
      "learning_rate": 1.4180935816525554e-05,
      "loss": 1.0746,
      "step": 997
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.83677077293396,
      "learning_rate": 1.4169607827613284e-05,
      "loss": 1.117,
      "step": 998
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9537527561187744,
      "learning_rate": 1.415827335842033e-05,
      "loss": 1.2246,
      "step": 999
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1023037433624268,
      "learning_rate": 1.4146932426562391e-05,
      "loss": 1.1628,
      "step": 1000
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.115691661834717,
      "learning_rate": 1.4135585049665207e-05,
      "loss": 1.0918,
      "step": 1001
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9690712690353394,
      "learning_rate": 1.4124231245364534e-05,
      "loss": 1.0992,
      "step": 1002
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0973830223083496,
      "learning_rate": 1.4112871031306118e-05,
      "loss": 1.1091,
      "step": 1003
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9564716815948486,
      "learning_rate": 1.4101504425145662e-05,
      "loss": 1.1108,
      "step": 1004
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0316810607910156,
      "learning_rate": 1.4090131444548814e-05,
      "loss": 1.1108,
      "step": 1005
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2079052925109863,
      "learning_rate": 1.4078752107191118e-05,
      "loss": 1.1162,
      "step": 1006
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0630481243133545,
      "learning_rate": 1.4067366430758004e-05,
      "loss": 1.1688,
      "step": 1007
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.03653883934021,
      "learning_rate": 1.4055974432944753e-05,
      "loss": 1.1534,
      "step": 1008
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1164796352386475,
      "learning_rate": 1.4044576131456466e-05,
      "loss": 1.1451,
      "step": 1009
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2972097396850586,
      "learning_rate": 1.4033171544008053e-05,
      "loss": 1.1668,
      "step": 1010
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1991159915924072,
      "learning_rate": 1.4021760688324175e-05,
      "loss": 1.1641,
      "step": 1011
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1452999114990234,
      "learning_rate": 1.4010343582139252e-05,
      "loss": 1.1412,
      "step": 1012
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9767870903015137,
      "learning_rate": 1.3998920243197408e-05,
      "loss": 1.1228,
      "step": 1013
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.229879140853882,
      "learning_rate": 1.3987490689252463e-05,
      "loss": 1.1608,
      "step": 1014
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0112690925598145,
      "learning_rate": 1.3976054938067885e-05,
      "loss": 1.1075,
      "step": 1015
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1151771545410156,
      "learning_rate": 1.3964613007416777e-05,
      "loss": 1.082,
      "step": 1016
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1948070526123047,
      "learning_rate": 1.3953164915081852e-05,
      "loss": 1.1537,
      "step": 1017
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0088984966278076,
      "learning_rate": 1.3941710678855396e-05,
      "loss": 1.0773,
      "step": 1018
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1882665157318115,
      "learning_rate": 1.3930250316539237e-05,
      "loss": 1.0689,
      "step": 1019
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1300809383392334,
      "learning_rate": 1.3918783845944737e-05,
      "loss": 1.0841,
      "step": 1020
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.386584520339966,
      "learning_rate": 1.3907311284892737e-05,
      "loss": 1.1757,
      "step": 1021
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.271986484527588,
      "learning_rate": 1.389583265121356e-05,
      "loss": 1.1256,
      "step": 1022
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9217442274093628,
      "learning_rate": 1.3884347962746949e-05,
      "loss": 1.1125,
      "step": 1023
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0429067611694336,
      "learning_rate": 1.3872857237342067e-05,
      "loss": 1.1685,
      "step": 1024
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8601691722869873,
      "learning_rate": 1.3861360492857464e-05,
      "loss": 1.1035,
      "step": 1025
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.062239170074463,
      "learning_rate": 1.3849857747161036e-05,
      "loss": 1.1483,
      "step": 1026
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0188026428222656,
      "learning_rate": 1.3838349018130007e-05,
      "loss": 1.1642,
      "step": 1027
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9894661903381348,
      "learning_rate": 1.3826834323650899e-05,
      "loss": 1.1909,
      "step": 1028
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1853253841400146,
      "learning_rate": 1.3815313681619515e-05,
      "loss": 1.2124,
      "step": 1029
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9315125942230225,
      "learning_rate": 1.3803787109940891e-05,
      "loss": 1.1434,
      "step": 1030
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.112725019454956,
      "learning_rate": 1.3792254626529286e-05,
      "loss": 1.1329,
      "step": 1031
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8992259502410889,
      "learning_rate": 1.378071624930814e-05,
      "loss": 1.0857,
      "step": 1032
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.324817657470703,
      "learning_rate": 1.3769171996210053e-05,
      "loss": 1.1786,
      "step": 1033
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.926531434059143,
      "learning_rate": 1.3757621885176768e-05,
      "loss": 1.0898,
      "step": 1034
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2120587825775146,
      "learning_rate": 1.3746065934159123e-05,
      "loss": 1.1823,
      "step": 1035
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0410256385803223,
      "learning_rate": 1.3734504161117028e-05,
      "loss": 1.113,
      "step": 1036
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.174452066421509,
      "learning_rate": 1.3722936584019453e-05,
      "loss": 1.1223,
      "step": 1037
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9703847169876099,
      "learning_rate": 1.371136322084438e-05,
      "loss": 1.1649,
      "step": 1038
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1911723613739014,
      "learning_rate": 1.3699784089578791e-05,
      "loss": 1.1836,
      "step": 1039
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0332579612731934,
      "learning_rate": 1.3688199208218623e-05,
      "loss": 1.1397,
      "step": 1040
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9998358488082886,
      "learning_rate": 1.3676608594768754e-05,
      "loss": 1.1662,
      "step": 1041
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0530648231506348,
      "learning_rate": 1.3665012267242974e-05,
      "loss": 1.1175,
      "step": 1042
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9367700815200806,
      "learning_rate": 1.3653410243663953e-05,
      "loss": 1.1276,
      "step": 1043
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1106598377227783,
      "learning_rate": 1.3641802542063208e-05,
      "loss": 1.1665,
      "step": 1044
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2009687423706055,
      "learning_rate": 1.3630189180481083e-05,
      "loss": 1.1019,
      "step": 1045
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.048473834991455,
      "learning_rate": 1.3618570176966723e-05,
      "loss": 1.0992,
      "step": 1046
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1818771362304688,
      "learning_rate": 1.3606945549578039e-05,
      "loss": 1.0123,
      "step": 1047
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9891952276229858,
      "learning_rate": 1.3595315316381676e-05,
      "loss": 1.1862,
      "step": 1048
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1227941513061523,
      "learning_rate": 1.3583679495453e-05,
      "loss": 1.1475,
      "step": 1049
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9893293380737305,
      "learning_rate": 1.3572038104876066e-05,
      "loss": 1.1923,
      "step": 1050
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0434818267822266,
      "learning_rate": 1.356039116274357e-05,
      "loss": 1.1146,
      "step": 1051
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.023879051208496,
      "learning_rate": 1.3548738687156839e-05,
      "loss": 1.1368,
      "step": 1052
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.347719430923462,
      "learning_rate": 1.3537080696225815e-05,
      "loss": 1.1269,
      "step": 1053
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9522267580032349,
      "learning_rate": 1.3525417208068996e-05,
      "loss": 1.1102,
      "step": 1054
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9791761636734009,
      "learning_rate": 1.3513748240813429e-05,
      "loss": 1.1256,
      "step": 1055
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9621890783309937,
      "learning_rate": 1.3502073812594677e-05,
      "loss": 1.1331,
      "step": 1056
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1891326904296875,
      "learning_rate": 1.3490393941556787e-05,
      "loss": 1.1354,
      "step": 1057
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1180191040039062,
      "learning_rate": 1.3478708645852272e-05,
      "loss": 1.1159,
      "step": 1058
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.056706666946411,
      "learning_rate": 1.3467017943642074e-05,
      "loss": 1.0965,
      "step": 1059
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9511005878448486,
      "learning_rate": 1.3455321853095527e-05,
      "loss": 1.0553,
      "step": 1060
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0664875507354736,
      "learning_rate": 1.3443620392390352e-05,
      "loss": 1.1234,
      "step": 1061
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0491373538970947,
      "learning_rate": 1.3431913579712612e-05,
      "loss": 1.1227,
      "step": 1062
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.160956382751465,
      "learning_rate": 1.342020143325669e-05,
      "loss": 1.1111,
      "step": 1063
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.058295488357544,
      "learning_rate": 1.340848397122525e-05,
      "loss": 1.0911,
      "step": 1064
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1995246410369873,
      "learning_rate": 1.3396761211829229e-05,
      "loss": 1.1191,
      "step": 1065
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1108579635620117,
      "learning_rate": 1.338503317328779e-05,
      "loss": 1.1247,
      "step": 1066
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.124035596847534,
      "learning_rate": 1.3373299873828303e-05,
      "loss": 1.1147,
      "step": 1067
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.109598398208618,
      "learning_rate": 1.336156133168631e-05,
      "loss": 1.1224,
      "step": 1068
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8874613046646118,
      "learning_rate": 1.3349817565105507e-05,
      "loss": 1.1607,
      "step": 1069
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.040125608444214,
      "learning_rate": 1.333806859233771e-05,
      "loss": 1.1422,
      "step": 1070
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0927445888519287,
      "learning_rate": 1.3326314431642821e-05,
      "loss": 1.1777,
      "step": 1071
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.129880428314209,
      "learning_rate": 1.3314555101288808e-05,
      "loss": 1.1068,
      "step": 1072
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.051346778869629,
      "learning_rate": 1.3302790619551673e-05,
      "loss": 1.1588,
      "step": 1073
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.071974992752075,
      "learning_rate": 1.3291021004715423e-05,
      "loss": 1.0812,
      "step": 1074
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9831247329711914,
      "learning_rate": 1.3279246275072046e-05,
      "loss": 1.1598,
      "step": 1075
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.213240385055542,
      "learning_rate": 1.3267466448921475e-05,
      "loss": 1.112,
      "step": 1076
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0195043087005615,
      "learning_rate": 1.3255681544571568e-05,
      "loss": 0.9986,
      "step": 1077
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1000263690948486,
      "learning_rate": 1.3243891580338074e-05,
      "loss": 1.1359,
      "step": 1078
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0627083778381348,
      "learning_rate": 1.3232096574544602e-05,
      "loss": 1.1656,
      "step": 1079
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.05391526222229,
      "learning_rate": 1.3220296545522601e-05,
      "loss": 1.1781,
      "step": 1080
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9792522192001343,
      "learning_rate": 1.320849151161133e-05,
      "loss": 1.0752,
      "step": 1081
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2187604904174805,
      "learning_rate": 1.3196681491157816e-05,
      "loss": 1.1383,
      "step": 1082
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1255452632904053,
      "learning_rate": 1.3184866502516846e-05,
      "loss": 1.0807,
      "step": 1083
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1010947227478027,
      "learning_rate": 1.3173046564050923e-05,
      "loss": 1.1084,
      "step": 1084
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.184443473815918,
      "learning_rate": 1.3161221694130247e-05,
      "loss": 1.0468,
      "step": 1085
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1017353534698486,
      "learning_rate": 1.3149391911132674e-05,
      "loss": 1.0816,
      "step": 1086
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9621886014938354,
      "learning_rate": 1.3137557233443707e-05,
      "loss": 1.1733,
      "step": 1087
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.09228777885437,
      "learning_rate": 1.3125717679456447e-05,
      "loss": 1.211,
      "step": 1088
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9787389039993286,
      "learning_rate": 1.3113873267571577e-05,
      "loss": 1.0744,
      "step": 1089
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1268293857574463,
      "learning_rate": 1.3102024016197337e-05,
      "loss": 1.1154,
      "step": 1090
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.088141918182373,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 1.1358,
      "step": 1091
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9886983633041382,
      "learning_rate": 1.3078311068651241e-05,
      "loss": 1.1475,
      "step": 1092
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.099916458129883,
      "learning_rate": 1.3066447409333345e-05,
      "loss": 1.1868,
      "step": 1093
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.220027446746826,
      "learning_rate": 1.3054578984233935e-05,
      "loss": 1.169,
      "step": 1094
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.1247763633728027,
      "learning_rate": 1.3042705811798565e-05,
      "loss": 1.1751,
      "step": 1095
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.103421688079834,
      "learning_rate": 1.3030827910480167e-05,
      "loss": 1.1475,
      "step": 1096
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.080113410949707,
      "learning_rate": 1.3018945298739022e-05,
      "loss": 1.1667,
      "step": 1097
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.853403091430664,
      "learning_rate": 1.300705799504273e-05,
      "loss": 1.1437,
      "step": 1098
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9002360105514526,
      "learning_rate": 1.2995166017866194e-05,
      "loss": 1.1346,
      "step": 1099
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9794301986694336,
      "learning_rate": 1.2983269385691562e-05,
      "loss": 1.1219,
      "step": 1100
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7105000019073486,
      "learning_rate": 1.2971368117008232e-05,
      "loss": 1.1486,
      "step": 1101
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8645259141921997,
      "learning_rate": 1.29594622303128e-05,
      "loss": 1.1735,
      "step": 1102
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8241422176361084,
      "learning_rate": 1.2947551744109044e-05,
      "loss": 1.1372,
      "step": 1103
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9360711574554443,
      "learning_rate": 1.2935636676907886e-05,
      "loss": 1.1235,
      "step": 1104
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0931246280670166,
      "learning_rate": 1.2923717047227368e-05,
      "loss": 1.1053,
      "step": 1105
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.900255799293518,
      "learning_rate": 1.2911792873592628e-05,
      "loss": 1.0718,
      "step": 1106
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.328380584716797,
      "learning_rate": 1.2899864174535863e-05,
      "loss": 1.1433,
      "step": 1107
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1223771572113037,
      "learning_rate": 1.28879309685963e-05,
      "loss": 1.0638,
      "step": 1108
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0530197620391846,
      "learning_rate": 1.2875993274320173e-05,
      "loss": 1.1273,
      "step": 1109
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.054980516433716,
      "learning_rate": 1.2864051110260697e-05,
      "loss": 1.111,
      "step": 1110
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9360896348953247,
      "learning_rate": 1.2852104494978024e-05,
      "loss": 1.1521,
      "step": 1111
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1880109310150146,
      "learning_rate": 1.284015344703923e-05,
      "loss": 1.0471,
      "step": 1112
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.849260687828064,
      "learning_rate": 1.2828197985018276e-05,
      "loss": 1.0854,
      "step": 1113
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9605604410171509,
      "learning_rate": 1.281623812749599e-05,
      "loss": 1.1646,
      "step": 1114
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.062492609024048,
      "learning_rate": 1.2804273893060028e-05,
      "loss": 1.0823,
      "step": 1115
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0816633701324463,
      "learning_rate": 1.2792305300304848e-05,
      "loss": 1.1013,
      "step": 1116
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9736711978912354,
      "learning_rate": 1.2780332367831678e-05,
      "loss": 1.1932,
      "step": 1117
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0933001041412354,
      "learning_rate": 1.2768355114248493e-05,
      "loss": 1.1438,
      "step": 1118
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0702292919158936,
      "learning_rate": 1.2756373558169992e-05,
      "loss": 1.1694,
      "step": 1119
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9584721326828003,
      "learning_rate": 1.274438771821755e-05,
      "loss": 1.1441,
      "step": 1120
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8502579927444458,
      "learning_rate": 1.2732397613019203e-05,
      "loss": 1.1128,
      "step": 1121
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.051954507827759,
      "learning_rate": 1.2720403261209615e-05,
      "loss": 1.1406,
      "step": 1122
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9614557027816772,
      "learning_rate": 1.2708404681430054e-05,
      "loss": 1.0981,
      "step": 1123
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.884958028793335,
      "learning_rate": 1.2696401892328356e-05,
      "loss": 1.083,
      "step": 1124
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9530218839645386,
      "learning_rate": 1.2684394912558898e-05,
      "loss": 1.1233,
      "step": 1125
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7797476053237915,
      "learning_rate": 1.267238376078257e-05,
      "loss": 1.0432,
      "step": 1126
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2489640712738037,
      "learning_rate": 1.2660368455666752e-05,
      "loss": 1.1415,
      "step": 1127
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0390803813934326,
      "learning_rate": 1.2648349015885272e-05,
      "loss": 1.0823,
      "step": 1128
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1665191650390625,
      "learning_rate": 1.2636325460118388e-05,
      "loss": 1.1945,
      "step": 1129
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8445687294006348,
      "learning_rate": 1.262429780705275e-05,
      "loss": 1.1131,
      "step": 1130
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8968560695648193,
      "learning_rate": 1.2612266075381385e-05,
      "loss": 1.0437,
      "step": 1131
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.21235728263855,
      "learning_rate": 1.2600230283803648e-05,
      "loss": 1.0721,
      "step": 1132
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9546791315078735,
      "learning_rate": 1.2588190451025209e-05,
      "loss": 1.0455,
      "step": 1133
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.070918083190918,
      "learning_rate": 1.2576146595758019e-05,
      "loss": 1.0873,
      "step": 1134
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2329609394073486,
      "learning_rate": 1.2564098736720286e-05,
      "loss": 1.0915,
      "step": 1135
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0559260845184326,
      "learning_rate": 1.2552046892636427e-05,
      "loss": 1.1015,
      "step": 1136
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.078958511352539,
      "learning_rate": 1.2539991082237062e-05,
      "loss": 1.0717,
      "step": 1137
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9968777894973755,
      "learning_rate": 1.2527931324258975e-05,
      "loss": 1.0807,
      "step": 1138
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.06479811668396,
      "learning_rate": 1.2515867637445088e-05,
      "loss": 1.1154,
      "step": 1139
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1567223072052,
      "learning_rate": 1.2503800040544417e-05,
      "loss": 1.1509,
      "step": 1140
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9639226198196411,
      "learning_rate": 1.2491728552312066e-05,
      "loss": 1.158,
      "step": 1141
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.011441946029663,
      "learning_rate": 1.2479653191509182e-05,
      "loss": 1.1787,
      "step": 1142
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2201945781707764,
      "learning_rate": 1.2467573976902936e-05,
      "loss": 1.0903,
      "step": 1143
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8386290073394775,
      "learning_rate": 1.2455490927266485e-05,
      "loss": 1.1123,
      "step": 1144
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.161186695098877,
      "learning_rate": 1.2443404061378941e-05,
      "loss": 1.1329,
      "step": 1145
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1396498680114746,
      "learning_rate": 1.2431313398025356e-05,
      "loss": 1.1417,
      "step": 1146
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.384620428085327,
      "learning_rate": 1.2419218955996677e-05,
      "loss": 1.05,
      "step": 1147
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.197754144668579,
      "learning_rate": 1.2407120754089733e-05,
      "loss": 1.2377,
      "step": 1148
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9910527467727661,
      "learning_rate": 1.239501881110718e-05,
      "loss": 1.1654,
      "step": 1149
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.095726490020752,
      "learning_rate": 1.2382913145857507e-05,
      "loss": 1.1994,
      "step": 1150
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.015230894088745,
      "learning_rate": 1.2370803777154976e-05,
      "loss": 1.1091,
      "step": 1151
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.224205493927002,
      "learning_rate": 1.2358690723819611e-05,
      "loss": 1.0867,
      "step": 1152
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.011887311935425,
      "learning_rate": 1.2346574004677154e-05,
      "loss": 1.0608,
      "step": 1153
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.3180363178253174,
      "learning_rate": 1.2334453638559057e-05,
      "loss": 1.1484,
      "step": 1154
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8666905164718628,
      "learning_rate": 1.2322329644302426e-05,
      "loss": 1.1171,
      "step": 1155
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.930542230606079,
      "learning_rate": 1.2310202040750022e-05,
      "loss": 1.1395,
      "step": 1156
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.207210063934326,
      "learning_rate": 1.2298070846750197e-05,
      "loss": 1.0629,
      "step": 1157
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0602211952209473,
      "learning_rate": 1.2285936081156897e-05,
      "loss": 1.0979,
      "step": 1158
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1004958152770996,
      "learning_rate": 1.2273797762829615e-05,
      "loss": 1.1132,
      "step": 1159
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.940547227859497,
      "learning_rate": 1.2261655910633365e-05,
      "loss": 1.2024,
      "step": 1160
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1186866760253906,
      "learning_rate": 1.2249510543438652e-05,
      "loss": 1.1286,
      "step": 1161
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1833655834198,
      "learning_rate": 1.2237361680121445e-05,
      "loss": 1.1067,
      "step": 1162
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9286000728607178,
      "learning_rate": 1.2225209339563144e-05,
      "loss": 1.092,
      "step": 1163
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9424619674682617,
      "learning_rate": 1.2213053540650566e-05,
      "loss": 1.1172,
      "step": 1164
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.091325521469116,
      "learning_rate": 1.2200894302275878e-05,
      "loss": 1.1359,
      "step": 1165
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.863377571105957,
      "learning_rate": 1.2188731643336615e-05,
      "loss": 1.1353,
      "step": 1166
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.2955470085144043,
      "learning_rate": 1.2176565582735624e-05,
      "loss": 1.176,
      "step": 1167
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1768903732299805,
      "learning_rate": 1.2164396139381029e-05,
      "loss": 1.0885,
      "step": 1168
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.963748574256897,
      "learning_rate": 1.2152223332186222e-05,
      "loss": 1.0903,
      "step": 1169
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0246946811676025,
      "learning_rate": 1.2140047180069814e-05,
      "loss": 1.0538,
      "step": 1170
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.187525510787964,
      "learning_rate": 1.2127867701955622e-05,
      "loss": 1.1178,
      "step": 1171
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1092066764831543,
      "learning_rate": 1.211568491677263e-05,
      "loss": 1.0804,
      "step": 1172
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.083913803100586,
      "learning_rate": 1.210349884345496e-05,
      "loss": 1.179,
      "step": 1173
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.050808906555176,
      "learning_rate": 1.2091309500941841e-05,
      "loss": 1.1137,
      "step": 1174
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.0212974548339844,
      "learning_rate": 1.2079116908177592e-05,
      "loss": 1.1713,
      "step": 1175
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.142390727996826,
      "learning_rate": 1.2066921084111582e-05,
      "loss": 1.1669,
      "step": 1176
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.820248007774353,
      "learning_rate": 1.2054722047698192e-05,
      "loss": 1.1665,
      "step": 1177
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8918629884719849,
      "learning_rate": 1.2042519817896805e-05,
      "loss": 1.0977,
      "step": 1178
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.215043544769287,
      "learning_rate": 1.2030314413671763e-05,
      "loss": 1.1235,
      "step": 1179
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.214589834213257,
      "learning_rate": 1.2018105853992346e-05,
      "loss": 1.146,
      "step": 1180
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.3156394958496094,
      "learning_rate": 1.200589415783273e-05,
      "loss": 1.1248,
      "step": 1181
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.928248405456543,
      "learning_rate": 1.1993679344171973e-05,
      "loss": 1.0599,
      "step": 1182
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9847966432571411,
      "learning_rate": 1.1981461431993978e-05,
      "loss": 1.0867,
      "step": 1183
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9135801792144775,
      "learning_rate": 1.1969240440287457e-05,
      "loss": 1.0867,
      "step": 1184
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8942394256591797,
      "learning_rate": 1.1957016388045917e-05,
      "loss": 1.1251,
      "step": 1185
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2909390926361084,
      "learning_rate": 1.1944789294267614e-05,
      "loss": 1.116,
      "step": 1186
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1683433055877686,
      "learning_rate": 1.1932559177955533e-05,
      "loss": 1.1381,
      "step": 1187
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1342670917510986,
      "learning_rate": 1.1920326058117364e-05,
      "loss": 1.0657,
      "step": 1188
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0786728858947754,
      "learning_rate": 1.190808995376545e-05,
      "loss": 1.0848,
      "step": 1189
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8350193500518799,
      "learning_rate": 1.1895850883916786e-05,
      "loss": 1.1633,
      "step": 1190
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2230052947998047,
      "learning_rate": 1.188360886759297e-05,
      "loss": 1.1457,
      "step": 1191
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1968648433685303,
      "learning_rate": 1.1871363923820182e-05,
      "loss": 1.0571,
      "step": 1192
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0614938735961914,
      "learning_rate": 1.1859116071629148e-05,
      "loss": 1.1675,
      "step": 1193
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.3264434337615967,
      "learning_rate": 1.1846865330055115e-05,
      "loss": 1.1134,
      "step": 1194
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0267574787139893,
      "learning_rate": 1.1834611718137825e-05,
      "loss": 1.1891,
      "step": 1195
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9409390687942505,
      "learning_rate": 1.1822355254921478e-05,
      "loss": 1.1146,
      "step": 1196
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1423439979553223,
      "learning_rate": 1.18100959594547e-05,
      "loss": 1.0912,
      "step": 1197
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.863655686378479,
      "learning_rate": 1.1797833850790527e-05,
      "loss": 1.1544,
      "step": 1198
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1801278591156006,
      "learning_rate": 1.1785568947986368e-05,
      "loss": 1.0963,
      "step": 1199
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2240066528320312,
      "learning_rate": 1.1773301270103968e-05,
      "loss": 1.1348,
      "step": 1200
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9000307321548462,
      "learning_rate": 1.1761030836209384e-05,
      "loss": 1.1681,
      "step": 1201
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.113084316253662,
      "learning_rate": 1.1748757665372965e-05,
      "loss": 1.0888,
      "step": 1202
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0428051948547363,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 1.0777,
      "step": 1203
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1791746616363525,
      "learning_rate": 1.1724203189177228e-05,
      "loss": 1.1106,
      "step": 1204
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1877098083496094,
      "learning_rate": 1.1711921921979754e-05,
      "loss": 1.1783,
      "step": 1205
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1576292514801025,
      "learning_rate": 1.1699637994164053e-05,
      "loss": 1.108,
      "step": 1206
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.726638674736023,
      "learning_rate": 1.1687351424821448e-05,
      "loss": 1.1042,
      "step": 1207
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9788721799850464,
      "learning_rate": 1.1675062233047365e-05,
      "loss": 1.1111,
      "step": 1208
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9948902130126953,
      "learning_rate": 1.1662770437941293e-05,
      "loss": 1.1486,
      "step": 1209
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1016578674316406,
      "learning_rate": 1.1650476058606776e-05,
      "loss": 1.1615,
      "step": 1210
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.026261568069458,
      "learning_rate": 1.1638179114151378e-05,
      "loss": 1.1646,
      "step": 1211
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0555472373962402,
      "learning_rate": 1.162587962368664e-05,
      "loss": 1.0893,
      "step": 1212
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0920801162719727,
      "learning_rate": 1.1613577606328068e-05,
      "loss": 1.086,
      "step": 1213
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0995540618896484,
      "learning_rate": 1.1601273081195087e-05,
      "loss": 1.1265,
      "step": 1214
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.321056604385376,
      "learning_rate": 1.1588966067411033e-05,
      "loss": 1.1423,
      "step": 1215
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.2245306968688965,
      "learning_rate": 1.1576656584103097e-05,
      "loss": 1.1019,
      "step": 1216
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8195040225982666,
      "learning_rate": 1.156434465040231e-05,
      "loss": 1.1305,
      "step": 1217
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0537068843841553,
      "learning_rate": 1.1552030285443516e-05,
      "loss": 1.1072,
      "step": 1218
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0263965129852295,
      "learning_rate": 1.1539713508365336e-05,
      "loss": 1.1747,
      "step": 1219
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9907608032226562,
      "learning_rate": 1.152739433831014e-05,
      "loss": 1.1149,
      "step": 1220
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1033456325531006,
      "learning_rate": 1.1515072794424013e-05,
      "loss": 1.0478,
      "step": 1221
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0900046825408936,
      "learning_rate": 1.1502748895856737e-05,
      "loss": 1.0767,
      "step": 1222
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.3248932361602783,
      "learning_rate": 1.1490422661761744e-05,
      "loss": 1.0912,
      "step": 1223
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0459542274475098,
      "learning_rate": 1.1478094111296109e-05,
      "loss": 1.1579,
      "step": 1224
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.899797797203064,
      "learning_rate": 1.146576326362049e-05,
      "loss": 1.1106,
      "step": 1225
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.881415605545044,
      "learning_rate": 1.1453430137899129e-05,
      "loss": 1.1014,
      "step": 1226
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1118178367614746,
      "learning_rate": 1.1441094753299802e-05,
      "loss": 1.1351,
      "step": 1227
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1176304817199707,
      "learning_rate": 1.1428757128993801e-05,
      "loss": 1.0677,
      "step": 1228
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9699676036834717,
      "learning_rate": 1.1416417284155892e-05,
      "loss": 1.1585,
      "step": 1229
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9920886754989624,
      "learning_rate": 1.1404075237964296e-05,
      "loss": 1.0614,
      "step": 1230
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.021756172180176,
      "learning_rate": 1.1391731009600655e-05,
      "loss": 1.1175,
      "step": 1231
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9650356769561768,
      "learning_rate": 1.1379384618250002e-05,
      "loss": 1.1382,
      "step": 1232
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9744888544082642,
      "learning_rate": 1.1367036083100735e-05,
      "loss": 1.1518,
      "step": 1233
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.092381477355957,
      "learning_rate": 1.1354685423344577e-05,
      "loss": 1.0523,
      "step": 1234
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0164592266082764,
      "learning_rate": 1.1342332658176556e-05,
      "loss": 1.1219,
      "step": 1235
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0943291187286377,
      "learning_rate": 1.1329977806794976e-05,
      "loss": 1.0702,
      "step": 1236
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0104851722717285,
      "learning_rate": 1.1317620888401379e-05,
      "loss": 1.1197,
      "step": 1237
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.089838981628418,
      "learning_rate": 1.130526192220052e-05,
      "loss": 1.1954,
      "step": 1238
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0972421169281006,
      "learning_rate": 1.1292900927400334e-05,
      "loss": 1.1155,
      "step": 1239
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.1477532386779785,
      "learning_rate": 1.1280537923211919e-05,
      "loss": 1.1113,
      "step": 1240
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0457348823547363,
      "learning_rate": 1.1268172928849486e-05,
      "loss": 1.109,
      "step": 1241
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0238020420074463,
      "learning_rate": 1.1255805963530339e-05,
      "loss": 1.0706,
      "step": 1242
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7477166652679443,
      "learning_rate": 1.1243437046474854e-05,
      "loss": 1.146,
      "step": 1243
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.071218252182007,
      "learning_rate": 1.123106619690643e-05,
      "loss": 1.1528,
      "step": 1244
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.233156204223633,
      "learning_rate": 1.1218693434051475e-05,
      "loss": 1.0874,
      "step": 1245
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.17651629447937,
      "learning_rate": 1.120631877713937e-05,
      "loss": 1.0746,
      "step": 1246
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0279769897460938,
      "learning_rate": 1.1193942245402443e-05,
      "loss": 1.0424,
      "step": 1247
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8711894750595093,
      "learning_rate": 1.118156385807593e-05,
      "loss": 1.1371,
      "step": 1248
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9808789491653442,
      "learning_rate": 1.1169183634397948e-05,
      "loss": 1.0586,
      "step": 1249
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0595874786376953,
      "learning_rate": 1.1156801593609479e-05,
      "loss": 1.0414,
      "step": 1250
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.008655071258545,
      "learning_rate": 1.114441775495432e-05,
      "loss": 1.1162,
      "step": 1251
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.290306806564331,
      "learning_rate": 1.113203213767907e-05,
      "loss": 1.1543,
      "step": 1252
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.870664358139038,
      "learning_rate": 1.1119644761033079e-05,
      "loss": 1.1424,
      "step": 1253
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2094833850860596,
      "learning_rate": 1.1107255644268448e-05,
      "loss": 1.0446,
      "step": 1254
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.1483206748962402,
      "learning_rate": 1.1094864806639971e-05,
      "loss": 1.116,
      "step": 1255
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8754299879074097,
      "learning_rate": 1.1082472267405123e-05,
      "loss": 1.148,
      "step": 1256
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.144179344177246,
      "learning_rate": 1.1070078045824014e-05,
      "loss": 1.0951,
      "step": 1257
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2458534240722656,
      "learning_rate": 1.105768216115938e-05,
      "loss": 1.1014,
      "step": 1258
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.127579689025879,
      "learning_rate": 1.1045284632676535e-05,
      "loss": 1.1348,
      "step": 1259
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.1558127403259277,
      "learning_rate": 1.1032885479643351e-05,
      "loss": 1.0892,
      "step": 1260
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.948914885520935,
      "learning_rate": 1.1020484721330227e-05,
      "loss": 1.0686,
      "step": 1261
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9006043672561646,
      "learning_rate": 1.1008082377010045e-05,
      "loss": 1.0633,
      "step": 1262
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0040600299835205,
      "learning_rate": 1.0995678465958168e-05,
      "loss": 1.1007,
      "step": 1263
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.057093620300293,
      "learning_rate": 1.098327300745238e-05,
      "loss": 1.13,
      "step": 1264
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8492480516433716,
      "learning_rate": 1.0970866020772884e-05,
      "loss": 1.0893,
      "step": 1265
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.146859645843506,
      "learning_rate": 1.0958457525202241e-05,
      "loss": 1.1798,
      "step": 1266
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0424344539642334,
      "learning_rate": 1.0946047540025373e-05,
      "loss": 1.0626,
      "step": 1267
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.357059955596924,
      "learning_rate": 1.0933636084529507e-05,
      "loss": 1.0976,
      "step": 1268
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9416987895965576,
      "learning_rate": 1.0921223178004163e-05,
      "loss": 1.0353,
      "step": 1269
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1292002201080322,
      "learning_rate": 1.0908808839741108e-05,
      "loss": 1.1211,
      "step": 1270
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8614038228988647,
      "learning_rate": 1.0896393089034336e-05,
      "loss": 1.0725,
      "step": 1271
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9323099851608276,
      "learning_rate": 1.0883975945180043e-05,
      "loss": 1.0862,
      "step": 1272
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.044222831726074,
      "learning_rate": 1.0871557427476585e-05,
      "loss": 1.1286,
      "step": 1273
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0621914863586426,
      "learning_rate": 1.0859137555224448e-05,
      "loss": 1.1291,
      "step": 1274
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0058326721191406,
      "learning_rate": 1.0846716347726233e-05,
      "loss": 1.1496,
      "step": 1275
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.820318341255188,
      "learning_rate": 1.0834293824286612e-05,
      "loss": 1.0631,
      "step": 1276
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.019606113433838,
      "learning_rate": 1.0821870004212305e-05,
      "loss": 1.1305,
      "step": 1277
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9105875492095947,
      "learning_rate": 1.0809444906812034e-05,
      "loss": 1.0947,
      "step": 1278
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9217480421066284,
      "learning_rate": 1.0797018551396527e-05,
      "loss": 1.0895,
      "step": 1279
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9761483669281006,
      "learning_rate": 1.0784590957278452e-05,
      "loss": 1.017,
      "step": 1280
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8527408838272095,
      "learning_rate": 1.0772162143772407e-05,
      "loss": 1.0623,
      "step": 1281
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8169554471969604,
      "learning_rate": 1.0759732130194887e-05,
      "loss": 1.1167,
      "step": 1282
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.006256341934204,
      "learning_rate": 1.0747300935864245e-05,
      "loss": 1.0756,
      "step": 1283
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1085610389709473,
      "learning_rate": 1.0734868580100677e-05,
      "loss": 1.0859,
      "step": 1284
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.1516268253326416,
      "learning_rate": 1.0722435082226186e-05,
      "loss": 1.17,
      "step": 1285
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.041027069091797,
      "learning_rate": 1.0710000461564535e-05,
      "loss": 1.0965,
      "step": 1286
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.173938274383545,
      "learning_rate": 1.0697564737441254e-05,
      "loss": 1.0875,
      "step": 1287
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0023765563964844,
      "learning_rate": 1.0685127929183567e-05,
      "loss": 1.0533,
      "step": 1288
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8377718925476074,
      "learning_rate": 1.0672690056120398e-05,
      "loss": 1.1484,
      "step": 1289
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0493903160095215,
      "learning_rate": 1.0660251137582316e-05,
      "loss": 1.139,
      "step": 1290
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0895886421203613,
      "learning_rate": 1.0647811192901518e-05,
      "loss": 1.1455,
      "step": 1291
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.074472188949585,
      "learning_rate": 1.0635370241411803e-05,
      "loss": 1.113,
      "step": 1292
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9800500869750977,
      "learning_rate": 1.0622928302448523e-05,
      "loss": 1.146,
      "step": 1293
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.221726179122925,
      "learning_rate": 1.0610485395348571e-05,
      "loss": 1.1226,
      "step": 1294
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.019347667694092,
      "learning_rate": 1.0598041539450344e-05,
      "loss": 1.1309,
      "step": 1295
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8713867664337158,
      "learning_rate": 1.0585596754093712e-05,
      "loss": 1.1828,
      "step": 1296
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.039044141769409,
      "learning_rate": 1.0573151058619994e-05,
      "loss": 1.125,
      "step": 1297
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.3119657039642334,
      "learning_rate": 1.0560704472371919e-05,
      "loss": 1.1559,
      "step": 1298
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.057481527328491,
      "learning_rate": 1.0548257014693602e-05,
      "loss": 1.0511,
      "step": 1299
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8920626640319824,
      "learning_rate": 1.053580870493051e-05,
      "loss": 1.1063,
      "step": 1300
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0831172466278076,
      "learning_rate": 1.0523359562429441e-05,
      "loss": 1.0222,
      "step": 1301
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.116926431655884,
      "learning_rate": 1.0510909606538478e-05,
      "loss": 1.1364,
      "step": 1302
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.009341239929199,
      "learning_rate": 1.0498458856606972e-05,
      "loss": 1.1016,
      "step": 1303
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0504322052001953,
      "learning_rate": 1.0486007331985516e-05,
      "loss": 1.0924,
      "step": 1304
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0273244380950928,
      "learning_rate": 1.0473555052025893e-05,
      "loss": 1.0899,
      "step": 1305
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.093641996383667,
      "learning_rate": 1.0461102036081069e-05,
      "loss": 1.1647,
      "step": 1306
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.074085235595703,
      "learning_rate": 1.044864830350515e-05,
      "loss": 1.0456,
      "step": 1307
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2045114040374756,
      "learning_rate": 1.0436193873653362e-05,
      "loss": 1.1421,
      "step": 1308
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2101428508758545,
      "learning_rate": 1.0423738765882006e-05,
      "loss": 1.1977,
      "step": 1309
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.18107271194458,
      "learning_rate": 1.0411282999548442e-05,
      "loss": 1.0723,
      "step": 1310
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.3213613033294678,
      "learning_rate": 1.039882659401105e-05,
      "loss": 1.1235,
      "step": 1311
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8490278720855713,
      "learning_rate": 1.0386369568629209e-05,
      "loss": 1.1043,
      "step": 1312
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0954411029815674,
      "learning_rate": 1.037391194276326e-05,
      "loss": 1.0323,
      "step": 1313
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7474703788757324,
      "learning_rate": 1.0361453735774467e-05,
      "loss": 1.1051,
      "step": 1314
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.5527727603912354,
      "learning_rate": 1.0348994967025012e-05,
      "loss": 1.1667,
      "step": 1315
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1230320930480957,
      "learning_rate": 1.0336535655877942e-05,
      "loss": 1.0956,
      "step": 1316
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.160593032836914,
      "learning_rate": 1.0324075821697146e-05,
      "loss": 1.1566,
      "step": 1317
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8571885824203491,
      "learning_rate": 1.0311615483847333e-05,
      "loss": 1.0811,
      "step": 1318
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.077572822570801,
      "learning_rate": 1.0299154661693987e-05,
      "loss": 1.12,
      "step": 1319
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.989057183265686,
      "learning_rate": 1.0286693374603347e-05,
      "loss": 1.0953,
      "step": 1320
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.17189884185791,
      "learning_rate": 1.0274231641942378e-05,
      "loss": 1.1186,
      "step": 1321
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9612396955490112,
      "learning_rate": 1.0261769483078734e-05,
      "loss": 1.1286,
      "step": 1322
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.978851556777954,
      "learning_rate": 1.0249306917380731e-05,
      "loss": 1.0885,
      "step": 1323
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8862606287002563,
      "learning_rate": 1.0236843964217317e-05,
      "loss": 1.0726,
      "step": 1324
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9732320308685303,
      "learning_rate": 1.0224380642958052e-05,
      "loss": 1.1238,
      "step": 1325
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8520922660827637,
      "learning_rate": 1.0211916972973048e-05,
      "loss": 1.1254,
      "step": 1326
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.031344175338745,
      "learning_rate": 1.0199452973632982e-05,
      "loss": 1.1379,
      "step": 1327
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.2249510288238525,
      "learning_rate": 1.0186988664309023e-05,
      "loss": 1.1291,
      "step": 1328
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0507466793060303,
      "learning_rate": 1.0174524064372837e-05,
      "loss": 1.1085,
      "step": 1329
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.250545024871826,
      "learning_rate": 1.016205919319653e-05,
      "loss": 1.0855,
      "step": 1330
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9275566339492798,
      "learning_rate": 1.0149594070152638e-05,
      "loss": 1.1263,
      "step": 1331
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.910042405128479,
      "learning_rate": 1.0137128714614082e-05,
      "loss": 1.1276,
      "step": 1332
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.0219812393188477,
      "learning_rate": 1.0124663145954152e-05,
      "loss": 1.1385,
      "step": 1333
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1053659915924072,
      "learning_rate": 1.011219738354646e-05,
      "loss": 1.1486,
      "step": 1334
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7824970483779907,
      "learning_rate": 1.0099731446764927e-05,
      "loss": 1.0953,
      "step": 1335
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9251192808151245,
      "learning_rate": 1.008726535498374e-05,
      "loss": 1.0933,
      "step": 1336
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.118582010269165,
      "learning_rate": 1.007479912757733e-05,
      "loss": 1.1315,
      "step": 1337
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.135776996612549,
      "learning_rate": 1.0062332783920337e-05,
      "loss": 1.1779,
      "step": 1338
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.014690637588501,
      "learning_rate": 1.0049866343387582e-05,
      "loss": 1.0523,
      "step": 1339
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.2314987182617188,
      "learning_rate": 1.0037399825354036e-05,
      "loss": 1.21,
      "step": 1340
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1638410091400146,
      "learning_rate": 1.0024933249194792e-05,
      "loss": 1.0612,
      "step": 1341
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0756895542144775,
      "learning_rate": 1.0012466634285034e-05,
      "loss": 1.1808,
      "step": 1342
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.067077398300171,
      "learning_rate": 1e-05,
      "loss": 1.1112,
      "step": 1343
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0096442699432373,
      "learning_rate": 9.987533365714968e-06,
      "loss": 1.1088,
      "step": 1344
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9529904127120972,
      "learning_rate": 9.97506675080521e-06,
      "loss": 1.0434,
      "step": 1345
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.092536687850952,
      "learning_rate": 9.962600174645968e-06,
      "loss": 1.1491,
      "step": 1346
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.175205945968628,
      "learning_rate": 9.950133656612421e-06,
      "loss": 1.0601,
      "step": 1347
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1728768348693848,
      "learning_rate": 9.937667216079665e-06,
      "loss": 1.0819,
      "step": 1348
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9862674474716187,
      "learning_rate": 9.925200872422671e-06,
      "loss": 1.0997,
      "step": 1349
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1013998985290527,
      "learning_rate": 9.912734645016262e-06,
      "loss": 1.0991,
      "step": 1350
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.743171215057373,
      "learning_rate": 9.900268553235077e-06,
      "loss": 1.0715,
      "step": 1351
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1712660789489746,
      "learning_rate": 9.887802616453543e-06,
      "loss": 1.2004,
      "step": 1352
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.008445978164673,
      "learning_rate": 9.87533685404585e-06,
      "loss": 1.0722,
      "step": 1353
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7337957620620728,
      "learning_rate": 9.862871285385921e-06,
      "loss": 1.1221,
      "step": 1354
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1562325954437256,
      "learning_rate": 9.850405929847367e-06,
      "loss": 1.0292,
      "step": 1355
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.135002374649048,
      "learning_rate": 9.837940806803474e-06,
      "loss": 1.0705,
      "step": 1356
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.303969144821167,
      "learning_rate": 9.825475935627165e-06,
      "loss": 1.0936,
      "step": 1357
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.163717031478882,
      "learning_rate": 9.81301133569098e-06,
      "loss": 1.0822,
      "step": 1358
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1357333660125732,
      "learning_rate": 9.800547026367022e-06,
      "loss": 1.1719,
      "step": 1359
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9401179552078247,
      "learning_rate": 9.788083027026953e-06,
      "loss": 1.0894,
      "step": 1360
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8623323440551758,
      "learning_rate": 9.775619357041952e-06,
      "loss": 1.1117,
      "step": 1361
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.098562717437744,
      "learning_rate": 9.763156035782686e-06,
      "loss": 1.1395,
      "step": 1362
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.926034688949585,
      "learning_rate": 9.750693082619274e-06,
      "loss": 1.1183,
      "step": 1363
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.1446053981781006,
      "learning_rate": 9.738230516921272e-06,
      "loss": 1.1105,
      "step": 1364
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9071414470672607,
      "learning_rate": 9.725768358057625e-06,
      "loss": 1.1021,
      "step": 1365
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.1443896293640137,
      "learning_rate": 9.713306625396658e-06,
      "loss": 1.061,
      "step": 1366
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.265909194946289,
      "learning_rate": 9.700845338306018e-06,
      "loss": 1.0484,
      "step": 1367
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.199605703353882,
      "learning_rate": 9.68838451615267e-06,
      "loss": 1.1389,
      "step": 1368
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.249391794204712,
      "learning_rate": 9.675924178302857e-06,
      "loss": 1.1084,
      "step": 1369
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.163304567337036,
      "learning_rate": 9.663464344122064e-06,
      "loss": 1.0856,
      "step": 1370
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.1496803760528564,
      "learning_rate": 9.651005032974994e-06,
      "loss": 1.1587,
      "step": 1371
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9755088090896606,
      "learning_rate": 9.638546264225538e-06,
      "loss": 1.1264,
      "step": 1372
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0251593589782715,
      "learning_rate": 9.626088057236745e-06,
      "loss": 1.1668,
      "step": 1373
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.111109495162964,
      "learning_rate": 9.613630431370794e-06,
      "loss": 1.0914,
      "step": 1374
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.001190662384033,
      "learning_rate": 9.601173405988955e-06,
      "loss": 1.1459,
      "step": 1375
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9859609603881836,
      "learning_rate": 9.588717000451563e-06,
      "loss": 1.1193,
      "step": 1376
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.099839925765991,
      "learning_rate": 9.576261234117998e-06,
      "loss": 1.1148,
      "step": 1377
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0269525051116943,
      "learning_rate": 9.563806126346643e-06,
      "loss": 1.09,
      "step": 1378
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0289416313171387,
      "learning_rate": 9.551351696494854e-06,
      "loss": 1.1622,
      "step": 1379
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8733547925949097,
      "learning_rate": 9.538897963918936e-06,
      "loss": 1.1139,
      "step": 1380
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2315354347229004,
      "learning_rate": 9.526444947974112e-06,
      "loss": 1.0838,
      "step": 1381
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8606928586959839,
      "learning_rate": 9.51399266801449e-06,
      "loss": 1.1216,
      "step": 1382
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2280423641204834,
      "learning_rate": 9.501541143393028e-06,
      "loss": 1.0938,
      "step": 1383
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.191945791244507,
      "learning_rate": 9.489090393461523e-06,
      "loss": 1.0946,
      "step": 1384
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0256056785583496,
      "learning_rate": 9.476640437570562e-06,
      "loss": 1.1182,
      "step": 1385
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2547764778137207,
      "learning_rate": 9.46419129506949e-06,
      "loss": 1.1281,
      "step": 1386
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8796321153640747,
      "learning_rate": 9.4517429853064e-06,
      "loss": 1.1395,
      "step": 1387
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2095818519592285,
      "learning_rate": 9.439295527628083e-06,
      "loss": 1.1003,
      "step": 1388
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2739408016204834,
      "learning_rate": 9.426848941380007e-06,
      "loss": 1.1144,
      "step": 1389
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.2189135551452637,
      "learning_rate": 9.41440324590629e-06,
      "loss": 1.0707,
      "step": 1390
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.053323984146118,
      "learning_rate": 9.401958460549658e-06,
      "loss": 1.0781,
      "step": 1391
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1338002681732178,
      "learning_rate": 9.38951460465143e-06,
      "loss": 1.1465,
      "step": 1392
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9417824745178223,
      "learning_rate": 9.377071697551479e-06,
      "loss": 1.1832,
      "step": 1393
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9831329584121704,
      "learning_rate": 9.364629758588199e-06,
      "loss": 1.029,
      "step": 1394
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0059049129486084,
      "learning_rate": 9.352188807098482e-06,
      "loss": 1.0844,
      "step": 1395
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.3085122108459473,
      "learning_rate": 9.339748862417686e-06,
      "loss": 0.9733,
      "step": 1396
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.027102470397949,
      "learning_rate": 9.327309943879604e-06,
      "loss": 1.0949,
      "step": 1397
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9626731872558594,
      "learning_rate": 9.314872070816435e-06,
      "loss": 1.1509,
      "step": 1398
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1581201553344727,
      "learning_rate": 9.302435262558748e-06,
      "loss": 1.1064,
      "step": 1399
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9414223432540894,
      "learning_rate": 9.289999538435463e-06,
      "loss": 1.0315,
      "step": 1400
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0606589317321777,
      "learning_rate": 9.277564917773816e-06,
      "loss": 1.1087,
      "step": 1401
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.323589324951172,
      "learning_rate": 9.265131419899325e-06,
      "loss": 1.1716,
      "step": 1402
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9866007566452026,
      "learning_rate": 9.252699064135759e-06,
      "loss": 1.1458,
      "step": 1403
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7380342483520508,
      "learning_rate": 9.240267869805118e-06,
      "loss": 1.0904,
      "step": 1404
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.89060378074646,
      "learning_rate": 9.227837856227594e-06,
      "loss": 1.0586,
      "step": 1405
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.1215169429779053,
      "learning_rate": 9.215409042721553e-06,
      "loss": 1.1584,
      "step": 1406
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.126962184906006,
      "learning_rate": 9.202981448603477e-06,
      "loss": 1.0607,
      "step": 1407
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.138881206512451,
      "learning_rate": 9.190555093187968e-06,
      "loss": 1.0972,
      "step": 1408
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0895638465881348,
      "learning_rate": 9.178129995787698e-06,
      "loss": 1.0589,
      "step": 1409
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0439066886901855,
      "learning_rate": 9.165706175713391e-06,
      "loss": 1.0945,
      "step": 1410
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9255824089050293,
      "learning_rate": 9.153283652273768e-06,
      "loss": 1.1441,
      "step": 1411
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9608616828918457,
      "learning_rate": 9.140862444775554e-06,
      "loss": 1.1185,
      "step": 1412
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.038109064102173,
      "learning_rate": 9.128442572523418e-06,
      "loss": 1.1225,
      "step": 1413
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0757999420166016,
      "learning_rate": 9.11602405481996e-06,
      "loss": 1.1686,
      "step": 1414
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0886220932006836,
      "learning_rate": 9.103606910965666e-06,
      "loss": 1.0741,
      "step": 1415
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.023231267929077,
      "learning_rate": 9.091191160258895e-06,
      "loss": 0.9993,
      "step": 1416
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9976321458816528,
      "learning_rate": 9.078776821995839e-06,
      "loss": 1.0958,
      "step": 1417
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1904337406158447,
      "learning_rate": 9.066363915470494e-06,
      "loss": 1.0509,
      "step": 1418
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9033163785934448,
      "learning_rate": 9.05395245997463e-06,
      "loss": 1.0861,
      "step": 1419
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1159894466400146,
      "learning_rate": 9.04154247479776e-06,
      "loss": 1.184,
      "step": 1420
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.230778455734253,
      "learning_rate": 9.02913397922712e-06,
      "loss": 1.223,
      "step": 1421
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8978996276855469,
      "learning_rate": 9.016726992547621e-06,
      "loss": 1.1263,
      "step": 1422
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.095830202102661,
      "learning_rate": 9.004321534041836e-06,
      "loss": 1.0465,
      "step": 1423
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0191736221313477,
      "learning_rate": 8.991917622989957e-06,
      "loss": 1.1198,
      "step": 1424
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.016019344329834,
      "learning_rate": 8.979515278669776e-06,
      "loss": 1.0932,
      "step": 1425
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.134251356124878,
      "learning_rate": 8.967114520356652e-06,
      "loss": 1.0988,
      "step": 1426
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.093466281890869,
      "learning_rate": 8.954715367323468e-06,
      "loss": 1.0989,
      "step": 1427
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.141160011291504,
      "learning_rate": 8.942317838840625e-06,
      "loss": 1.1106,
      "step": 1428
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8558472394943237,
      "learning_rate": 8.92992195417599e-06,
      "loss": 1.0965,
      "step": 1429
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.003922700881958,
      "learning_rate": 8.917527732594882e-06,
      "loss": 1.1169,
      "step": 1430
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9853335618972778,
      "learning_rate": 8.905135193360032e-06,
      "loss": 1.2146,
      "step": 1431
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.059844493865967,
      "learning_rate": 8.892744355731555e-06,
      "loss": 1.0414,
      "step": 1432
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.1277239322662354,
      "learning_rate": 8.880355238966923e-06,
      "loss": 1.0857,
      "step": 1433
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9769431352615356,
      "learning_rate": 8.867967862320935e-06,
      "loss": 1.1218,
      "step": 1434
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9725464582443237,
      "learning_rate": 8.855582245045682e-06,
      "loss": 1.1177,
      "step": 1435
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.121272325515747,
      "learning_rate": 8.843198406390524e-06,
      "loss": 1.0748,
      "step": 1436
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8484469652175903,
      "learning_rate": 8.830816365602053e-06,
      "loss": 1.1023,
      "step": 1437
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.009852647781372,
      "learning_rate": 8.818436141924072e-06,
      "loss": 1.128,
      "step": 1438
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.163686752319336,
      "learning_rate": 8.806057754597559e-06,
      "loss": 1.1231,
      "step": 1439
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.028637170791626,
      "learning_rate": 8.793681222860631e-06,
      "loss": 1.0151,
      "step": 1440
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.0488662719726562,
      "learning_rate": 8.781306565948528e-06,
      "loss": 1.1557,
      "step": 1441
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.08610463142395,
      "learning_rate": 8.768933803093573e-06,
      "loss": 1.1493,
      "step": 1442
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.168633460998535,
      "learning_rate": 8.756562953525151e-06,
      "loss": 1.0922,
      "step": 1443
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6791796684265137,
      "learning_rate": 8.744194036469665e-06,
      "loss": 1.0962,
      "step": 1444
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1688127517700195,
      "learning_rate": 8.731827071150519e-06,
      "loss": 1.1182,
      "step": 1445
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9698238372802734,
      "learning_rate": 8.719462076788085e-06,
      "loss": 1.131,
      "step": 1446
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0287795066833496,
      "learning_rate": 8.70709907259967e-06,
      "loss": 1.1062,
      "step": 1447
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1382603645324707,
      "learning_rate": 8.694738077799487e-06,
      "loss": 1.0891,
      "step": 1448
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1598997116088867,
      "learning_rate": 8.682379111598626e-06,
      "loss": 1.1562,
      "step": 1449
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0097768306732178,
      "learning_rate": 8.670022193205027e-06,
      "loss": 1.0769,
      "step": 1450
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.217902660369873,
      "learning_rate": 8.657667341823449e-06,
      "loss": 1.0849,
      "step": 1451
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.973050594329834,
      "learning_rate": 8.645314576655428e-06,
      "loss": 1.0517,
      "step": 1452
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1404519081115723,
      "learning_rate": 8.632963916899268e-06,
      "loss": 0.9962,
      "step": 1453
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.149658441543579,
      "learning_rate": 8.62061538175e-06,
      "loss": 1.1473,
      "step": 1454
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.084038496017456,
      "learning_rate": 8.60826899039935e-06,
      "loss": 1.0999,
      "step": 1455
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.2137606143951416,
      "learning_rate": 8.59592476203571e-06,
      "loss": 1.0469,
      "step": 1456
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.092439651489258,
      "learning_rate": 8.583582715844113e-06,
      "loss": 1.1777,
      "step": 1457
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.023463726043701,
      "learning_rate": 8.571242871006202e-06,
      "loss": 1.1459,
      "step": 1458
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.136021375656128,
      "learning_rate": 8.558905246700202e-06,
      "loss": 1.1808,
      "step": 1459
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9554442167282104,
      "learning_rate": 8.546569862100876e-06,
      "loss": 1.1469,
      "step": 1460
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1295862197875977,
      "learning_rate": 8.534236736379515e-06,
      "loss": 1.1303,
      "step": 1461
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0788791179656982,
      "learning_rate": 8.521905888703894e-06,
      "loss": 1.1353,
      "step": 1462
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0439255237579346,
      "learning_rate": 8.509577338238255e-06,
      "loss": 1.1128,
      "step": 1463
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.180420398712158,
      "learning_rate": 8.497251104143263e-06,
      "loss": 1.12,
      "step": 1464
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0794341564178467,
      "learning_rate": 8.484927205575985e-06,
      "loss": 1.0845,
      "step": 1465
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9681614637374878,
      "learning_rate": 8.472605661689863e-06,
      "loss": 1.0949,
      "step": 1466
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.1266703605651855,
      "learning_rate": 8.460286491634664e-06,
      "loss": 1.0248,
      "step": 1467
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0616703033447266,
      "learning_rate": 8.447969714556484e-06,
      "loss": 1.1366,
      "step": 1468
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.978525161743164,
      "learning_rate": 8.43565534959769e-06,
      "loss": 1.103,
      "step": 1469
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9612858295440674,
      "learning_rate": 8.423343415896904e-06,
      "loss": 1.1204,
      "step": 1470
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.976824164390564,
      "learning_rate": 8.411033932588969e-06,
      "loss": 1.0468,
      "step": 1471
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9748729467391968,
      "learning_rate": 8.398726918804911e-06,
      "loss": 1.0332,
      "step": 1472
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.977769136428833,
      "learning_rate": 8.386422393671934e-06,
      "loss": 1.1643,
      "step": 1473
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9440882205963135,
      "learning_rate": 8.374120376313362e-06,
      "loss": 1.0564,
      "step": 1474
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.082899332046509,
      "learning_rate": 8.361820885848623e-06,
      "loss": 1.0524,
      "step": 1475
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.310333251953125,
      "learning_rate": 8.349523941393224e-06,
      "loss": 1.1111,
      "step": 1476
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.020038366317749,
      "learning_rate": 8.337229562058707e-06,
      "loss": 1.0776,
      "step": 1477
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.098550319671631,
      "learning_rate": 8.324937766952638e-06,
      "loss": 1.0846,
      "step": 1478
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2772510051727295,
      "learning_rate": 8.312648575178552e-06,
      "loss": 1.1322,
      "step": 1479
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0358498096466064,
      "learning_rate": 8.300362005835947e-06,
      "loss": 1.1434,
      "step": 1480
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0043091773986816,
      "learning_rate": 8.28807807802025e-06,
      "loss": 1.1858,
      "step": 1481
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.112215995788574,
      "learning_rate": 8.275796810822774e-06,
      "loss": 1.1397,
      "step": 1482
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.70342218875885,
      "learning_rate": 8.263518223330698e-06,
      "loss": 1.1698,
      "step": 1483
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.914313554763794,
      "learning_rate": 8.251242334627039e-06,
      "loss": 1.1149,
      "step": 1484
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0917694568634033,
      "learning_rate": 8.238969163790617e-06,
      "loss": 1.1958,
      "step": 1485
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.2785234451293945,
      "learning_rate": 8.226698729896036e-06,
      "loss": 1.0031,
      "step": 1486
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.118316173553467,
      "learning_rate": 8.214431052013636e-06,
      "loss": 1.144,
      "step": 1487
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8279763460159302,
      "learning_rate": 8.202166149209475e-06,
      "loss": 1.1058,
      "step": 1488
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9190444946289062,
      "learning_rate": 8.189904040545302e-06,
      "loss": 1.1116,
      "step": 1489
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.117196798324585,
      "learning_rate": 8.177644745078525e-06,
      "loss": 1.2082,
      "step": 1490
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.246647834777832,
      "learning_rate": 8.165388281862177e-06,
      "loss": 1.0294,
      "step": 1491
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8178882598876953,
      "learning_rate": 8.153134669944887e-06,
      "loss": 1.1519,
      "step": 1492
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0482490062713623,
      "learning_rate": 8.140883928370855e-06,
      "loss": 1.0691,
      "step": 1493
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9156575202941895,
      "learning_rate": 8.12863607617982e-06,
      "loss": 1.1163,
      "step": 1494
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0756282806396484,
      "learning_rate": 8.116391132407033e-06,
      "loss": 1.1049,
      "step": 1495
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.049535036087036,
      "learning_rate": 8.104149116083216e-06,
      "loss": 1.1242,
      "step": 1496
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9343962669372559,
      "learning_rate": 8.091910046234552e-06,
      "loss": 1.0793,
      "step": 1497
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0816025733947754,
      "learning_rate": 8.079673941882639e-06,
      "loss": 1.1204,
      "step": 1498
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.135411500930786,
      "learning_rate": 8.06744082204447e-06,
      "loss": 1.1128,
      "step": 1499
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.061978816986084,
      "learning_rate": 8.055210705732391e-06,
      "loss": 1.1148,
      "step": 1500
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9971214532852173,
      "learning_rate": 8.042983611954087e-06,
      "loss": 1.0977,
      "step": 1501
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9362305402755737,
      "learning_rate": 8.030759559712545e-06,
      "loss": 1.0507,
      "step": 1502
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0651655197143555,
      "learning_rate": 8.018538568006027e-06,
      "loss": 1.0868,
      "step": 1503
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8527884483337402,
      "learning_rate": 8.00632065582803e-06,
      "loss": 1.1278,
      "step": 1504
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.164207935333252,
      "learning_rate": 7.994105842167274e-06,
      "loss": 1.0956,
      "step": 1505
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.130094528198242,
      "learning_rate": 7.981894146007658e-06,
      "loss": 1.1191,
      "step": 1506
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0135304927825928,
      "learning_rate": 7.96968558632824e-06,
      "loss": 1.1832,
      "step": 1507
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0344009399414062,
      "learning_rate": 7.957480182103198e-06,
      "loss": 1.0448,
      "step": 1508
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.057497978210449,
      "learning_rate": 7.945277952301811e-06,
      "loss": 1.0196,
      "step": 1509
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.103379726409912,
      "learning_rate": 7.93307891588842e-06,
      "loss": 1.1369,
      "step": 1510
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9313547611236572,
      "learning_rate": 7.92088309182241e-06,
      "loss": 1.1678,
      "step": 1511
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0106332302093506,
      "learning_rate": 7.908690499058162e-06,
      "loss": 1.1147,
      "step": 1512
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9276609420776367,
      "learning_rate": 7.896501156545044e-06,
      "loss": 1.0385,
      "step": 1513
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0994088649749756,
      "learning_rate": 7.884315083227373e-06,
      "loss": 1.1046,
      "step": 1514
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9529263973236084,
      "learning_rate": 7.872132298044382e-06,
      "loss": 1.088,
      "step": 1515
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.042224884033203,
      "learning_rate": 7.859952819930188e-06,
      "loss": 1.1535,
      "step": 1516
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.2643330097198486,
      "learning_rate": 7.847776667813782e-06,
      "loss": 1.0663,
      "step": 1517
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9815683364868164,
      "learning_rate": 7.835603860618973e-06,
      "loss": 1.1144,
      "step": 1518
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8868129253387451,
      "learning_rate": 7.823434417264378e-06,
      "loss": 1.1122,
      "step": 1519
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.225114107131958,
      "learning_rate": 7.811268356663387e-06,
      "loss": 1.0428,
      "step": 1520
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9408117532730103,
      "learning_rate": 7.799105697724127e-06,
      "loss": 1.1087,
      "step": 1521
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.006227731704712,
      "learning_rate": 7.786946459349441e-06,
      "loss": 1.0925,
      "step": 1522
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2480838298797607,
      "learning_rate": 7.774790660436857e-06,
      "loss": 1.1098,
      "step": 1523
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.100397825241089,
      "learning_rate": 7.76263831987856e-06,
      "loss": 1.1078,
      "step": 1524
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.414620876312256,
      "learning_rate": 7.750489456561351e-06,
      "loss": 1.074,
      "step": 1525
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9122530221939087,
      "learning_rate": 7.738344089366638e-06,
      "loss": 1.0383,
      "step": 1526
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1369354724884033,
      "learning_rate": 7.726202237170387e-06,
      "loss": 1.1137,
      "step": 1527
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8929128646850586,
      "learning_rate": 7.714063918843106e-06,
      "loss": 1.1205,
      "step": 1528
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1241397857666016,
      "learning_rate": 7.701929153249808e-06,
      "loss": 1.0107,
      "step": 1529
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0766971111297607,
      "learning_rate": 7.689797959249984e-06,
      "loss": 1.1433,
      "step": 1530
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.02679443359375,
      "learning_rate": 7.677670355697577e-06,
      "loss": 1.0972,
      "step": 1531
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8820918798446655,
      "learning_rate": 7.66554636144095e-06,
      "loss": 1.0122,
      "step": 1532
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9866809844970703,
      "learning_rate": 7.653425995322852e-06,
      "loss": 1.0718,
      "step": 1533
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0311107635498047,
      "learning_rate": 7.641309276180394e-06,
      "loss": 1.1882,
      "step": 1534
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.2008237838745117,
      "learning_rate": 7.629196222845027e-06,
      "loss": 1.1283,
      "step": 1535
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0520412921905518,
      "learning_rate": 7.617086854142498e-06,
      "loss": 1.0742,
      "step": 1536
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8520824909210205,
      "learning_rate": 7.6049811888928235e-06,
      "loss": 1.1661,
      "step": 1537
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.3313441276550293,
      "learning_rate": 7.592879245910273e-06,
      "loss": 1.0617,
      "step": 1538
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.07185697555542,
      "learning_rate": 7.580781044003324e-06,
      "loss": 1.0776,
      "step": 1539
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.042532444000244,
      "learning_rate": 7.5686866019746444e-06,
      "loss": 1.0917,
      "step": 1540
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.214764356613159,
      "learning_rate": 7.556595938621058e-06,
      "loss": 1.1706,
      "step": 1541
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.0267531871795654,
      "learning_rate": 7.544509072733515e-06,
      "loss": 1.1035,
      "step": 1542
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1297056674957275,
      "learning_rate": 7.532426023097063e-06,
      "loss": 1.114,
      "step": 1543
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1139683723449707,
      "learning_rate": 7.5203468084908185e-06,
      "loss": 1.1216,
      "step": 1544
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.978652000427246,
      "learning_rate": 7.508271447687936e-06,
      "loss": 1.0389,
      "step": 1545
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1386473178863525,
      "learning_rate": 7.496199959455584e-06,
      "loss": 1.0779,
      "step": 1546
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1456310749053955,
      "learning_rate": 7.484132362554915e-06,
      "loss": 1.0843,
      "step": 1547
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9918345212936401,
      "learning_rate": 7.472068675741024e-06,
      "loss": 1.0977,
      "step": 1548
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.164090394973755,
      "learning_rate": 7.4600089177629384e-06,
      "loss": 1.0386,
      "step": 1549
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3182408809661865,
      "learning_rate": 7.447953107363574e-06,
      "loss": 1.0761,
      "step": 1550
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0967342853546143,
      "learning_rate": 7.435901263279717e-06,
      "loss": 1.1241,
      "step": 1551
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.048201322555542,
      "learning_rate": 7.4238534042419805e-06,
      "loss": 1.1783,
      "step": 1552
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.104344129562378,
      "learning_rate": 7.411809548974792e-06,
      "loss": 1.0974,
      "step": 1553
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.161365032196045,
      "learning_rate": 7.399769716196353e-06,
      "loss": 1.0494,
      "step": 1554
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.990423321723938,
      "learning_rate": 7.387733924618617e-06,
      "loss": 1.056,
      "step": 1555
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.2371435165405273,
      "learning_rate": 7.3757021929472495e-06,
      "loss": 1.0871,
      "step": 1556
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0566372871398926,
      "learning_rate": 7.3636745398816135e-06,
      "loss": 1.0711,
      "step": 1557
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.048182964324951,
      "learning_rate": 7.3516509841147276e-06,
      "loss": 1.135,
      "step": 1558
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0956366062164307,
      "learning_rate": 7.33963154433325e-06,
      "loss": 1.1348,
      "step": 1559
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.026340961456299,
      "learning_rate": 7.327616239217432e-06,
      "loss": 1.0848,
      "step": 1560
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9602152109146118,
      "learning_rate": 7.315605087441107e-06,
      "loss": 1.167,
      "step": 1561
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.089892864227295,
      "learning_rate": 7.303598107671648e-06,
      "loss": 1.1609,
      "step": 1562
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9736313819885254,
      "learning_rate": 7.291595318569951e-06,
      "loss": 1.162,
      "step": 1563
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.950594425201416,
      "learning_rate": 7.279596738790389e-06,
      "loss": 1.0783,
      "step": 1564
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9996474981307983,
      "learning_rate": 7.267602386980801e-06,
      "loss": 1.1177,
      "step": 1565
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9544799327850342,
      "learning_rate": 7.255612281782451e-06,
      "loss": 1.1085,
      "step": 1566
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.871524453163147,
      "learning_rate": 7.243626441830009e-06,
      "loss": 1.1256,
      "step": 1567
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0104820728302,
      "learning_rate": 7.2316448857515076e-06,
      "loss": 1.1461,
      "step": 1568
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8180543184280396,
      "learning_rate": 7.219667632168326e-06,
      "loss": 1.0873,
      "step": 1569
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.083705186843872,
      "learning_rate": 7.207694699695155e-06,
      "loss": 1.1012,
      "step": 1570
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0506584644317627,
      "learning_rate": 7.1957261069399745e-06,
      "loss": 1.0916,
      "step": 1571
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.125929832458496,
      "learning_rate": 7.183761872504011e-06,
      "loss": 1.1776,
      "step": 1572
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9218811988830566,
      "learning_rate": 7.171802014981726e-06,
      "loss": 1.1107,
      "step": 1573
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9667363166809082,
      "learning_rate": 7.159846552960774e-06,
      "loss": 1.1425,
      "step": 1574
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1067311763763428,
      "learning_rate": 7.14789550502198e-06,
      "loss": 1.1266,
      "step": 1575
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9808349609375,
      "learning_rate": 7.135948889739307e-06,
      "loss": 1.0807,
      "step": 1576
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.023986577987671,
      "learning_rate": 7.124006725679828e-06,
      "loss": 1.0713,
      "step": 1577
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9587496519088745,
      "learning_rate": 7.112069031403704e-06,
      "loss": 1.1015,
      "step": 1578
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1089558601379395,
      "learning_rate": 7.100135825464138e-06,
      "loss": 1.0335,
      "step": 1579
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.039599895477295,
      "learning_rate": 7.088207126407374e-06,
      "loss": 1.08,
      "step": 1580
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0800459384918213,
      "learning_rate": 7.076282952772634e-06,
      "loss": 1.1199,
      "step": 1581
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0322885513305664,
      "learning_rate": 7.064363323092117e-06,
      "loss": 1.0873,
      "step": 1582
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0144588947296143,
      "learning_rate": 7.052448255890958e-06,
      "loss": 1.064,
      "step": 1583
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0187156200408936,
      "learning_rate": 7.040537769687204e-06,
      "loss": 1.1454,
      "step": 1584
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.1463465690612793,
      "learning_rate": 7.028631882991771e-06,
      "loss": 1.0949,
      "step": 1585
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.00339412689209,
      "learning_rate": 7.01673061430844e-06,
      "loss": 1.0744,
      "step": 1586
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0913426876068115,
      "learning_rate": 7.004833982133808e-06,
      "loss": 1.0692,
      "step": 1587
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2648189067840576,
      "learning_rate": 6.992942004957271e-06,
      "loss": 1.1653,
      "step": 1588
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0065925121307373,
      "learning_rate": 6.981054701260981e-06,
      "loss": 1.1084,
      "step": 1589
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.041378974914551,
      "learning_rate": 6.969172089519836e-06,
      "loss": 1.1419,
      "step": 1590
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.920621395111084,
      "learning_rate": 6.957294188201438e-06,
      "loss": 1.0781,
      "step": 1591
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.128154993057251,
      "learning_rate": 6.9454210157660694e-06,
      "loss": 1.0723,
      "step": 1592
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9094117879867554,
      "learning_rate": 6.933552590666659e-06,
      "loss": 1.0723,
      "step": 1593
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.094926357269287,
      "learning_rate": 6.921688931348762e-06,
      "loss": 1.0538,
      "step": 1594
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8331307172775269,
      "learning_rate": 6.909830056250527e-06,
      "loss": 1.152,
      "step": 1595
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.047940731048584,
      "learning_rate": 6.8979759838026676e-06,
      "loss": 1.0484,
      "step": 1596
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.17697811126709,
      "learning_rate": 6.886126732428424e-06,
      "loss": 1.1248,
      "step": 1597
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9507261514663696,
      "learning_rate": 6.874282320543557e-06,
      "loss": 1.1071,
      "step": 1598
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9298338890075684,
      "learning_rate": 6.862442766556297e-06,
      "loss": 1.0701,
      "step": 1599
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9668244123458862,
      "learning_rate": 6.850608088867332e-06,
      "loss": 1.0899,
      "step": 1600
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.814441204071045,
      "learning_rate": 6.83877830586976e-06,
      "loss": 1.086,
      "step": 1601
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0865113735198975,
      "learning_rate": 6.826953435949081e-06,
      "loss": 1.1438,
      "step": 1602
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0027682781219482,
      "learning_rate": 6.815133497483157e-06,
      "loss": 1.0402,
      "step": 1603
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1299386024475098,
      "learning_rate": 6.8033185088421874e-06,
      "loss": 1.1415,
      "step": 1604
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2721800804138184,
      "learning_rate": 6.791508488388675e-06,
      "loss": 1.0075,
      "step": 1605
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1821746826171875,
      "learning_rate": 6.779703454477402e-06,
      "loss": 1.0808,
      "step": 1606
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9186224937438965,
      "learning_rate": 6.767903425455402e-06,
      "loss": 1.0468,
      "step": 1607
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.991197109222412,
      "learning_rate": 6.7561084196619306e-06,
      "loss": 1.0433,
      "step": 1608
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8954776525497437,
      "learning_rate": 6.744318455428436e-06,
      "loss": 1.0229,
      "step": 1609
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.120370388031006,
      "learning_rate": 6.732533551078528e-06,
      "loss": 1.0727,
      "step": 1610
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9359171390533447,
      "learning_rate": 6.720753724927957e-06,
      "loss": 1.0891,
      "step": 1611
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.015676975250244,
      "learning_rate": 6.708978995284581e-06,
      "loss": 1.102,
      "step": 1612
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.770275354385376,
      "learning_rate": 6.697209380448333e-06,
      "loss": 1.1267,
      "step": 1613
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.078923463821411,
      "learning_rate": 6.685444898711196e-06,
      "loss": 1.108,
      "step": 1614
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1396501064300537,
      "learning_rate": 6.673685568357182e-06,
      "loss": 1.0591,
      "step": 1615
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.047515869140625,
      "learning_rate": 6.661931407662292e-06,
      "loss": 1.139,
      "step": 1616
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1159801483154297,
      "learning_rate": 6.650182434894496e-06,
      "loss": 1.1311,
      "step": 1617
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.087707996368408,
      "learning_rate": 6.638438668313695e-06,
      "loss": 1.1361,
      "step": 1618
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.2029693126678467,
      "learning_rate": 6.6267001261717015e-06,
      "loss": 1.0606,
      "step": 1619
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.1774814128875732,
      "learning_rate": 6.614966826712212e-06,
      "loss": 1.1299,
      "step": 1620
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9464373588562012,
      "learning_rate": 6.603238788170771e-06,
      "loss": 1.0744,
      "step": 1621
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9495941400527954,
      "learning_rate": 6.59151602877475e-06,
      "loss": 1.1155,
      "step": 1622
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.0874831676483154,
      "learning_rate": 6.579798566743314e-06,
      "loss": 1.0785,
      "step": 1623
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.051748275756836,
      "learning_rate": 6.568086420287389e-06,
      "loss": 1.0498,
      "step": 1624
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.104167938232422,
      "learning_rate": 6.5563796076096484e-06,
      "loss": 1.1728,
      "step": 1625
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9477722644805908,
      "learning_rate": 6.544678146904475e-06,
      "loss": 1.1,
      "step": 1626
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.052161693572998,
      "learning_rate": 6.532982056357928e-06,
      "loss": 1.061,
      "step": 1627
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1822869777679443,
      "learning_rate": 6.521291354147727e-06,
      "loss": 1.0603,
      "step": 1628
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.168095588684082,
      "learning_rate": 6.5096060584432134e-06,
      "loss": 1.102,
      "step": 1629
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.070427656173706,
      "learning_rate": 6.497926187405326e-06,
      "loss": 1.0599,
      "step": 1630
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1799020767211914,
      "learning_rate": 6.486251759186573e-06,
      "loss": 1.0632,
      "step": 1631
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0340375900268555,
      "learning_rate": 6.474582791931006e-06,
      "loss": 1.0315,
      "step": 1632
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9689500331878662,
      "learning_rate": 6.462919303774186e-06,
      "loss": 1.0129,
      "step": 1633
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9503164291381836,
      "learning_rate": 6.45126131284316e-06,
      "loss": 1.1089,
      "step": 1634
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1219639778137207,
      "learning_rate": 6.439608837256432e-06,
      "loss": 1.1458,
      "step": 1635
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1236953735351562,
      "learning_rate": 6.427961895123935e-06,
      "loss": 1.0734,
      "step": 1636
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.028430938720703,
      "learning_rate": 6.4163205045469975e-06,
      "loss": 1.0846,
      "step": 1637
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2715771198272705,
      "learning_rate": 6.404684683618325e-06,
      "loss": 1.0834,
      "step": 1638
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9743119478225708,
      "learning_rate": 6.393054450421963e-06,
      "loss": 1.1146,
      "step": 1639
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2376039028167725,
      "learning_rate": 6.381429823033281e-06,
      "loss": 1.1606,
      "step": 1640
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0413501262664795,
      "learning_rate": 6.36981081951892e-06,
      "loss": 1.0894,
      "step": 1641
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0880486965179443,
      "learning_rate": 6.358197457936796e-06,
      "loss": 1.0549,
      "step": 1642
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.813349723815918,
      "learning_rate": 6.34658975633605e-06,
      "loss": 1.0925,
      "step": 1643
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.972325086593628,
      "learning_rate": 6.334987732757028e-06,
      "loss": 1.1118,
      "step": 1644
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.1375391483306885,
      "learning_rate": 6.32339140523125e-06,
      "loss": 1.1183,
      "step": 1645
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0853328704833984,
      "learning_rate": 6.311800791781381e-06,
      "loss": 1.0702,
      "step": 1646
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.223700523376465,
      "learning_rate": 6.300215910421212e-06,
      "loss": 1.1212,
      "step": 1647
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.174569606781006,
      "learning_rate": 6.288636779155621e-06,
      "loss": 1.0618,
      "step": 1648
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2569713592529297,
      "learning_rate": 6.277063415980549e-06,
      "loss": 1.1847,
      "step": 1649
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2951107025146484,
      "learning_rate": 6.265495838882973e-06,
      "loss": 1.085,
      "step": 1650
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.004568338394165,
      "learning_rate": 6.25393406584088e-06,
      "loss": 1.0876,
      "step": 1651
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1090683937072754,
      "learning_rate": 6.2423781148232345e-06,
      "loss": 1.0736,
      "step": 1652
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9403154850006104,
      "learning_rate": 6.230828003789949e-06,
      "loss": 1.0924,
      "step": 1653
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.799635410308838,
      "learning_rate": 6.219283750691864e-06,
      "loss": 1.1132,
      "step": 1654
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0976476669311523,
      "learning_rate": 6.207745373470717e-06,
      "loss": 1.1276,
      "step": 1655
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.3663783073425293,
      "learning_rate": 6.196212890059111e-06,
      "loss": 1.0907,
      "step": 1656
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0144829750061035,
      "learning_rate": 6.184686318380488e-06,
      "loss": 1.1265,
      "step": 1657
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1192829608917236,
      "learning_rate": 6.173165676349103e-06,
      "loss": 1.0619,
      "step": 1658
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.004643201828003,
      "learning_rate": 6.1616509818699975e-06,
      "loss": 1.0942,
      "step": 1659
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9866899251937866,
      "learning_rate": 6.150142252838967e-06,
      "loss": 1.0969,
      "step": 1660
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.215085506439209,
      "learning_rate": 6.138639507142539e-06,
      "loss": 1.1494,
      "step": 1661
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0401766300201416,
      "learning_rate": 6.127142762657934e-06,
      "loss": 1.0599,
      "step": 1662
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.87773859500885,
      "learning_rate": 6.115652037253054e-06,
      "loss": 1.094,
      "step": 1663
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1740565299987793,
      "learning_rate": 6.104167348786443e-06,
      "loss": 1.1386,
      "step": 1664
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0981645584106445,
      "learning_rate": 6.092688715107265e-06,
      "loss": 1.0836,
      "step": 1665
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0547423362731934,
      "learning_rate": 6.0812161540552675e-06,
      "loss": 1.0733,
      "step": 1666
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9700754880905151,
      "learning_rate": 6.069749683460765e-06,
      "loss": 1.0826,
      "step": 1667
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8821861743927002,
      "learning_rate": 6.058289321144608e-06,
      "loss": 1.1037,
      "step": 1668
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.146040916442871,
      "learning_rate": 6.046835084918152e-06,
      "loss": 1.1489,
      "step": 1669
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0803847312927246,
      "learning_rate": 6.0353869925832255e-06,
      "loss": 1.1299,
      "step": 1670
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0954434871673584,
      "learning_rate": 6.023945061932119e-06,
      "loss": 1.0189,
      "step": 1671
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.273372173309326,
      "learning_rate": 6.0125093107475385e-06,
      "loss": 1.1072,
      "step": 1672
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.2314305305480957,
      "learning_rate": 6.001079756802592e-06,
      "loss": 1.0845,
      "step": 1673
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.014892101287842,
      "learning_rate": 5.98965641786075e-06,
      "loss": 1.0715,
      "step": 1674
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1618101596832275,
      "learning_rate": 5.978239311675826e-06,
      "loss": 1.1601,
      "step": 1675
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.232137680053711,
      "learning_rate": 5.966828455991951e-06,
      "loss": 1.1328,
      "step": 1676
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1128127574920654,
      "learning_rate": 5.955423868543537e-06,
      "loss": 1.1339,
      "step": 1677
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8662935495376587,
      "learning_rate": 5.944025567055251e-06,
      "loss": 1.13,
      "step": 1678
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.058858633041382,
      "learning_rate": 5.932633569242e-06,
      "loss": 1.1085,
      "step": 1679
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.1305861473083496,
      "learning_rate": 5.921247892808886e-06,
      "loss": 1.0801,
      "step": 1680
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.074005365371704,
      "learning_rate": 5.909868555451191e-06,
      "loss": 1.0863,
      "step": 1681
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2353601455688477,
      "learning_rate": 5.898495574854342e-06,
      "loss": 1.0844,
      "step": 1682
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.827561855316162,
      "learning_rate": 5.887128968693887e-06,
      "loss": 1.1071,
      "step": 1683
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0747570991516113,
      "learning_rate": 5.875768754635469e-06,
      "loss": 1.0654,
      "step": 1684
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0640437602996826,
      "learning_rate": 5.864414950334796e-06,
      "loss": 1.0606,
      "step": 1685
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9681077003479004,
      "learning_rate": 5.853067573437612e-06,
      "loss": 1.0902,
      "step": 1686
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9823013544082642,
      "learning_rate": 5.8417266415796745e-06,
      "loss": 1.0776,
      "step": 1687
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8577474355697632,
      "learning_rate": 5.830392172386723e-06,
      "loss": 1.1182,
      "step": 1688
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.035902738571167,
      "learning_rate": 5.819064183474451e-06,
      "loss": 1.1362,
      "step": 1689
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8354504108428955,
      "learning_rate": 5.807742692448483e-06,
      "loss": 1.073,
      "step": 1690
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.187523365020752,
      "learning_rate": 5.796427716904347e-06,
      "loss": 1.0407,
      "step": 1691
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0962114334106445,
      "learning_rate": 5.78511927442744e-06,
      "loss": 1.1261,
      "step": 1692
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.215414047241211,
      "learning_rate": 5.773817382593008e-06,
      "loss": 1.1038,
      "step": 1693
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.19148588180542,
      "learning_rate": 5.7625220589661136e-06,
      "loss": 1.0876,
      "step": 1694
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.27380633354187,
      "learning_rate": 5.751233321101617e-06,
      "loss": 1.1029,
      "step": 1695
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.003251791000366,
      "learning_rate": 5.739951186544141e-06,
      "loss": 1.1343,
      "step": 1696
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9356578588485718,
      "learning_rate": 5.728675672828037e-06,
      "loss": 1.0914,
      "step": 1697
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9251477718353271,
      "learning_rate": 5.717406797477371e-06,
      "loss": 1.0401,
      "step": 1698
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3825197219848633,
      "learning_rate": 5.706144578005908e-06,
      "loss": 1.1183,
      "step": 1699
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0672779083251953,
      "learning_rate": 5.694889031917047e-06,
      "loss": 1.1102,
      "step": 1700
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3228559494018555,
      "learning_rate": 5.683640176703824e-06,
      "loss": 1.0996,
      "step": 1701
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3158223628997803,
      "learning_rate": 5.672398029848881e-06,
      "loss": 1.0558,
      "step": 1702
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9911483526229858,
      "learning_rate": 5.66116260882442e-06,
      "loss": 1.1257,
      "step": 1703
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2464706897735596,
      "learning_rate": 5.6499339310922015e-06,
      "loss": 0.989,
      "step": 1704
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.213170289993286,
      "learning_rate": 5.638712014103507e-06,
      "loss": 1.1117,
      "step": 1705
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.026341438293457,
      "learning_rate": 5.627496875299102e-06,
      "loss": 1.0548,
      "step": 1706
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.029353141784668,
      "learning_rate": 5.616288532109225e-06,
      "loss": 1.1159,
      "step": 1707
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0057480335235596,
      "learning_rate": 5.6050870019535496e-06,
      "loss": 1.1327,
      "step": 1708
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1011106967926025,
      "learning_rate": 5.5938923022411615e-06,
      "loss": 1.0842,
      "step": 1709
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.970698595046997,
      "learning_rate": 5.582704450370533e-06,
      "loss": 1.0713,
      "step": 1710
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.193359613418579,
      "learning_rate": 5.571523463729487e-06,
      "loss": 1.1289,
      "step": 1711
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.913373351097107,
      "learning_rate": 5.560349359695181e-06,
      "loss": 1.1375,
      "step": 1712
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.086493730545044,
      "learning_rate": 5.549182155634076e-06,
      "loss": 1.0997,
      "step": 1713
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.010148286819458,
      "learning_rate": 5.5380218689019125e-06,
      "loss": 1.1264,
      "step": 1714
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.113147020339966,
      "learning_rate": 5.526868516843673e-06,
      "loss": 1.1132,
      "step": 1715
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7228913307189941,
      "learning_rate": 5.51572211679357e-06,
      "loss": 1.0834,
      "step": 1716
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.3988699913024902,
      "learning_rate": 5.504582686075002e-06,
      "loss": 1.0087,
      "step": 1717
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0355377197265625,
      "learning_rate": 5.493450242000546e-06,
      "loss": 1.084,
      "step": 1718
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.1305768489837646,
      "learning_rate": 5.4823248018719184e-06,
      "loss": 1.1452,
      "step": 1719
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7497082948684692,
      "learning_rate": 5.471206382979941e-06,
      "loss": 1.1516,
      "step": 1720
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.0818862915039062,
      "learning_rate": 5.460095002604533e-06,
      "loss": 1.1154,
      "step": 1721
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.3669931888580322,
      "learning_rate": 5.4489906780146716e-06,
      "loss": 1.1046,
      "step": 1722
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9429916143417358,
      "learning_rate": 5.43789342646837e-06,
      "loss": 1.1427,
      "step": 1723
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.196155071258545,
      "learning_rate": 5.4268032652126465e-06,
      "loss": 1.1088,
      "step": 1724
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.096585750579834,
      "learning_rate": 5.415720211483499e-06,
      "loss": 1.1484,
      "step": 1725
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.939610242843628,
      "learning_rate": 5.404644282505882e-06,
      "loss": 1.109,
      "step": 1726
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2942538261413574,
      "learning_rate": 5.393575495493679e-06,
      "loss": 1.1598,
      "step": 1727
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9425272941589355,
      "learning_rate": 5.382513867649663e-06,
      "loss": 1.118,
      "step": 1728
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8539479970932007,
      "learning_rate": 5.37145941616549e-06,
      "loss": 1.0643,
      "step": 1729
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9609973430633545,
      "learning_rate": 5.360412158221661e-06,
      "loss": 1.1485,
      "step": 1730
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9605740308761597,
      "learning_rate": 5.349372110987496e-06,
      "loss": 1.0728,
      "step": 1731
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1405398845672607,
      "learning_rate": 5.338339291621109e-06,
      "loss": 1.0547,
      "step": 1732
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.086383819580078,
      "learning_rate": 5.32731371726938e-06,
      "loss": 1.1074,
      "step": 1733
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1521904468536377,
      "learning_rate": 5.316295405067929e-06,
      "loss": 1.0062,
      "step": 1734
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1046409606933594,
      "learning_rate": 5.305284372141095e-06,
      "loss": 1.1227,
      "step": 1735
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.989786982536316,
      "learning_rate": 5.294280635601889e-06,
      "loss": 1.1158,
      "step": 1736
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9036221504211426,
      "learning_rate": 5.283284212551997e-06,
      "loss": 1.1449,
      "step": 1737
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1115336418151855,
      "learning_rate": 5.2722951200817315e-06,
      "loss": 1.0908,
      "step": 1738
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.228163003921509,
      "learning_rate": 5.2613133752700145e-06,
      "loss": 1.1406,
      "step": 1739
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9933525323867798,
      "learning_rate": 5.250338995184348e-06,
      "loss": 1.0763,
      "step": 1740
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0925755500793457,
      "learning_rate": 5.239371996880786e-06,
      "loss": 1.0446,
      "step": 1741
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1182453632354736,
      "learning_rate": 5.228412397403916e-06,
      "loss": 1.0329,
      "step": 1742
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9731640815734863,
      "learning_rate": 5.217460213786822e-06,
      "loss": 1.126,
      "step": 1743
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.994879961013794,
      "learning_rate": 5.206515463051054e-06,
      "loss": 1.0271,
      "step": 1744
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9355889558792114,
      "learning_rate": 5.195578162206627e-06,
      "loss": 1.0461,
      "step": 1745
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9235438108444214,
      "learning_rate": 5.184648328251965e-06,
      "loss": 1.0561,
      "step": 1746
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7448362112045288,
      "learning_rate": 5.1737259781738934e-06,
      "loss": 1.111,
      "step": 1747
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.104656219482422,
      "learning_rate": 5.1628111289476025e-06,
      "loss": 1.1106,
      "step": 1748
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0473151206970215,
      "learning_rate": 5.151903797536631e-06,
      "loss": 1.0818,
      "step": 1749
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.999972939491272,
      "learning_rate": 5.141004000892824e-06,
      "loss": 1.1391,
      "step": 1750
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.090437889099121,
      "learning_rate": 5.130111755956327e-06,
      "loss": 1.1006,
      "step": 1751
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.168076515197754,
      "learning_rate": 5.119227079655544e-06,
      "loss": 1.1013,
      "step": 1752
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.992275357246399,
      "learning_rate": 5.108349988907111e-06,
      "loss": 1.095,
      "step": 1753
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0318305492401123,
      "learning_rate": 5.097480500615879e-06,
      "loss": 1.096,
      "step": 1754
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.063845157623291,
      "learning_rate": 5.086618631674888e-06,
      "loss": 1.0486,
      "step": 1755
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0832042694091797,
      "learning_rate": 5.075764398965331e-06,
      "loss": 1.0918,
      "step": 1756
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0776469707489014,
      "learning_rate": 5.064917819356532e-06,
      "loss": 1.1477,
      "step": 1757
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0468738079071045,
      "learning_rate": 5.054078909705926e-06,
      "loss": 1.178,
      "step": 1758
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0597331523895264,
      "learning_rate": 5.043247686859024e-06,
      "loss": 1.1249,
      "step": 1759
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.205166816711426,
      "learning_rate": 5.032424167649394e-06,
      "loss": 1.1023,
      "step": 1760
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.040487051010132,
      "learning_rate": 5.021608368898621e-06,
      "loss": 1.0626,
      "step": 1761
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.406477212905884,
      "learning_rate": 5.0108003074163005e-06,
      "loss": 1.0748,
      "step": 1762
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.139605760574341,
      "learning_rate": 5.000000000000003e-06,
      "loss": 1.1251,
      "step": 1763
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9485657215118408,
      "learning_rate": 4.989207463435244e-06,
      "loss": 1.1091,
      "step": 1764
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.2621524333953857,
      "learning_rate": 4.978422714495465e-06,
      "loss": 1.0858,
      "step": 1765
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1738579273223877,
      "learning_rate": 4.967645769942e-06,
      "loss": 1.0308,
      "step": 1766
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1339192390441895,
      "learning_rate": 4.956876646524059e-06,
      "loss": 1.0919,
      "step": 1767
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.005560874938965,
      "learning_rate": 4.946115360978696e-06,
      "loss": 1.0777,
      "step": 1768
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.038527727127075,
      "learning_rate": 4.935361930030774e-06,
      "loss": 1.046,
      "step": 1769
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9045964479446411,
      "learning_rate": 4.924616370392962e-06,
      "loss": 1.1063,
      "step": 1770
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7185484170913696,
      "learning_rate": 4.9138786987656865e-06,
      "loss": 1.0605,
      "step": 1771
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8573479652404785,
      "learning_rate": 4.9031489318371195e-06,
      "loss": 1.0819,
      "step": 1772
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.096778631210327,
      "learning_rate": 4.892427086283147e-06,
      "loss": 1.1713,
      "step": 1773
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9790879487991333,
      "learning_rate": 4.881713178767341e-06,
      "loss": 1.092,
      "step": 1774
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.172245502471924,
      "learning_rate": 4.871007225940939e-06,
      "loss": 1.0713,
      "step": 1775
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9747918844223022,
      "learning_rate": 4.860309244442821e-06,
      "loss": 1.0467,
      "step": 1776
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.1390485763549805,
      "learning_rate": 4.849619250899458e-06,
      "loss": 1.1264,
      "step": 1777
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.025855541229248,
      "learning_rate": 4.838937261924933e-06,
      "loss": 1.0842,
      "step": 1778
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0367910861968994,
      "learning_rate": 4.8282632941208725e-06,
      "loss": 1.1544,
      "step": 1779
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8249406814575195,
      "learning_rate": 4.817597364076441e-06,
      "loss": 1.1008,
      "step": 1780
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8820219039916992,
      "learning_rate": 4.806939488368308e-06,
      "loss": 1.1105,
      "step": 1781
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.98818039894104,
      "learning_rate": 4.796289683560631e-06,
      "loss": 1.1068,
      "step": 1782
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0090770721435547,
      "learning_rate": 4.78564796620502e-06,
      "loss": 1.1042,
      "step": 1783
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.137737274169922,
      "learning_rate": 4.775014352840512e-06,
      "loss": 1.1477,
      "step": 1784
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9740898609161377,
      "learning_rate": 4.764388859993556e-06,
      "loss": 1.1238,
      "step": 1785
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9674060344696045,
      "learning_rate": 4.753771504177976e-06,
      "loss": 1.0549,
      "step": 1786
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0028605461120605,
      "learning_rate": 4.743162301894952e-06,
      "loss": 1.1494,
      "step": 1787
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1349518299102783,
      "learning_rate": 4.732561269632992e-06,
      "loss": 1.0908,
      "step": 1788
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.1477420330047607,
      "learning_rate": 4.7219684238679066e-06,
      "loss": 1.0319,
      "step": 1789
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.2491891384124756,
      "learning_rate": 4.711383781062779e-06,
      "loss": 1.1351,
      "step": 1790
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0702829360961914,
      "learning_rate": 4.700807357667953e-06,
      "loss": 1.0575,
      "step": 1791
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0327067375183105,
      "learning_rate": 4.6902391701209826e-06,
      "loss": 1.1481,
      "step": 1792
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9329017400741577,
      "learning_rate": 4.679679234846636e-06,
      "loss": 1.0514,
      "step": 1793
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.140753984451294,
      "learning_rate": 4.669127568256851e-06,
      "loss": 1.0718,
      "step": 1794
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.142291784286499,
      "learning_rate": 4.658584186750713e-06,
      "loss": 1.0144,
      "step": 1795
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8659164905548096,
      "learning_rate": 4.648049106714434e-06,
      "loss": 1.1063,
      "step": 1796
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.189004898071289,
      "learning_rate": 4.637522344521323e-06,
      "loss": 1.007,
      "step": 1797
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9614797830581665,
      "learning_rate": 4.627003916531761e-06,
      "loss": 1.0969,
      "step": 1798
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.806142807006836,
      "learning_rate": 4.616493839093179e-06,
      "loss": 1.0994,
      "step": 1799
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8966248035430908,
      "learning_rate": 4.60599212854002e-06,
      "loss": 1.1083,
      "step": 1800
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.017648458480835,
      "learning_rate": 4.595498801193736e-06,
      "loss": 1.1469,
      "step": 1801
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.984197974205017,
      "learning_rate": 4.5850138733627435e-06,
      "loss": 1.0539,
      "step": 1802
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9302130937576294,
      "learning_rate": 4.5745373613424075e-06,
      "loss": 1.0819,
      "step": 1803
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0928430557250977,
      "learning_rate": 4.564069281415009e-06,
      "loss": 1.1472,
      "step": 1804
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9319603443145752,
      "learning_rate": 4.5536096498497295e-06,
      "loss": 1.1082,
      "step": 1805
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0319430828094482,
      "learning_rate": 4.543158482902617e-06,
      "loss": 1.0752,
      "step": 1806
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1002352237701416,
      "learning_rate": 4.532715796816565e-06,
      "loss": 1.1031,
      "step": 1807
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.097454786300659,
      "learning_rate": 4.522281607821288e-06,
      "loss": 1.0853,
      "step": 1808
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0000736713409424,
      "learning_rate": 4.511855932133289e-06,
      "loss": 1.0922,
      "step": 1809
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9575142860412598,
      "learning_rate": 4.501438785955845e-06,
      "loss": 1.1015,
      "step": 1810
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9900134801864624,
      "learning_rate": 4.491030185478976e-06,
      "loss": 1.0639,
      "step": 1811
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0368077754974365,
      "learning_rate": 4.480630146879419e-06,
      "loss": 1.1004,
      "step": 1812
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.165717840194702,
      "learning_rate": 4.470238686320606e-06,
      "loss": 1.1441,
      "step": 1813
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9778984785079956,
      "learning_rate": 4.459855819952635e-06,
      "loss": 1.0226,
      "step": 1814
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0772528648376465,
      "learning_rate": 4.449481563912252e-06,
      "loss": 1.0918,
      "step": 1815
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.946523904800415,
      "learning_rate": 4.439115934322816e-06,
      "loss": 1.0628,
      "step": 1816
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1224365234375,
      "learning_rate": 4.428758947294278e-06,
      "loss": 1.0426,
      "step": 1817
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1366336345672607,
      "learning_rate": 4.418410618923163e-06,
      "loss": 1.1085,
      "step": 1818
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7378625869750977,
      "learning_rate": 4.408070965292534e-06,
      "loss": 1.1375,
      "step": 1819
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0879054069519043,
      "learning_rate": 4.397740002471973e-06,
      "loss": 1.0729,
      "step": 1820
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.230639696121216,
      "learning_rate": 4.387417746517557e-06,
      "loss": 1.0564,
      "step": 1821
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0533108711242676,
      "learning_rate": 4.377104213471831e-06,
      "loss": 0.9246,
      "step": 1822
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9193230867385864,
      "learning_rate": 4.3667994193637794e-06,
      "loss": 1.0649,
      "step": 1823
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0609843730926514,
      "learning_rate": 4.356503380208812e-06,
      "loss": 1.1282,
      "step": 1824
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9227945804595947,
      "learning_rate": 4.34621611200872e-06,
      "loss": 1.0165,
      "step": 1825
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.003931999206543,
      "learning_rate": 4.335937630751675e-06,
      "loss": 1.097,
      "step": 1826
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0936031341552734,
      "learning_rate": 4.325667952412184e-06,
      "loss": 1.0767,
      "step": 1827
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1678357124328613,
      "learning_rate": 4.315407092951078e-06,
      "loss": 1.0917,
      "step": 1828
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8280673027038574,
      "learning_rate": 4.305155068315481e-06,
      "loss": 1.1402,
      "step": 1829
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9860889911651611,
      "learning_rate": 4.2949118944387846e-06,
      "loss": 1.0402,
      "step": 1830
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.1179068088531494,
      "learning_rate": 4.284677587240625e-06,
      "loss": 1.0933,
      "step": 1831
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7848842144012451,
      "learning_rate": 4.27445216262686e-06,
      "loss": 1.0917,
      "step": 1832
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9051471948623657,
      "learning_rate": 4.264235636489542e-06,
      "loss": 1.0435,
      "step": 1833
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0317513942718506,
      "learning_rate": 4.254028024706887e-06,
      "loss": 1.1181,
      "step": 1834
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1253409385681152,
      "learning_rate": 4.2438293431432665e-06,
      "loss": 1.1826,
      "step": 1835
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7846280336380005,
      "learning_rate": 4.233639607649168e-06,
      "loss": 1.0847,
      "step": 1836
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.031002998352051,
      "learning_rate": 4.223458834061175e-06,
      "loss": 1.0163,
      "step": 1837
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.925840139389038,
      "learning_rate": 4.213287038201943e-06,
      "loss": 1.0395,
      "step": 1838
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.230156183242798,
      "learning_rate": 4.203124235880179e-06,
      "loss": 1.1085,
      "step": 1839
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0899596214294434,
      "learning_rate": 4.192970442890602e-06,
      "loss": 0.9987,
      "step": 1840
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9482628107070923,
      "learning_rate": 4.182825675013945e-06,
      "loss": 1.0356,
      "step": 1841
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1738624572753906,
      "learning_rate": 4.172689948016896e-06,
      "loss": 1.0023,
      "step": 1842
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.987499475479126,
      "learning_rate": 4.162563277652104e-06,
      "loss": 1.0673,
      "step": 1843
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.032559633255005,
      "learning_rate": 4.152445679658139e-06,
      "loss": 1.1135,
      "step": 1844
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.241908550262451,
      "learning_rate": 4.142337169759472e-06,
      "loss": 1.0507,
      "step": 1845
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.088334560394287,
      "learning_rate": 4.132237763666451e-06,
      "loss": 1.0378,
      "step": 1846
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.4197518825531006,
      "learning_rate": 4.12214747707527e-06,
      "loss": 1.0497,
      "step": 1847
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9269376993179321,
      "learning_rate": 4.112066325667954e-06,
      "loss": 1.0724,
      "step": 1848
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2002551555633545,
      "learning_rate": 4.101994325112332e-06,
      "loss": 1.0088,
      "step": 1849
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.21146297454834,
      "learning_rate": 4.0919314910620025e-06,
      "loss": 1.1158,
      "step": 1850
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.1694178581237793,
      "learning_rate": 4.0818778391563255e-06,
      "loss": 1.0528,
      "step": 1851
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.225661277770996,
      "learning_rate": 4.07183338502039e-06,
      "loss": 1.08,
      "step": 1852
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.081386089324951,
      "learning_rate": 4.061798144264986e-06,
      "loss": 1.0643,
      "step": 1853
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9421201944351196,
      "learning_rate": 4.051772132486589e-06,
      "loss": 1.1059,
      "step": 1854
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.960248351097107,
      "learning_rate": 4.041755365267323e-06,
      "loss": 1.1183,
      "step": 1855
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.055168867111206,
      "learning_rate": 4.0317478581749644e-06,
      "loss": 1.1009,
      "step": 1856
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0702321529388428,
      "learning_rate": 4.021749626762869e-06,
      "loss": 1.1301,
      "step": 1857
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2303805351257324,
      "learning_rate": 4.0117606865699975e-06,
      "loss": 1.072,
      "step": 1858
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1404776573181152,
      "learning_rate": 4.001781053120863e-06,
      "loss": 1.0452,
      "step": 1859
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.25935435295105,
      "learning_rate": 3.991810741925515e-06,
      "loss": 1.1136,
      "step": 1860
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.018381118774414,
      "learning_rate": 3.981849768479516e-06,
      "loss": 1.0477,
      "step": 1861
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.2800936698913574,
      "learning_rate": 3.971898148263915e-06,
      "loss": 1.0735,
      "step": 1862
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1086249351501465,
      "learning_rate": 3.961955896745224e-06,
      "loss": 1.107,
      "step": 1863
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8856889009475708,
      "learning_rate": 3.9520230293754e-06,
      "loss": 1.1029,
      "step": 1864
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.077592134475708,
      "learning_rate": 3.942099561591802e-06,
      "loss": 1.0585,
      "step": 1865
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1065855026245117,
      "learning_rate": 3.9321855088171934e-06,
      "loss": 1.0958,
      "step": 1866
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0198168754577637,
      "learning_rate": 3.922280886459701e-06,
      "loss": 1.1511,
      "step": 1867
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9405418634414673,
      "learning_rate": 3.912385709912794e-06,
      "loss": 1.0246,
      "step": 1868
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9860402345657349,
      "learning_rate": 3.902499994555261e-06,
      "loss": 1.0939,
      "step": 1869
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0301761627197266,
      "learning_rate": 3.892623755751189e-06,
      "loss": 1.0945,
      "step": 1870
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.094388008117676,
      "learning_rate": 3.882757008849936e-06,
      "loss": 1.1229,
      "step": 1871
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1418917179107666,
      "learning_rate": 3.872899769186107e-06,
      "loss": 1.1281,
      "step": 1872
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.045501470565796,
      "learning_rate": 3.8630520520795275e-06,
      "loss": 1.0759,
      "step": 1873
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.013611316680908,
      "learning_rate": 3.853213872835229e-06,
      "loss": 0.9822,
      "step": 1874
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.065305471420288,
      "learning_rate": 3.8433852467434175e-06,
      "loss": 1.1072,
      "step": 1875
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.094135284423828,
      "learning_rate": 3.833566189079453e-06,
      "loss": 1.107,
      "step": 1876
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.221890449523926,
      "learning_rate": 3.823756715103822e-06,
      "loss": 1.0408,
      "step": 1877
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.054564952850342,
      "learning_rate": 3.8139568400621184e-06,
      "loss": 1.0251,
      "step": 1878
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.099581003189087,
      "learning_rate": 3.804166579185018e-06,
      "loss": 1.0972,
      "step": 1879
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9709281921386719,
      "learning_rate": 3.794385947688255e-06,
      "loss": 1.125,
      "step": 1880
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.996031403541565,
      "learning_rate": 3.78461496077259e-06,
      "loss": 1.0963,
      "step": 1881
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1745455265045166,
      "learning_rate": 3.774853633623806e-06,
      "loss": 1.1052,
      "step": 1882
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0069360733032227,
      "learning_rate": 3.7651019814126656e-06,
      "loss": 1.1187,
      "step": 1883
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.109524965286255,
      "learning_rate": 3.755360019294898e-06,
      "loss": 1.0781,
      "step": 1884
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.037043571472168,
      "learning_rate": 3.7456277624111725e-06,
      "loss": 1.1192,
      "step": 1885
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9782909154891968,
      "learning_rate": 3.735905225887072e-06,
      "loss": 1.0713,
      "step": 1886
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9802428483963013,
      "learning_rate": 3.726192424833074e-06,
      "loss": 1.027,
      "step": 1887
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0952064990997314,
      "learning_rate": 3.7164893743445274e-06,
      "loss": 1.0746,
      "step": 1888
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.064514636993408,
      "learning_rate": 3.7067960895016277e-06,
      "loss": 1.0433,
      "step": 1889
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0164601802825928,
      "learning_rate": 3.6971125853693834e-06,
      "loss": 1.0528,
      "step": 1890
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.238691806793213,
      "learning_rate": 3.687438876997612e-06,
      "loss": 1.0541,
      "step": 1891
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9820799827575684,
      "learning_rate": 3.677774979420904e-06,
      "loss": 1.1145,
      "step": 1892
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.2317981719970703,
      "learning_rate": 3.6681209076586035e-06,
      "loss": 1.0047,
      "step": 1893
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9848343133926392,
      "learning_rate": 3.6584766767147796e-06,
      "loss": 1.0765,
      "step": 1894
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9645864963531494,
      "learning_rate": 3.6488423015782128e-06,
      "loss": 1.0826,
      "step": 1895
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.143014907836914,
      "learning_rate": 3.6392177972223596e-06,
      "loss": 1.0205,
      "step": 1896
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.034555673599243,
      "learning_rate": 3.6296031786053455e-06,
      "loss": 1.1038,
      "step": 1897
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.165067195892334,
      "learning_rate": 3.619998460669916e-06,
      "loss": 1.0023,
      "step": 1898
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1658997535705566,
      "learning_rate": 3.610403658343443e-06,
      "loss": 1.0361,
      "step": 1899
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.161109209060669,
      "learning_rate": 3.600818786537883e-06,
      "loss": 1.131,
      "step": 1900
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1736698150634766,
      "learning_rate": 3.591243860149759e-06,
      "loss": 1.1379,
      "step": 1901
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.056490421295166,
      "learning_rate": 3.5816788940601387e-06,
      "loss": 1.0797,
      "step": 1902
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8134005069732666,
      "learning_rate": 3.5721239031346067e-06,
      "loss": 1.0927,
      "step": 1903
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.1097288131713867,
      "learning_rate": 3.5625789022232473e-06,
      "loss": 1.0565,
      "step": 1904
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.0622751712799072,
      "learning_rate": 3.5530439061606202e-06,
      "loss": 1.0304,
      "step": 1905
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.126830816268921,
      "learning_rate": 3.5435189297657267e-06,
      "loss": 1.0957,
      "step": 1906
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.082766056060791,
      "learning_rate": 3.534003987842005e-06,
      "loss": 1.0526,
      "step": 1907
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9150887727737427,
      "learning_rate": 3.5244990951772972e-06,
      "loss": 1.0777,
      "step": 1908
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.976219654083252,
      "learning_rate": 3.5150042665438233e-06,
      "loss": 1.093,
      "step": 1909
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.3244826793670654,
      "learning_rate": 3.505519516698165e-06,
      "loss": 1.0603,
      "step": 1910
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.137023687362671,
      "learning_rate": 3.496044860381238e-06,
      "loss": 1.0955,
      "step": 1911
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0748074054718018,
      "learning_rate": 3.486580312318273e-06,
      "loss": 1.0391,
      "step": 1912
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1463418006896973,
      "learning_rate": 3.4771258872187917e-06,
      "loss": 1.105,
      "step": 1913
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1751866340637207,
      "learning_rate": 3.4676815997765756e-06,
      "loss": 1.1154,
      "step": 1914
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.906115174293518,
      "learning_rate": 3.4582474646696575e-06,
      "loss": 1.0378,
      "step": 1915
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.1036808490753174,
      "learning_rate": 3.448823496560292e-06,
      "loss": 1.0594,
      "step": 1916
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.309847354888916,
      "learning_rate": 3.4394097100949286e-06,
      "loss": 1.0564,
      "step": 1917
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.3234987258911133,
      "learning_rate": 3.4300061199041967e-06,
      "loss": 1.0908,
      "step": 1918
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.045914649963379,
      "learning_rate": 3.4206127406028744e-06,
      "loss": 1.1082,
      "step": 1919
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0051355361938477,
      "learning_rate": 3.411229586789876e-06,
      "loss": 1.1402,
      "step": 1920
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.2013471126556396,
      "learning_rate": 3.401856673048217e-06,
      "loss": 1.0525,
      "step": 1921
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8010168075561523,
      "learning_rate": 3.392494013945009e-06,
      "loss": 1.1254,
      "step": 1922
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0478837490081787,
      "learning_rate": 3.3831416240314085e-06,
      "loss": 1.0685,
      "step": 1923
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.049607753753662,
      "learning_rate": 3.3737995178426276e-06,
      "loss": 1.1755,
      "step": 1924
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8641812801361084,
      "learning_rate": 3.3644677098978894e-06,
      "loss": 1.0952,
      "step": 1925
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0942866802215576,
      "learning_rate": 3.3551462147004132e-06,
      "loss": 1.0486,
      "step": 1926
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.13684344291687,
      "learning_rate": 3.3458350467373914e-06,
      "loss": 1.133,
      "step": 1927
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0116803646087646,
      "learning_rate": 3.3365342204799613e-06,
      "loss": 1.0265,
      "step": 1928
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9069188833236694,
      "learning_rate": 3.3272437503831945e-06,
      "loss": 1.0369,
      "step": 1929
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.118513822555542,
      "learning_rate": 3.3179636508860645e-06,
      "loss": 1.0569,
      "step": 1930
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.444251537322998,
      "learning_rate": 3.308693936411421e-06,
      "loss": 1.1846,
      "step": 1931
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9954161643981934,
      "learning_rate": 3.2994346213659824e-06,
      "loss": 1.0431,
      "step": 1932
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.037930965423584,
      "learning_rate": 3.290185720140301e-06,
      "loss": 1.0999,
      "step": 1933
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.036816358566284,
      "learning_rate": 3.280947247108741e-06,
      "loss": 1.0375,
      "step": 1934
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.263140916824341,
      "learning_rate": 3.2717192166294685e-06,
      "loss": 1.0652,
      "step": 1935
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.041914463043213,
      "learning_rate": 3.262501643044417e-06,
      "loss": 1.1135,
      "step": 1936
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7905452251434326,
      "learning_rate": 3.2532945406792573e-06,
      "loss": 1.0932,
      "step": 1937
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1496798992156982,
      "learning_rate": 3.2440979238433977e-06,
      "loss": 1.0559,
      "step": 1938
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.019996166229248,
      "learning_rate": 3.234911806829948e-06,
      "loss": 1.0642,
      "step": 1939
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7940446138381958,
      "learning_rate": 3.225736203915698e-06,
      "loss": 1.1722,
      "step": 1940
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.876781940460205,
      "learning_rate": 3.216571129361097e-06,
      "loss": 1.0877,
      "step": 1941
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2207353115081787,
      "learning_rate": 3.2074165974102324e-06,
      "loss": 1.0985,
      "step": 1942
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0208237171173096,
      "learning_rate": 3.1982726222908046e-06,
      "loss": 1.0987,
      "step": 1943
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2734079360961914,
      "learning_rate": 3.1891392182141103e-06,
      "loss": 1.1053,
      "step": 1944
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0520012378692627,
      "learning_rate": 3.1800163993750166e-06,
      "loss": 1.0202,
      "step": 1945
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1368093490600586,
      "learning_rate": 3.1709041799519312e-06,
      "loss": 1.1069,
      "step": 1946
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.904400110244751,
      "learning_rate": 3.161802574106799e-06,
      "loss": 1.119,
      "step": 1947
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1012165546417236,
      "learning_rate": 3.152711595985065e-06,
      "loss": 1.0586,
      "step": 1948
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1333816051483154,
      "learning_rate": 3.143631259715658e-06,
      "loss": 1.0945,
      "step": 1949
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.047309637069702,
      "learning_rate": 3.134561579410966e-06,
      "loss": 1.142,
      "step": 1950
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9430721998214722,
      "learning_rate": 3.1255025691668184e-06,
      "loss": 1.1107,
      "step": 1951
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.117534637451172,
      "learning_rate": 3.116454243062459e-06,
      "loss": 1.0795,
      "step": 1952
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.138334035873413,
      "learning_rate": 3.10741661516053e-06,
      "loss": 1.068,
      "step": 1953
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1115400791168213,
      "learning_rate": 3.0983896995070397e-06,
      "loss": 1.0774,
      "step": 1954
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1781904697418213,
      "learning_rate": 3.089373510131354e-06,
      "loss": 1.0294,
      "step": 1955
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0763144493103027,
      "learning_rate": 3.0803680610461674e-06,
      "loss": 1.1213,
      "step": 1956
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.201507091522217,
      "learning_rate": 3.071373366247482e-06,
      "loss": 0.9985,
      "step": 1957
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.018613576889038,
      "learning_rate": 3.0623894397145837e-06,
      "loss": 1.0601,
      "step": 1958
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9831788539886475,
      "learning_rate": 3.0534162954100264e-06,
      "loss": 1.031,
      "step": 1959
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.2358272075653076,
      "learning_rate": 3.044453947279603e-06,
      "loss": 1.0834,
      "step": 1960
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0005476474761963,
      "learning_rate": 3.0355024092523334e-06,
      "loss": 1.0345,
      "step": 1961
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.0345916748046875,
      "learning_rate": 3.0265616952404252e-06,
      "loss": 1.0688,
      "step": 1962
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1344540119171143,
      "learning_rate": 3.017631819139273e-06,
      "loss": 1.1367,
      "step": 1963
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9567629098892212,
      "learning_rate": 3.0087127948274264e-06,
      "loss": 1.1748,
      "step": 1964
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.061760663986206,
      "learning_rate": 2.999804636166567e-06,
      "loss": 1.0844,
      "step": 1965
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9265661239624023,
      "learning_rate": 2.990907357001491e-06,
      "loss": 1.0638,
      "step": 1966
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.028034210205078,
      "learning_rate": 2.9820209711600858e-06,
      "loss": 1.0969,
      "step": 1967
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.113457679748535,
      "learning_rate": 2.9731454924533086e-06,
      "loss": 1.089,
      "step": 1968
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.119246244430542,
      "learning_rate": 2.9642809346751677e-06,
      "loss": 1.0494,
      "step": 1969
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9707144498825073,
      "learning_rate": 2.9554273116026887e-06,
      "loss": 1.1726,
      "step": 1970
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9218149185180664,
      "learning_rate": 2.9465846369959126e-06,
      "loss": 1.0832,
      "step": 1971
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.789323091506958,
      "learning_rate": 2.937752924597864e-06,
      "loss": 1.1099,
      "step": 1972
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0854337215423584,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 1.0856,
      "step": 1973
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1082921028137207,
      "learning_rate": 2.9201224413148254e-06,
      "loss": 1.1148,
      "step": 1974
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8733550310134888,
      "learning_rate": 2.91132369783061e-06,
      "loss": 1.0904,
      "step": 1975
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.077897548675537,
      "learning_rate": 2.902535971356625e-06,
      "loss": 1.1334,
      "step": 1976
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9850120544433594,
      "learning_rate": 2.893759275550494e-06,
      "loss": 1.0494,
      "step": 1977
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9902393817901611,
      "learning_rate": 2.884993624052701e-06,
      "loss": 1.0937,
      "step": 1978
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8868184089660645,
      "learning_rate": 2.876239030486554e-06,
      "loss": 1.1015,
      "step": 1979
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1456332206726074,
      "learning_rate": 2.867495508458186e-06,
      "loss": 1.055,
      "step": 1980
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.138429880142212,
      "learning_rate": 2.8587630715565185e-06,
      "loss": 1.1376,
      "step": 1981
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.863605260848999,
      "learning_rate": 2.850041733353247e-06,
      "loss": 1.1004,
      "step": 1982
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9080727100372314,
      "learning_rate": 2.8413315074028157e-06,
      "loss": 1.0448,
      "step": 1983
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1216044425964355,
      "learning_rate": 2.8326324072423994e-06,
      "loss": 1.1037,
      "step": 1984
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.08176589012146,
      "learning_rate": 2.823944446391881e-06,
      "loss": 1.1189,
      "step": 1985
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9596679210662842,
      "learning_rate": 2.8152676383538346e-06,
      "loss": 1.0709,
      "step": 1986
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.1317408084869385,
      "learning_rate": 2.8066019966134907e-06,
      "loss": 1.1217,
      "step": 1987
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.068991184234619,
      "learning_rate": 2.7979475346387363e-06,
      "loss": 1.1075,
      "step": 1988
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.184727907180786,
      "learning_rate": 2.7893042658800793e-06,
      "loss": 1.0799,
      "step": 1989
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9974550008773804,
      "learning_rate": 2.78067220377063e-06,
      "loss": 1.082,
      "step": 1990
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.928688406944275,
      "learning_rate": 2.7720513617260857e-06,
      "loss": 1.2163,
      "step": 1991
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0356600284576416,
      "learning_rate": 2.763441753144701e-06,
      "loss": 1.0412,
      "step": 1992
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.216644048690796,
      "learning_rate": 2.7548433914072736e-06,
      "loss": 1.1404,
      "step": 1993
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.156062602996826,
      "learning_rate": 2.746256289877126e-06,
      "loss": 1.1345,
      "step": 1994
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.068415641784668,
      "learning_rate": 2.7376804619000706e-06,
      "loss": 1.0519,
      "step": 1995
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0306575298309326,
      "learning_rate": 2.729115920804408e-06,
      "loss": 1.2019,
      "step": 1996
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9250686168670654,
      "learning_rate": 2.720562679900892e-06,
      "loss": 1.0623,
      "step": 1997
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9183446168899536,
      "learning_rate": 2.712020752482717e-06,
      "loss": 1.0416,
      "step": 1998
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1527929306030273,
      "learning_rate": 2.703490151825492e-06,
      "loss": 1.1136,
      "step": 1999
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.243713617324829,
      "learning_rate": 2.694970891187225e-06,
      "loss": 1.0875,
      "step": 2000
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9941197633743286,
      "learning_rate": 2.6864629838082957e-06,
      "loss": 1.1492,
      "step": 2001
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0557737350463867,
      "learning_rate": 2.6779664429114448e-06,
      "loss": 1.0518,
      "step": 2002
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.202550172805786,
      "learning_rate": 2.669481281701739e-06,
      "loss": 1.0756,
      "step": 2003
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9158265590667725,
      "learning_rate": 2.661007513366567e-06,
      "loss": 1.0736,
      "step": 2004
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.377363920211792,
      "learning_rate": 2.652545151075606e-06,
      "loss": 1.1326,
      "step": 2005
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.1276984214782715,
      "learning_rate": 2.6440942079808096e-06,
      "loss": 1.04,
      "step": 2006
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9255988597869873,
      "learning_rate": 2.635654697216382e-06,
      "loss": 1.0118,
      "step": 2007
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9377467632293701,
      "learning_rate": 2.6272266318987606e-06,
      "loss": 1.1352,
      "step": 2008
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.960485816001892,
      "learning_rate": 2.6188100251265947e-06,
      "loss": 1.1428,
      "step": 2009
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9914568662643433,
      "learning_rate": 2.6104048899807243e-06,
      "loss": 1.0461,
      "step": 2010
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9814711809158325,
      "learning_rate": 2.6020112395241627e-06,
      "loss": 1.0235,
      "step": 2011
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9382963180541992,
      "learning_rate": 2.5936290868020685e-06,
      "loss": 1.1076,
      "step": 2012
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9915140867233276,
      "learning_rate": 2.5852584448417327e-06,
      "loss": 1.0304,
      "step": 2013
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.309069871902466,
      "learning_rate": 2.5768993266525653e-06,
      "loss": 1.0985,
      "step": 2014
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.279567003250122,
      "learning_rate": 2.5685517452260566e-06,
      "loss": 1.0413,
      "step": 2015
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.235076904296875,
      "learning_rate": 2.5602157135357697e-06,
      "loss": 0.984,
      "step": 2016
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2019641399383545,
      "learning_rate": 2.55189124453732e-06,
      "loss": 1.016,
      "step": 2017
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0233139991760254,
      "learning_rate": 2.5435783511683444e-06,
      "loss": 1.0575,
      "step": 2018
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.173647403717041,
      "learning_rate": 2.5352770463484986e-06,
      "loss": 1.1137,
      "step": 2019
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.019589900970459,
      "learning_rate": 2.5269873429794235e-06,
      "loss": 1.1263,
      "step": 2020
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.186032772064209,
      "learning_rate": 2.51870925394473e-06,
      "loss": 1.0122,
      "step": 2021
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0525412559509277,
      "learning_rate": 2.5104427921099783e-06,
      "loss": 1.0898,
      "step": 2022
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0405092239379883,
      "learning_rate": 2.502187970322657e-06,
      "loss": 1.1094,
      "step": 2023
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.934238076210022,
      "learning_rate": 2.493944801412167e-06,
      "loss": 1.031,
      "step": 2024
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.2143795490264893,
      "learning_rate": 2.485713298189798e-06,
      "loss": 1.1138,
      "step": 2025
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.998497724533081,
      "learning_rate": 2.4774934734487034e-06,
      "loss": 1.0245,
      "step": 2026
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9704022407531738,
      "learning_rate": 2.469285339963892e-06,
      "loss": 1.0121,
      "step": 2027
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.075895309448242,
      "learning_rate": 2.461088910492202e-06,
      "loss": 1.0046,
      "step": 2028
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.173581600189209,
      "learning_rate": 2.45290419777228e-06,
      "loss": 1.0179,
      "step": 2029
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1675612926483154,
      "learning_rate": 2.444731214524565e-06,
      "loss": 1.0003,
      "step": 2030
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9163613319396973,
      "learning_rate": 2.436569973451264e-06,
      "loss": 1.0594,
      "step": 2031
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1370961666107178,
      "learning_rate": 2.428420487236336e-06,
      "loss": 1.1086,
      "step": 2032
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9650774002075195,
      "learning_rate": 2.420282768545469e-06,
      "loss": 1.1419,
      "step": 2033
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.007429599761963,
      "learning_rate": 2.412156830026069e-06,
      "loss": 1.1326,
      "step": 2034
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.12319016456604,
      "learning_rate": 2.4040426843072206e-06,
      "loss": 1.1055,
      "step": 2035
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0051164627075195,
      "learning_rate": 2.395940343999691e-06,
      "loss": 1.1318,
      "step": 2036
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9078313112258911,
      "learning_rate": 2.387849821695899e-06,
      "loss": 1.0148,
      "step": 2037
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.068796157836914,
      "learning_rate": 2.3797711299698924e-06,
      "loss": 1.1006,
      "step": 2038
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8358103036880493,
      "learning_rate": 2.371704281377335e-06,
      "loss": 1.0968,
      "step": 2039
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1067261695861816,
      "learning_rate": 2.363649288455482e-06,
      "loss": 1.0952,
      "step": 2040
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8204158544540405,
      "learning_rate": 2.3556061637231653e-06,
      "loss": 1.1132,
      "step": 2041
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8129912614822388,
      "learning_rate": 2.347574919680773e-06,
      "loss": 1.0935,
      "step": 2042
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.028385639190674,
      "learning_rate": 2.339555568810221e-06,
      "loss": 1.0915,
      "step": 2043
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.072126626968384,
      "learning_rate": 2.3315481235749484e-06,
      "loss": 1.1466,
      "step": 2044
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.093380928039551,
      "learning_rate": 2.323552596419889e-06,
      "loss": 1.0777,
      "step": 2045
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.093386650085449,
      "learning_rate": 2.315568999771455e-06,
      "loss": 1.1269,
      "step": 2046
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0541529655456543,
      "learning_rate": 2.3075973460375134e-06,
      "loss": 1.0502,
      "step": 2047
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.084561824798584,
      "learning_rate": 2.2996376476073724e-06,
      "loss": 1.0833,
      "step": 2048
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0166513919830322,
      "learning_rate": 2.291689916851758e-06,
      "loss": 1.1238,
      "step": 2049
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0867319107055664,
      "learning_rate": 2.2837541661228024e-06,
      "loss": 1.1328,
      "step": 2050
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.307003974914551,
      "learning_rate": 2.275830407754006e-06,
      "loss": 1.1207,
      "step": 2051
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1785788536071777,
      "learning_rate": 2.267918654060244e-06,
      "loss": 1.0981,
      "step": 2052
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9180868864059448,
      "learning_rate": 2.2600189173377263e-06,
      "loss": 1.0091,
      "step": 2053
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9918795824050903,
      "learning_rate": 2.2521312098639914e-06,
      "loss": 1.0774,
      "step": 2054
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1535873413085938,
      "learning_rate": 2.2442555438978774e-06,
      "loss": 1.0781,
      "step": 2055
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.977783203125,
      "learning_rate": 2.2363919316795135e-06,
      "loss": 1.1053,
      "step": 2056
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1209468841552734,
      "learning_rate": 2.2285403854302912e-06,
      "loss": 1.133,
      "step": 2057
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.961635947227478,
      "learning_rate": 2.2207009173528528e-06,
      "loss": 1.1401,
      "step": 2058
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8661292791366577,
      "learning_rate": 2.2128735396310606e-06,
      "loss": 1.028,
      "step": 2059
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9213451147079468,
      "learning_rate": 2.2050582644299976e-06,
      "loss": 1.0953,
      "step": 2060
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.1833877563476562,
      "learning_rate": 2.1972551038959313e-06,
      "loss": 1.0635,
      "step": 2061
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9852343797683716,
      "learning_rate": 2.189464070156302e-06,
      "loss": 1.1305,
      "step": 2062
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.131807327270508,
      "learning_rate": 2.1816851753197023e-06,
      "loss": 1.0834,
      "step": 2063
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0492334365844727,
      "learning_rate": 2.173918431475861e-06,
      "loss": 1.0952,
      "step": 2064
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.111006736755371,
      "learning_rate": 2.1661638506956208e-06,
      "loss": 1.0504,
      "step": 2065
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9400074481964111,
      "learning_rate": 2.1584214450309194e-06,
      "loss": 1.1257,
      "step": 2066
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1785190105438232,
      "learning_rate": 2.1506912265147772e-06,
      "loss": 1.0732,
      "step": 2067
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.963616132736206,
      "learning_rate": 2.1429732071612653e-06,
      "loss": 1.0958,
      "step": 2068
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0915346145629883,
      "learning_rate": 2.1352673989655026e-06,
      "loss": 1.0702,
      "step": 2069
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8960150480270386,
      "learning_rate": 2.1275738139036274e-06,
      "loss": 1.0729,
      "step": 2070
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.034881114959717,
      "learning_rate": 2.119892463932781e-06,
      "loss": 1.015,
      "step": 2071
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0039772987365723,
      "learning_rate": 2.1122233609910903e-06,
      "loss": 1.0487,
      "step": 2072
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.004584550857544,
      "learning_rate": 2.104566516997647e-06,
      "loss": 1.0299,
      "step": 2073
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.794167160987854,
      "learning_rate": 2.096921943852489e-06,
      "loss": 1.1258,
      "step": 2074
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9886587858200073,
      "learning_rate": 2.08928965343659e-06,
      "loss": 1.0862,
      "step": 2075
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8785330057144165,
      "learning_rate": 2.0816696576118223e-06,
      "loss": 1.0771,
      "step": 2076
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2036328315734863,
      "learning_rate": 2.0740619682209607e-06,
      "loss": 1.1361,
      "step": 2077
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1315627098083496,
      "learning_rate": 2.0664665970876496e-06,
      "loss": 1.0686,
      "step": 2078
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8478988409042358,
      "learning_rate": 2.05888355601639e-06,
      "loss": 1.0398,
      "step": 2079
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9928655624389648,
      "learning_rate": 2.0513128567925177e-06,
      "loss": 1.0298,
      "step": 2080
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1912498474121094,
      "learning_rate": 2.043754511182191e-06,
      "loss": 1.119,
      "step": 2081
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.177396774291992,
      "learning_rate": 2.036208530932363e-06,
      "loss": 1.17,
      "step": 2082
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.109976053237915,
      "learning_rate": 2.0286749277707783e-06,
      "loss": 1.0896,
      "step": 2083
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0971505641937256,
      "learning_rate": 2.021153713405931e-06,
      "loss": 1.0516,
      "step": 2084
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9471511840820312,
      "learning_rate": 2.013644899527074e-06,
      "loss": 1.1539,
      "step": 2085
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0618789196014404,
      "learning_rate": 2.0061484978041813e-06,
      "loss": 1.0951,
      "step": 2086
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0704874992370605,
      "learning_rate": 1.9986645198879385e-06,
      "loss": 1.0844,
      "step": 2087
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7902917861938477,
      "learning_rate": 1.9911929774097216e-06,
      "loss": 1.0593,
      "step": 2088
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.138139247894287,
      "learning_rate": 1.983733881981581e-06,
      "loss": 1.0618,
      "step": 2089
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3603248596191406,
      "learning_rate": 1.9762872451962214e-06,
      "loss": 1.0487,
      "step": 2090
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9984337091445923,
      "learning_rate": 1.9688530786269854e-06,
      "loss": 1.0638,
      "step": 2091
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2030344009399414,
      "learning_rate": 1.961431393827827e-06,
      "loss": 1.1064,
      "step": 2092
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0233442783355713,
      "learning_rate": 1.9540222023333165e-06,
      "loss": 1.0805,
      "step": 2093
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.093528985977173,
      "learning_rate": 1.9466255156585965e-06,
      "loss": 1.0734,
      "step": 2094
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9667054414749146,
      "learning_rate": 1.9392413452993787e-06,
      "loss": 1.119,
      "step": 2095
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1054129600524902,
      "learning_rate": 1.93186970273192e-06,
      "loss": 1.1174,
      "step": 2096
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.90326726436615,
      "learning_rate": 1.9245105994130086e-06,
      "loss": 1.0858,
      "step": 2097
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9260094165802002,
      "learning_rate": 1.9171640467799478e-06,
      "loss": 1.084,
      "step": 2098
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0994699001312256,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 1.0713,
      "step": 2099
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1617395877838135,
      "learning_rate": 1.9025086392230163e-06,
      "loss": 1.0249,
      "step": 2100
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9857853651046753,
      "learning_rate": 1.895199807076148e-06,
      "loss": 1.1048,
      "step": 2101
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.040562629699707,
      "learning_rate": 1.8879035711690897e-06,
      "loss": 1.1512,
      "step": 2102
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.2064340114593506,
      "learning_rate": 1.880619942841435e-06,
      "loss": 0.9988,
      "step": 2103
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.02236270904541,
      "learning_rate": 1.8733489334131827e-06,
      "loss": 1.0436,
      "step": 2104
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8686761856079102,
      "learning_rate": 1.8660905541847208e-06,
      "loss": 1.0696,
      "step": 2105
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.1587555408477783,
      "learning_rate": 1.858844816436809e-06,
      "loss": 1.0991,
      "step": 2106
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.933213710784912,
      "learning_rate": 1.8516117314305526e-06,
      "loss": 1.0656,
      "step": 2107
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0804402828216553,
      "learning_rate": 1.8443913104073984e-06,
      "loss": 1.0658,
      "step": 2108
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.049659252166748,
      "learning_rate": 1.8371835645891134e-06,
      "loss": 1.0285,
      "step": 2109
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0305473804473877,
      "learning_rate": 1.82998850517776e-06,
      "loss": 1.1464,
      "step": 2110
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9408175945281982,
      "learning_rate": 1.8228061433556866e-06,
      "loss": 1.0626,
      "step": 2111
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0285484790802,
      "learning_rate": 1.815636490285506e-06,
      "loss": 1.0545,
      "step": 2112
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.955892562866211,
      "learning_rate": 1.808479557110081e-06,
      "loss": 1.0866,
      "step": 2113
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.114067316055298,
      "learning_rate": 1.801335354952507e-06,
      "loss": 1.0203,
      "step": 2114
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.889113426208496,
      "learning_rate": 1.7942038949160857e-06,
      "loss": 1.1105,
      "step": 2115
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0481467247009277,
      "learning_rate": 1.7870851880843233e-06,
      "loss": 1.1034,
      "step": 2116
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0907087326049805,
      "learning_rate": 1.7799792455209019e-06,
      "loss": 1.0292,
      "step": 2117
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0300443172454834,
      "learning_rate": 1.7728860782696666e-06,
      "loss": 1.054,
      "step": 2118
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.030100107192993,
      "learning_rate": 1.765805697354608e-06,
      "loss": 1.0641,
      "step": 2119
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.202105760574341,
      "learning_rate": 1.7587381137798432e-06,
      "loss": 1.097,
      "step": 2120
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.867055892944336,
      "learning_rate": 1.7516833385296016e-06,
      "loss": 1.1135,
      "step": 2121
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0856893062591553,
      "learning_rate": 1.744641382568204e-06,
      "loss": 1.0763,
      "step": 2122
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1351118087768555,
      "learning_rate": 1.7376122568400533e-06,
      "loss": 1.1318,
      "step": 2123
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9626833200454712,
      "learning_rate": 1.730595972269602e-06,
      "loss": 1.0609,
      "step": 2124
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0367257595062256,
      "learning_rate": 1.7235925397613529e-06,
      "loss": 1.1361,
      "step": 2125
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9432381391525269,
      "learning_rate": 1.716601970199836e-06,
      "loss": 1.0174,
      "step": 2126
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9720298051834106,
      "learning_rate": 1.709624274449584e-06,
      "loss": 1.068,
      "step": 2127
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.092616319656372,
      "learning_rate": 1.7026594633551252e-06,
      "loss": 1.1029,
      "step": 2128
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1237807273864746,
      "learning_rate": 1.6957075477409623e-06,
      "loss": 0.9697,
      "step": 2129
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1068990230560303,
      "learning_rate": 1.688768538411556e-06,
      "loss": 1.1077,
      "step": 2130
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.1293509006500244,
      "learning_rate": 1.6818424461513129e-06,
      "loss": 1.1117,
      "step": 2131
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.091648817062378,
      "learning_rate": 1.674929281724551e-06,
      "loss": 1.1245,
      "step": 2132
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2272872924804688,
      "learning_rate": 1.6680290558755119e-06,
      "loss": 1.0947,
      "step": 2133
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.196308135986328,
      "learning_rate": 1.6611417793283192e-06,
      "loss": 1.0429,
      "step": 2134
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.0779333114624023,
      "learning_rate": 1.6542674627869738e-06,
      "loss": 1.0902,
      "step": 2135
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.830299973487854,
      "learning_rate": 1.6474061169353362e-06,
      "loss": 1.1004,
      "step": 2136
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.215440034866333,
      "learning_rate": 1.640557752437103e-06,
      "loss": 1.1,
      "step": 2137
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.110886812210083,
      "learning_rate": 1.6337223799358025e-06,
      "loss": 1.0573,
      "step": 2138
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.303654909133911,
      "learning_rate": 1.6269000100547682e-06,
      "loss": 1.0655,
      "step": 2139
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.008995532989502,
      "learning_rate": 1.6200906533971206e-06,
      "loss": 1.0336,
      "step": 2140
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.199241876602173,
      "learning_rate": 1.6132943205457607e-06,
      "loss": 1.0747,
      "step": 2141
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2171287536621094,
      "learning_rate": 1.6065110220633495e-06,
      "loss": 1.0871,
      "step": 2142
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.934096097946167,
      "learning_rate": 1.599740768492286e-06,
      "loss": 1.004,
      "step": 2143
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.889215111732483,
      "learning_rate": 1.5929835703546992e-06,
      "loss": 1.1003,
      "step": 2144
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9214625358581543,
      "learning_rate": 1.5862394381524239e-06,
      "loss": 1.0602,
      "step": 2145
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1523196697235107,
      "learning_rate": 1.579508382366992e-06,
      "loss": 1.0343,
      "step": 2146
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.963254451751709,
      "learning_rate": 1.5727904134596084e-06,
      "loss": 1.1166,
      "step": 2147
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.025282382965088,
      "learning_rate": 1.566085541871145e-06,
      "loss": 1.0706,
      "step": 2148
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9535044431686401,
      "learning_rate": 1.5593937780221092e-06,
      "loss": 1.0617,
      "step": 2149
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.06207537651062,
      "learning_rate": 1.5527151323126422e-06,
      "loss": 1.0509,
      "step": 2150
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.024340867996216,
      "learning_rate": 1.5460496151225002e-06,
      "loss": 1.0806,
      "step": 2151
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1910572052001953,
      "learning_rate": 1.5393972368110287e-06,
      "loss": 1.0747,
      "step": 2152
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0068399906158447,
      "learning_rate": 1.5327580077171589e-06,
      "loss": 1.026,
      "step": 2153
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0331673622131348,
      "learning_rate": 1.5261319381593831e-06,
      "loss": 1.1467,
      "step": 2154
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0585806369781494,
      "learning_rate": 1.5195190384357405e-06,
      "loss": 1.1538,
      "step": 2155
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9101479053497314,
      "learning_rate": 1.5129193188238068e-06,
      "loss": 1.1446,
      "step": 2156
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.156921863555908,
      "learning_rate": 1.5063327895806668e-06,
      "loss": 1.0251,
      "step": 2157
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8132952451705933,
      "learning_rate": 1.499759460942909e-06,
      "loss": 1.0109,
      "step": 2158
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0310792922973633,
      "learning_rate": 1.4931993431266056e-06,
      "loss": 1.0578,
      "step": 2159
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.109267473220825,
      "learning_rate": 1.486652446327299e-06,
      "loss": 1.0832,
      "step": 2160
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0667452812194824,
      "learning_rate": 1.48011878071998e-06,
      "loss": 1.079,
      "step": 2161
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0776727199554443,
      "learning_rate": 1.4735983564590784e-06,
      "loss": 1.1049,
      "step": 2162
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.079890251159668,
      "learning_rate": 1.467091183678444e-06,
      "loss": 1.0604,
      "step": 2163
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9230848550796509,
      "learning_rate": 1.460597272491333e-06,
      "loss": 1.1363,
      "step": 2164
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.1073877811431885,
      "learning_rate": 1.4541166329903856e-06,
      "loss": 1.1102,
      "step": 2165
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8848835229873657,
      "learning_rate": 1.4476492752476223e-06,
      "loss": 1.0897,
      "step": 2166
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0256147384643555,
      "learning_rate": 1.4411952093144167e-06,
      "loss": 1.0213,
      "step": 2167
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8747097253799438,
      "learning_rate": 1.4347544452214869e-06,
      "loss": 1.1208,
      "step": 2168
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9030245542526245,
      "learning_rate": 1.4283269929788779e-06,
      "loss": 1.0911,
      "step": 2169
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.028526782989502,
      "learning_rate": 1.4219128625759426e-06,
      "loss": 1.1063,
      "step": 2170
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0991570949554443,
      "learning_rate": 1.4155120639813392e-06,
      "loss": 1.027,
      "step": 2171
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.046793222427368,
      "learning_rate": 1.409124607142992e-06,
      "loss": 1.0682,
      "step": 2172
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.350033760070801,
      "learning_rate": 1.4027505019880972e-06,
      "loss": 1.1146,
      "step": 2173
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1326277256011963,
      "learning_rate": 1.3963897584231023e-06,
      "loss": 1.1538,
      "step": 2174
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9357757568359375,
      "learning_rate": 1.3900423863336842e-06,
      "loss": 1.0146,
      "step": 2175
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.014263153076172,
      "learning_rate": 1.3837083955847418e-06,
      "loss": 0.9848,
      "step": 2176
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.143441677093506,
      "learning_rate": 1.377387796020374e-06,
      "loss": 1.0896,
      "step": 2177
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.163198709487915,
      "learning_rate": 1.3710805974638697e-06,
      "loss": 1.1224,
      "step": 2178
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0704548358917236,
      "learning_rate": 1.364786809717692e-06,
      "loss": 1.0718,
      "step": 2179
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.008725881576538,
      "learning_rate": 1.3585064425634542e-06,
      "loss": 1.0661,
      "step": 2180
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0685529708862305,
      "learning_rate": 1.3522395057619186e-06,
      "loss": 1.0683,
      "step": 2181
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.065948963165283,
      "learning_rate": 1.345986009052972e-06,
      "loss": 1.0897,
      "step": 2182
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0238430500030518,
      "learning_rate": 1.339745962155613e-06,
      "loss": 1.109,
      "step": 2183
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1469929218292236,
      "learning_rate": 1.3335193747679387e-06,
      "loss": 1.0662,
      "step": 2184
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.188908338546753,
      "learning_rate": 1.3273062565671258e-06,
      "loss": 1.128,
      "step": 2185
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1393096446990967,
      "learning_rate": 1.3211066172094178e-06,
      "loss": 1.1718,
      "step": 2186
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1209185123443604,
      "learning_rate": 1.3149204663301118e-06,
      "loss": 1.1029,
      "step": 2187
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1069631576538086,
      "learning_rate": 1.3087478135435361e-06,
      "loss": 1.0957,
      "step": 2188
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.260504722595215,
      "learning_rate": 1.3025886684430467e-06,
      "loss": 1.1043,
      "step": 2189
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9879469871520996,
      "learning_rate": 1.2964430406010032e-06,
      "loss": 1.0567,
      "step": 2190
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9505807161331177,
      "learning_rate": 1.2903109395687597e-06,
      "loss": 1.121,
      "step": 2191
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8520638942718506,
      "learning_rate": 1.2841923748766427e-06,
      "loss": 1.0975,
      "step": 2192
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.036651611328125,
      "learning_rate": 1.278087356033947e-06,
      "loss": 1.1049,
      "step": 2193
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.879692792892456,
      "learning_rate": 1.271995892528909e-06,
      "loss": 1.0059,
      "step": 2194
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1137635707855225,
      "learning_rate": 1.2659179938287035e-06,
      "loss": 1.1104,
      "step": 2195
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.0090227127075195,
      "learning_rate": 1.2598536693794171e-06,
      "loss": 1.0931,
      "step": 2196
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9011852741241455,
      "learning_rate": 1.2538029286060428e-06,
      "loss": 1.0371,
      "step": 2197
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.138629198074341,
      "learning_rate": 1.2477657809124632e-06,
      "loss": 1.064,
      "step": 2198
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0290091037750244,
      "learning_rate": 1.2417422356814345e-06,
      "loss": 1.0773,
      "step": 2199
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0091402530670166,
      "learning_rate": 1.2357323022745716e-06,
      "loss": 1.0574,
      "step": 2200
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0669007301330566,
      "learning_rate": 1.2297359900323346e-06,
      "loss": 1.0955,
      "step": 2201
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9753942489624023,
      "learning_rate": 1.2237533082740127e-06,
      "loss": 1.083,
      "step": 2202
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.345106840133667,
      "learning_rate": 1.2177842662977136e-06,
      "loss": 1.0721,
      "step": 2203
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.055177688598633,
      "learning_rate": 1.2118288733803474e-06,
      "loss": 1.0519,
      "step": 2204
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1050448417663574,
      "learning_rate": 1.2058871387776039e-06,
      "loss": 1.0794,
      "step": 2205
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.012935161590576,
      "learning_rate": 1.199959071723953e-06,
      "loss": 1.0371,
      "step": 2206
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.243584394454956,
      "learning_rate": 1.19404468143262e-06,
      "loss": 1.0651,
      "step": 2207
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.146897554397583,
      "learning_rate": 1.188143977095576e-06,
      "loss": 1.1465,
      "step": 2208
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1145238876342773,
      "learning_rate": 1.1822569678835195e-06,
      "loss": 1.0194,
      "step": 2209
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9626801013946533,
      "learning_rate": 1.176383662945866e-06,
      "loss": 1.0247,
      "step": 2210
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.1420493125915527,
      "learning_rate": 1.1705240714107301e-06,
      "loss": 1.1021,
      "step": 2211
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.034756660461426,
      "learning_rate": 1.164678202384919e-06,
      "loss": 1.025,
      "step": 2212
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.162412166595459,
      "learning_rate": 1.1588460649539036e-06,
      "loss": 1.1265,
      "step": 2213
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.946412205696106,
      "learning_rate": 1.153027668181821e-06,
      "loss": 1.0446,
      "step": 2214
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.939348816871643,
      "learning_rate": 1.1472230211114498e-06,
      "loss": 1.092,
      "step": 2215
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.000009298324585,
      "learning_rate": 1.1414321327642019e-06,
      "loss": 1.0393,
      "step": 2216
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0100507736206055,
      "learning_rate": 1.1356550121401033e-06,
      "loss": 1.0648,
      "step": 2217
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.050122022628784,
      "learning_rate": 1.129891668217783e-06,
      "loss": 1.1097,
      "step": 2218
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.231600046157837,
      "learning_rate": 1.124142109954459e-06,
      "loss": 1.1382,
      "step": 2219
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.143968343734741,
      "learning_rate": 1.1184063462859262e-06,
      "loss": 1.0886,
      "step": 2220
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8700674772262573,
      "learning_rate": 1.1126843861265347e-06,
      "loss": 1.0924,
      "step": 2221
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8149470090866089,
      "learning_rate": 1.1069762383691862e-06,
      "loss": 1.0958,
      "step": 2222
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.140536308288574,
      "learning_rate": 1.1012819118853147e-06,
      "loss": 1.0449,
      "step": 2223
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.17146897315979,
      "learning_rate": 1.095601415524874e-06,
      "loss": 1.0849,
      "step": 2224
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9741861820220947,
      "learning_rate": 1.0899347581163222e-06,
      "loss": 1.0996,
      "step": 2225
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.95802640914917,
      "learning_rate": 1.08428194846661e-06,
      "loss": 1.0796,
      "step": 2226
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.226527690887451,
      "learning_rate": 1.0786429953611665e-06,
      "loss": 1.0664,
      "step": 2227
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9510045051574707,
      "learning_rate": 1.073017907563887e-06,
      "loss": 1.0981,
      "step": 2228
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0927810668945312,
      "learning_rate": 1.0674066938171123e-06,
      "loss": 1.1059,
      "step": 2229
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.412574052810669,
      "learning_rate": 1.061809362841626e-06,
      "loss": 1.1613,
      "step": 2230
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0362143516540527,
      "learning_rate": 1.0562259233366334e-06,
      "loss": 1.1,
      "step": 2231
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.145831346511841,
      "learning_rate": 1.0506563839797501e-06,
      "loss": 1.0809,
      "step": 2232
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.891459345817566,
      "learning_rate": 1.0451007534269908e-06,
      "loss": 1.0567,
      "step": 2233
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.041980743408203,
      "learning_rate": 1.0395590403127487e-06,
      "loss": 1.1105,
      "step": 2234
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1914401054382324,
      "learning_rate": 1.034031253249792e-06,
      "loss": 1.051,
      "step": 2235
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8129850625991821,
      "learning_rate": 1.0285174008292431e-06,
      "loss": 1.1056,
      "step": 2236
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0140349864959717,
      "learning_rate": 1.0230174916205681e-06,
      "loss": 1.0653,
      "step": 2237
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9827483892440796,
      "learning_rate": 1.0175315341715598e-06,
      "loss": 1.0502,
      "step": 2238
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.185121536254883,
      "learning_rate": 1.012059537008332e-06,
      "loss": 1.0522,
      "step": 2239
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.18265438079834,
      "learning_rate": 1.0066015086353e-06,
      "loss": 1.0585,
      "step": 2240
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.997373342514038,
      "learning_rate": 1.0011574575351702e-06,
      "loss": 1.1126,
      "step": 2241
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0864250659942627,
      "learning_rate": 9.957273921689237e-07,
      "loss": 1.053,
      "step": 2242
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.429229736328125,
      "learning_rate": 9.903113209758098e-07,
      "loss": 1.03,
      "step": 2243
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0856714248657227,
      "learning_rate": 9.849092523733217e-07,
      "loss": 1.0584,
      "step": 2244
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9060333967208862,
      "learning_rate": 9.795211947571993e-07,
      "loss": 1.0725,
      "step": 2245
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.011547327041626,
      "learning_rate": 9.74147156501396e-07,
      "loss": 1.0268,
      "step": 2246
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.037419080734253,
      "learning_rate": 9.687871459580845e-07,
      "loss": 1.1002,
      "step": 2247
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1190500259399414,
      "learning_rate": 9.634411714576353e-07,
      "loss": 1.0808,
      "step": 2248
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1224915981292725,
      "learning_rate": 9.58109241308599e-07,
      "loss": 1.1201,
      "step": 2249
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.025413990020752,
      "learning_rate": 9.527913637977093e-07,
      "loss": 0.9949,
      "step": 2250
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1313462257385254,
      "learning_rate": 9.474875471898526e-07,
      "loss": 1.039,
      "step": 2251
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0565345287323,
      "learning_rate": 9.421977997280596e-07,
      "loss": 1.1147,
      "step": 2252
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0440540313720703,
      "learning_rate": 9.369221296335007e-07,
      "loss": 1.0524,
      "step": 2253
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9696065187454224,
      "learning_rate": 9.31660545105465e-07,
      "loss": 1.0155,
      "step": 2254
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.204986810684204,
      "learning_rate": 9.264130543213512e-07,
      "loss": 0.9845,
      "step": 2255
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0204014778137207,
      "learning_rate": 9.211796654366545e-07,
      "loss": 1.0524,
      "step": 2256
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.032377243041992,
      "learning_rate": 9.15960386584952e-07,
      "loss": 1.0658,
      "step": 2257
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8374688625335693,
      "learning_rate": 9.107552258778907e-07,
      "loss": 1.0714,
      "step": 2258
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.3109209537506104,
      "learning_rate": 9.055641914051783e-07,
      "loss": 1.0687,
      "step": 2259
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8712282180786133,
      "learning_rate": 9.00387291234569e-07,
      "loss": 1.0721,
      "step": 2260
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9640578031539917,
      "learning_rate": 8.952245334118415e-07,
      "loss": 1.0744,
      "step": 2261
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.013775110244751,
      "learning_rate": 8.900759259608038e-07,
      "loss": 1.0481,
      "step": 2262
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0952086448669434,
      "learning_rate": 8.849414768832687e-07,
      "loss": 1.071,
      "step": 2263
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.023747205734253,
      "learning_rate": 8.79821194159044e-07,
      "loss": 0.9885,
      "step": 2264
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0883166790008545,
      "learning_rate": 8.747150857459219e-07,
      "loss": 1.0453,
      "step": 2265
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0477912425994873,
      "learning_rate": 8.696231595796633e-07,
      "loss": 1.0736,
      "step": 2266
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.084082841873169,
      "learning_rate": 8.645454235739903e-07,
      "loss": 1.0837,
      "step": 2267
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.111997127532959,
      "learning_rate": 8.5948188562057e-07,
      "loss": 1.0739,
      "step": 2268
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.1199848651885986,
      "learning_rate": 8.544325535889997e-07,
      "loss": 1.0656,
      "step": 2269
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0807812213897705,
      "learning_rate": 8.493974353268019e-07,
      "loss": 1.0965,
      "step": 2270
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.974009394645691,
      "learning_rate": 8.443765386594094e-07,
      "loss": 1.1146,
      "step": 2271
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.0679733753204346,
      "learning_rate": 8.393698713901499e-07,
      "loss": 1.1608,
      "step": 2272
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7803610563278198,
      "learning_rate": 8.343774413002382e-07,
      "loss": 1.0548,
      "step": 2273
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8302640914916992,
      "learning_rate": 8.293992561487596e-07,
      "loss": 1.0233,
      "step": 2274
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.059798002243042,
      "learning_rate": 8.24435323672661e-07,
      "loss": 1.0603,
      "step": 2275
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0559043884277344,
      "learning_rate": 8.194856515867433e-07,
      "loss": 1.0392,
      "step": 2276
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.862945795059204,
      "learning_rate": 8.145502475836331e-07,
      "loss": 1.0614,
      "step": 2277
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7939507961273193,
      "learning_rate": 8.096291193337935e-07,
      "loss": 1.0377,
      "step": 2278
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.17577862739563,
      "learning_rate": 8.047222744854943e-07,
      "loss": 1.0672,
      "step": 2279
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.036130905151367,
      "learning_rate": 7.998297206648075e-07,
      "loss": 1.0628,
      "step": 2280
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.190807342529297,
      "learning_rate": 7.949514654755963e-07,
      "loss": 1.0332,
      "step": 2281
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9127849340438843,
      "learning_rate": 7.900875164995003e-07,
      "loss": 0.9497,
      "step": 2282
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9593311548233032,
      "learning_rate": 7.852378812959227e-07,
      "loss": 0.984,
      "step": 2283
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.088101625442505,
      "learning_rate": 7.804025674020266e-07,
      "loss": 1.107,
      "step": 2284
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9637494087219238,
      "learning_rate": 7.755815823327084e-07,
      "loss": 1.054,
      "step": 2285
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8121821880340576,
      "learning_rate": 7.707749335806014e-07,
      "loss": 1.1599,
      "step": 2286
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9897934198379517,
      "learning_rate": 7.659826286160565e-07,
      "loss": 1.0588,
      "step": 2287
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.107254981994629,
      "learning_rate": 7.612046748871327e-07,
      "loss": 1.1113,
      "step": 2288
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2427737712860107,
      "learning_rate": 7.564410798195832e-07,
      "loss": 1.0776,
      "step": 2289
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0249686241149902,
      "learning_rate": 7.516918508168469e-07,
      "loss": 1.1112,
      "step": 2290
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.2287402153015137,
      "learning_rate": 7.46956995260033e-07,
      "loss": 1.0967,
      "step": 2291
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9392672777175903,
      "learning_rate": 7.422365205079141e-07,
      "loss": 1.1347,
      "step": 2292
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.045301914215088,
      "learning_rate": 7.375304338969135e-07,
      "loss": 1.0931,
      "step": 2293
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.170807361602783,
      "learning_rate": 7.328387427410877e-07,
      "loss": 1.0796,
      "step": 2294
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.016481399536133,
      "learning_rate": 7.281614543321269e-07,
      "loss": 1.0368,
      "step": 2295
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0962867736816406,
      "learning_rate": 7.234985759393321e-07,
      "loss": 1.1004,
      "step": 2296
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1926701068878174,
      "learning_rate": 7.188501148096117e-07,
      "loss": 1.1033,
      "step": 2297
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.0737500190734863,
      "learning_rate": 7.142160781674645e-07,
      "loss": 1.0647,
      "step": 2298
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.291659116744995,
      "learning_rate": 7.095964732149741e-07,
      "loss": 1.0718,
      "step": 2299
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.697638988494873,
      "learning_rate": 7.049913071317927e-07,
      "loss": 1.148,
      "step": 2300
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0080626010894775,
      "learning_rate": 7.004005870751341e-07,
      "loss": 1.0637,
      "step": 2301
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0152671337127686,
      "learning_rate": 6.958243201797554e-07,
      "loss": 1.0846,
      "step": 2302
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0115272998809814,
      "learning_rate": 6.912625135579587e-07,
      "loss": 1.0645,
      "step": 2303
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7418760061264038,
      "learning_rate": 6.867151742995659e-07,
      "loss": 1.0258,
      "step": 2304
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.089531660079956,
      "learning_rate": 6.821823094719171e-07,
      "loss": 1.0719,
      "step": 2305
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0292317867279053,
      "learning_rate": 6.776639261198581e-07,
      "loss": 1.0363,
      "step": 2306
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.093168020248413,
      "learning_rate": 6.731600312657238e-07,
      "loss": 1.1466,
      "step": 2307
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1573081016540527,
      "learning_rate": 6.68670631909335e-07,
      "loss": 1.0728,
      "step": 2308
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9670580625534058,
      "learning_rate": 6.641957350279838e-07,
      "loss": 1.0909,
      "step": 2309
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2009267807006836,
      "learning_rate": 6.597353475764201e-07,
      "loss": 1.0464,
      "step": 2310
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8985291719436646,
      "learning_rate": 6.552894764868456e-07,
      "loss": 0.9887,
      "step": 2311
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9667905569076538,
      "learning_rate": 6.508581286689031e-07,
      "loss": 1.0682,
      "step": 2312
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9581094980239868,
      "learning_rate": 6.464413110096601e-07,
      "loss": 1.1011,
      "step": 2313
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.992676019668579,
      "learning_rate": 6.420390303736034e-07,
      "loss": 1.0485,
      "step": 2314
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8992412090301514,
      "learning_rate": 6.37651293602628e-07,
      "loss": 1.0851,
      "step": 2315
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8477810621261597,
      "learning_rate": 6.332781075160244e-07,
      "loss": 0.9614,
      "step": 2316
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0865042209625244,
      "learning_rate": 6.28919478910468e-07,
      "loss": 1.1044,
      "step": 2317
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.013518810272217,
      "learning_rate": 6.245754145600091e-07,
      "loss": 1.0431,
      "step": 2318
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.1170589923858643,
      "learning_rate": 6.202459212160638e-07,
      "loss": 1.0597,
      "step": 2319
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.3192379474639893,
      "learning_rate": 6.15931005607403e-07,
      "loss": 1.0608,
      "step": 2320
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.0935842990875244,
      "learning_rate": 6.116306744401391e-07,
      "loss": 1.054,
      "step": 2321
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.086225986480713,
      "learning_rate": 6.073449343977211e-07,
      "loss": 1.0625,
      "step": 2322
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.019973039627075,
      "learning_rate": 6.030737921409169e-07,
      "loss": 1.1542,
      "step": 2323
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.229466199874878,
      "learning_rate": 5.988172543078097e-07,
      "loss": 1.0617,
      "step": 2324
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.076089859008789,
      "learning_rate": 5.945753275137844e-07,
      "loss": 1.1159,
      "step": 2325
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.048313617706299,
      "learning_rate": 5.903480183515197e-07,
      "loss": 1.1077,
      "step": 2326
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.140153169631958,
      "learning_rate": 5.861353333909692e-07,
      "loss": 1.0854,
      "step": 2327
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.049351692199707,
      "learning_rate": 5.819372791793654e-07,
      "loss": 1.0249,
      "step": 2328
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.049881935119629,
      "learning_rate": 5.777538622412005e-07,
      "loss": 1.1159,
      "step": 2329
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.123831033706665,
      "learning_rate": 5.735850890782158e-07,
      "loss": 1.1101,
      "step": 2330
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0053601264953613,
      "learning_rate": 5.694309661693942e-07,
      "loss": 1.075,
      "step": 2331
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.256922960281372,
      "learning_rate": 5.652914999709535e-07,
      "loss": 1.0569,
      "step": 2332
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.110792636871338,
      "learning_rate": 5.611666969163243e-07,
      "loss": 1.0724,
      "step": 2333
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.3687119483947754,
      "learning_rate": 5.570565634161562e-07,
      "loss": 1.1015,
      "step": 2334
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1923677921295166,
      "learning_rate": 5.529611058582951e-07,
      "loss": 1.0991,
      "step": 2335
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2981626987457275,
      "learning_rate": 5.488803306077806e-07,
      "loss": 1.1048,
      "step": 2336
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0987839698791504,
      "learning_rate": 5.448142440068316e-07,
      "loss": 1.0777,
      "step": 2337
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0344395637512207,
      "learning_rate": 5.407628523748398e-07,
      "loss": 1.0964,
      "step": 2338
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8401011228561401,
      "learning_rate": 5.367261620083575e-07,
      "loss": 1.0733,
      "step": 2339
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.2768774032592773,
      "learning_rate": 5.327041791810905e-07,
      "loss": 1.1034,
      "step": 2340
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.910409927368164,
      "learning_rate": 5.286969101438821e-07,
      "loss": 0.9979,
      "step": 2341
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9292501211166382,
      "learning_rate": 5.247043611247127e-07,
      "loss": 1.1476,
      "step": 2342
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.198500394821167,
      "learning_rate": 5.207265383286831e-07,
      "loss": 1.0723,
      "step": 2343
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8000935316085815,
      "learning_rate": 5.167634479380068e-07,
      "loss": 1.0985,
      "step": 2344
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.109813690185547,
      "learning_rate": 5.128150961120026e-07,
      "loss": 1.0915,
      "step": 2345
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0207180976867676,
      "learning_rate": 5.088814889870808e-07,
      "loss": 1.0622,
      "step": 2346
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.038322687149048,
      "learning_rate": 5.049626326767366e-07,
      "loss": 1.0576,
      "step": 2347
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.070478677749634,
      "learning_rate": 5.010585332715401e-07,
      "loss": 1.0528,
      "step": 2348
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0810468196868896,
      "learning_rate": 4.97169196839129e-07,
      "loss": 1.131,
      "step": 2349
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1072449684143066,
      "learning_rate": 4.932946294241903e-07,
      "loss": 0.9887,
      "step": 2350
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0414559841156006,
      "learning_rate": 4.894348370484648e-07,
      "loss": 1.0335,
      "step": 2351
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.133049964904785,
      "learning_rate": 4.855898257107261e-07,
      "loss": 1.0331,
      "step": 2352
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0265235900878906,
      "learning_rate": 4.817596013867765e-07,
      "loss": 1.0819,
      "step": 2353
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0449888706207275,
      "learning_rate": 4.779441700294362e-07,
      "loss": 1.0938,
      "step": 2354
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.038867950439453,
      "learning_rate": 4.7414353756853773e-07,
      "loss": 1.1149,
      "step": 2355
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8237578868865967,
      "learning_rate": 4.703577099109091e-07,
      "loss": 1.0327,
      "step": 2356
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8971409797668457,
      "learning_rate": 4.6658669294037393e-07,
      "loss": 1.0908,
      "step": 2357
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.237762451171875,
      "learning_rate": 4.628304925177318e-07,
      "loss": 1.0381,
      "step": 2358
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9325631856918335,
      "learning_rate": 4.5908911448075746e-07,
      "loss": 1.1211,
      "step": 2359
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6598244905471802,
      "learning_rate": 4.553625646441928e-07,
      "loss": 1.127,
      "step": 2360
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9364616870880127,
      "learning_rate": 4.5165084879972844e-07,
      "loss": 1.1074,
      "step": 2361
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.069039821624756,
      "learning_rate": 4.4795397271600314e-07,
      "loss": 1.063,
      "step": 2362
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.239853858947754,
      "learning_rate": 4.4427194213859216e-07,
      "loss": 1.0851,
      "step": 2363
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.222581148147583,
      "learning_rate": 4.4060476278999897e-07,
      "loss": 1.034,
      "step": 2364
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2812037467956543,
      "learning_rate": 4.3695244036964567e-07,
      "loss": 1.0996,
      "step": 2365
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9026345014572144,
      "learning_rate": 4.3331498055386057e-07,
      "loss": 1.0387,
      "step": 2366
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9953440427780151,
      "learning_rate": 4.296923889958771e-07,
      "loss": 0.9965,
      "step": 2367
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0452935695648193,
      "learning_rate": 4.2608467132581934e-07,
      "loss": 1.0191,
      "step": 2368
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.091887950897217,
      "learning_rate": 4.224918331506955e-07,
      "loss": 1.0577,
      "step": 2369
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.0989646911621094,
      "learning_rate": 4.189138800543868e-07,
      "loss": 1.0393,
      "step": 2370
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.035747528076172,
      "learning_rate": 4.1535081759764286e-07,
      "loss": 1.1303,
      "step": 2371
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.128170967102051,
      "learning_rate": 4.118026513180695e-07,
      "loss": 1.0498,
      "step": 2372
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.275786876678467,
      "learning_rate": 4.082693867301224e-07,
      "loss": 1.0794,
      "step": 2373
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9808770418167114,
      "learning_rate": 4.0475102932509336e-07,
      "loss": 1.0775,
      "step": 2374
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8925813436508179,
      "learning_rate": 4.012475845711106e-07,
      "loss": 1.0103,
      "step": 2375
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9656004905700684,
      "learning_rate": 3.9775905791312407e-07,
      "loss": 1.0319,
      "step": 2376
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.265615224838257,
      "learning_rate": 3.9428545477289913e-07,
      "loss": 1.0996,
      "step": 2377
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.006573438644409,
      "learning_rate": 3.908267805490051e-07,
      "loss": 1.0412,
      "step": 2378
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9896271228790283,
      "learning_rate": 3.8738304061681107e-07,
      "loss": 1.0606,
      "step": 2379
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1295104026794434,
      "learning_rate": 3.8395424032847686e-07,
      "loss": 1.11,
      "step": 2380
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9212133884429932,
      "learning_rate": 3.8054038501294077e-07,
      "loss": 1.0659,
      "step": 2381
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9252375364303589,
      "learning_rate": 3.7714147997591655e-07,
      "loss": 1.0716,
      "step": 2382
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1474685668945312,
      "learning_rate": 3.7375753049987974e-07,
      "loss": 1.0997,
      "step": 2383
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9588407278060913,
      "learning_rate": 3.703885418440656e-07,
      "loss": 1.078,
      "step": 2384
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7770421504974365,
      "learning_rate": 3.6703451924445467e-07,
      "loss": 1.0949,
      "step": 2385
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9683630466461182,
      "learning_rate": 3.6369546791377054e-07,
      "loss": 1.0769,
      "step": 2386
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0799646377563477,
      "learning_rate": 3.603713930414676e-07,
      "loss": 1.0878,
      "step": 2387
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9410487413406372,
      "learning_rate": 3.570622997937234e-07,
      "loss": 1.0577,
      "step": 2388
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.126528024673462,
      "learning_rate": 3.5376819331343404e-07,
      "loss": 1.0947,
      "step": 2389
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.794063925743103,
      "learning_rate": 3.50489078720202e-07,
      "loss": 1.0356,
      "step": 2390
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.047170877456665,
      "learning_rate": 3.472249611103273e-07,
      "loss": 1.1459,
      "step": 2391
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0276432037353516,
      "learning_rate": 3.4397584555680653e-07,
      "loss": 1.0575,
      "step": 2392
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.10845685005188,
      "learning_rate": 3.4074173710931804e-07,
      "loss": 1.0813,
      "step": 2393
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0322184562683105,
      "learning_rate": 3.375226407942178e-07,
      "loss": 1.0825,
      "step": 2394
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.882827877998352,
      "learning_rate": 3.3431856161452835e-07,
      "loss": 1.0365,
      "step": 2395
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1573452949523926,
      "learning_rate": 3.3112950454993633e-07,
      "loss": 1.0988,
      "step": 2396
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.021303653717041,
      "learning_rate": 3.2795547455677813e-07,
      "loss": 1.0844,
      "step": 2397
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.980573058128357,
      "learning_rate": 3.247964765680389e-07,
      "loss": 1.0519,
      "step": 2398
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.849236249923706,
      "learning_rate": 3.2165251549333585e-07,
      "loss": 1.0518,
      "step": 2399
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.1389212608337402,
      "learning_rate": 3.185235962189237e-07,
      "loss": 1.0025,
      "step": 2400
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.837730884552002,
      "learning_rate": 3.1540972360767254e-07,
      "loss": 1.0212,
      "step": 2401
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.110355854034424,
      "learning_rate": 3.1231090249907336e-07,
      "loss": 1.0474,
      "step": 2402
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.140359878540039,
      "learning_rate": 3.0922713770922155e-07,
      "loss": 0.9382,
      "step": 2403
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9713770151138306,
      "learning_rate": 3.06158434030811e-07,
      "loss": 1.0731,
      "step": 2404
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0847887992858887,
      "learning_rate": 3.0310479623313125e-07,
      "loss": 1.1201,
      "step": 2405
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9484189748764038,
      "learning_rate": 3.0006622906205707e-07,
      "loss": 1.0504,
      "step": 2406
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.925686240196228,
      "learning_rate": 2.970427372400353e-07,
      "loss": 1.1226,
      "step": 2407
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8018156290054321,
      "learning_rate": 2.940343254660905e-07,
      "loss": 1.0784,
      "step": 2408
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9751100540161133,
      "learning_rate": 2.910409984158058e-07,
      "loss": 1.0884,
      "step": 2409
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.311815023422241,
      "learning_rate": 2.880627607413211e-07,
      "loss": 1.0355,
      "step": 2410
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.182737350463867,
      "learning_rate": 2.8509961707132496e-07,
      "loss": 1.0633,
      "step": 2411
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.1470437049865723,
      "learning_rate": 2.821515720110479e-07,
      "loss": 1.1069,
      "step": 2412
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0461294651031494,
      "learning_rate": 2.7921863014225504e-07,
      "loss": 1.0784,
      "step": 2413
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9097718000411987,
      "learning_rate": 2.7630079602323447e-07,
      "loss": 1.1072,
      "step": 2414
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9900500774383545,
      "learning_rate": 2.733980741887987e-07,
      "loss": 1.1123,
      "step": 2415
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.1558196544647217,
      "learning_rate": 2.7051046915027244e-07,
      "loss": 1.0656,
      "step": 2416
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.172513246536255,
      "learning_rate": 2.676379853954858e-07,
      "loss": 1.0407,
      "step": 2417
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0111191272735596,
      "learning_rate": 2.6478062738876654e-07,
      "loss": 1.0794,
      "step": 2418
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0015742778778076,
      "learning_rate": 2.6193839957093683e-07,
      "loss": 1.0812,
      "step": 2419
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.2460951805114746,
      "learning_rate": 2.5911130635930203e-07,
      "loss": 1.091,
      "step": 2420
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.073729991912842,
      "learning_rate": 2.5629935214764866e-07,
      "loss": 1.0523,
      "step": 2421
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.888290524482727,
      "learning_rate": 2.5350254130622866e-07,
      "loss": 1.0851,
      "step": 2422
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9870355129241943,
      "learning_rate": 2.507208781817638e-07,
      "loss": 1.0974,
      "step": 2423
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.005234718322754,
      "learning_rate": 2.4795436709743384e-07,
      "loss": 1.0392,
      "step": 2424
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0189740657806396,
      "learning_rate": 2.45203012352867e-07,
      "loss": 1.0789,
      "step": 2425
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9499077796936035,
      "learning_rate": 2.424668182241363e-07,
      "loss": 1.076,
      "step": 2426
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.083486318588257,
      "learning_rate": 2.3974578896375555e-07,
      "loss": 1.0328,
      "step": 2427
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.0778701305389404,
      "learning_rate": 2.370399288006664e-07,
      "loss": 1.0608,
      "step": 2428
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.134263753890991,
      "learning_rate": 2.3434924194023712e-07,
      "loss": 1.0484,
      "step": 2429
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2767274379730225,
      "learning_rate": 2.3167373256425152e-07,
      "loss": 1.0893,
      "step": 2430
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.701598882675171,
      "learning_rate": 2.2901340483090785e-07,
      "loss": 1.0606,
      "step": 2431
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.017697334289551,
      "learning_rate": 2.2636826287480872e-07,
      "loss": 1.0668,
      "step": 2432
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.8839207887649536,
      "learning_rate": 2.2373831080695463e-07,
      "loss": 1.1023,
      "step": 2433
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.043126344680786,
      "learning_rate": 2.211235527147404e-07,
      "loss": 1.0819,
      "step": 2434
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9526861906051636,
      "learning_rate": 2.1852399266194312e-07,
      "loss": 1.0848,
      "step": 2435
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9554346799850464,
      "learning_rate": 2.1593963468872435e-07,
      "loss": 1.0204,
      "step": 2436
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.3021304607391357,
      "learning_rate": 2.1337048281161565e-07,
      "loss": 1.074,
      "step": 2437
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.0305123329162598,
      "learning_rate": 2.1081654102351634e-07,
      "loss": 1.0801,
      "step": 2438
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9925532341003418,
      "learning_rate": 2.082778132936858e-07,
      "loss": 1.0754,
      "step": 2439
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9017348289489746,
      "learning_rate": 2.057543035677412e-07,
      "loss": 1.0859,
      "step": 2440
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1010007858276367,
      "learning_rate": 2.0324601576764525e-07,
      "loss": 1.0757,
      "step": 2441
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.935064435005188,
      "learning_rate": 2.0075295379170413e-07,
      "loss": 1.1041,
      "step": 2442
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.06764817237854,
      "learning_rate": 1.9827512151456175e-07,
      "loss": 1.0956,
      "step": 2443
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.085620164871216,
      "learning_rate": 1.95812522787191e-07,
      "loss": 1.1266,
      "step": 2444
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6828104257583618,
      "learning_rate": 1.933651614368892e-07,
      "loss": 1.094,
      "step": 2445
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9124735593795776,
      "learning_rate": 1.9093304126727497e-07,
      "loss": 1.0802,
      "step": 2446
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1738624572753906,
      "learning_rate": 1.885161660582746e-07,
      "loss": 1.1414,
      "step": 2447
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.26924204826355,
      "learning_rate": 1.8611453956612346e-07,
      "loss": 1.0793,
      "step": 2448
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1126022338867188,
      "learning_rate": 1.8372816552336025e-07,
      "loss": 1.0846,
      "step": 2449
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9607864618301392,
      "learning_rate": 1.81357047638816e-07,
      "loss": 1.0689,
      "step": 2450
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1523258686065674,
      "learning_rate": 1.7900118959761181e-07,
      "loss": 1.0748,
      "step": 2451
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.050507068634033,
      "learning_rate": 1.7666059506115108e-07,
      "loss": 1.0147,
      "step": 2452
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.2121455669403076,
      "learning_rate": 1.7433526766711727e-07,
      "loss": 1.1447,
      "step": 2453
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.009307384490967,
      "learning_rate": 1.7202521102946624e-07,
      "loss": 1.0201,
      "step": 2454
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.025744676589966,
      "learning_rate": 1.6973042873841827e-07,
      "loss": 1.0353,
      "step": 2455
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.169806957244873,
      "learning_rate": 1.6745092436045495e-07,
      "loss": 1.0997,
      "step": 2456
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8906662464141846,
      "learning_rate": 1.651867014383146e-07,
      "loss": 1.0934,
      "step": 2457
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.911436676979065,
      "learning_rate": 1.629377634909868e-07,
      "loss": 1.0961,
      "step": 2458
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2790775299072266,
      "learning_rate": 1.6070411401370335e-07,
      "loss": 1.1441,
      "step": 2459
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.14805269241333,
      "learning_rate": 1.584857564779363e-07,
      "loss": 1.072,
      "step": 2460
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.150108814239502,
      "learning_rate": 1.5628269433139e-07,
      "loss": 1.1347,
      "step": 2461
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1906936168670654,
      "learning_rate": 1.5409493099799888e-07,
      "loss": 1.0838,
      "step": 2462
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.09163761138916,
      "learning_rate": 1.519224698779198e-07,
      "loss": 1.0913,
      "step": 2463
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9614250659942627,
      "learning_rate": 1.497653143475275e-07,
      "loss": 1.1053,
      "step": 2464
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9306327104568481,
      "learning_rate": 1.4762346775940794e-07,
      "loss": 1.0794,
      "step": 2465
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.227121591567993,
      "learning_rate": 1.4549693344235726e-07,
      "loss": 1.0802,
      "step": 2466
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0046396255493164,
      "learning_rate": 1.4338571470137063e-07,
      "loss": 1.0618,
      "step": 2467
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.018825054168701,
      "learning_rate": 1.4128981481764115e-07,
      "loss": 1.0512,
      "step": 2468
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9474681615829468,
      "learning_rate": 1.3920923704855648e-07,
      "loss": 1.0444,
      "step": 2469
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8143775463104248,
      "learning_rate": 1.3714398462768563e-07,
      "loss": 1.068,
      "step": 2470
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.1274454593658447,
      "learning_rate": 1.350940607647866e-07,
      "loss": 1.0854,
      "step": 2471
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.104363203048706,
      "learning_rate": 1.3305946864578645e-07,
      "loss": 1.0997,
      "step": 2472
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0627822875976562,
      "learning_rate": 1.3104021143278911e-07,
      "loss": 1.089,
      "step": 2473
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8147544860839844,
      "learning_rate": 1.2903629226406423e-07,
      "loss": 1.0824,
      "step": 2474
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.2368528842926025,
      "learning_rate": 1.2704771425404382e-07,
      "loss": 1.1401,
      "step": 2475
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9573479890823364,
      "learning_rate": 1.2507448049331571e-07,
      "loss": 1.0322,
      "step": 2476
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.112328290939331,
      "learning_rate": 1.231165940486234e-07,
      "loss": 1.0611,
      "step": 2477
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8626047372817993,
      "learning_rate": 1.2117405796285286e-07,
      "loss": 1.0151,
      "step": 2478
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9831112623214722,
      "learning_rate": 1.192468752550402e-07,
      "loss": 0.9917,
      "step": 2479
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.795022964477539,
      "learning_rate": 1.1733504892035064e-07,
      "loss": 1.1235,
      "step": 2480
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0961997509002686,
      "learning_rate": 1.1543858193009183e-07,
      "loss": 1.0554,
      "step": 2481
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1731226444244385,
      "learning_rate": 1.1355747723169386e-07,
      "loss": 1.0848,
      "step": 2482
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.895052194595337,
      "learning_rate": 1.1169173774871478e-07,
      "loss": 1.1007,
      "step": 2483
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0253288745880127,
      "learning_rate": 1.0984136638083176e-07,
      "loss": 1.0721,
      "step": 2484
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0866169929504395,
      "learning_rate": 1.0800636600383662e-07,
      "loss": 1.0972,
      "step": 2485
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.3037004470825195,
      "learning_rate": 1.0618673946963365e-07,
      "loss": 1.0401,
      "step": 2486
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.89889657497406,
      "learning_rate": 1.0438248960623065e-07,
      "loss": 1.0305,
      "step": 2487
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.968326210975647,
      "learning_rate": 1.0259361921774014e-07,
      "loss": 1.1318,
      "step": 2488
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8982306718826294,
      "learning_rate": 1.0082013108437038e-07,
      "loss": 1.104,
      "step": 2489
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1765213012695312,
      "learning_rate": 9.906202796242547e-08,
      "loss": 1.014,
      "step": 2490
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.173142433166504,
      "learning_rate": 9.731931258429638e-08,
      "loss": 1.0356,
      "step": 2491
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0092382431030273,
      "learning_rate": 9.55919876584599e-08,
      "loss": 1.0961,
      "step": 2492
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1994028091430664,
      "learning_rate": 9.388005586947191e-08,
      "loss": 1.1098,
      "step": 2493
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0307843685150146,
      "learning_rate": 9.218351987796859e-08,
      "loss": 1.033,
      "step": 2494
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0328028202056885,
      "learning_rate": 9.0502382320653e-08,
      "loss": 1.1307,
      "step": 2495
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2205312252044678,
      "learning_rate": 8.88366458102996e-08,
      "loss": 1.0793,
      "step": 2496
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.159393548965454,
      "learning_rate": 8.718631293574753e-08,
      "loss": 1.0719,
      "step": 2497
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0513756275177,
      "learning_rate": 8.555138626189619e-08,
      "loss": 1.0355,
      "step": 2498
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0853822231292725,
      "learning_rate": 8.393186832969746e-08,
      "loss": 1.0679,
      "step": 2499
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0627925395965576,
      "learning_rate": 8.232776165616019e-08,
      "loss": 1.0539,
      "step": 2500
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.170922040939331,
      "learning_rate": 8.07390687343379e-08,
      "loss": 1.0203,
      "step": 2501
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8435313701629639,
      "learning_rate": 7.916579203333107e-08,
      "loss": 1.14,
      "step": 2502
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0338761806488037,
      "learning_rate": 7.760793399827937e-08,
      "loss": 1.1038,
      "step": 2503
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.090447187423706,
      "learning_rate": 7.606549705035937e-08,
      "loss": 1.1506,
      "step": 2504
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.803004264831543,
      "learning_rate": 7.453848358678018e-08,
      "loss": 1.125,
      "step": 2505
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1259775161743164,
      "learning_rate": 7.302689598078117e-08,
      "loss": 1.0381,
      "step": 2506
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.2720749378204346,
      "learning_rate": 7.153073658162646e-08,
      "loss": 1.1518,
      "step": 2507
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0580289363861084,
      "learning_rate": 7.00500077146038e-08,
      "loss": 1.0111,
      "step": 2508
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0478155612945557,
      "learning_rate": 6.858471168101788e-08,
      "loss": 1.1509,
      "step": 2509
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.087164878845215,
      "learning_rate": 6.713485075818815e-08,
      "loss": 1.0572,
      "step": 2510
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9714109897613525,
      "learning_rate": 6.570042719944436e-08,
      "loss": 1.0994,
      "step": 2511
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.789076566696167,
      "learning_rate": 6.428144323412544e-08,
      "loss": 1.0542,
      "step": 2512
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.009892463684082,
      "learning_rate": 6.287790106757396e-08,
      "loss": 1.1219,
      "step": 2513
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0267467498779297,
      "learning_rate": 6.148980288113504e-08,
      "loss": 1.0895,
      "step": 2514
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.174973726272583,
      "learning_rate": 6.011715083214742e-08,
      "loss": 1.0424,
      "step": 2515
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.125401020050049,
      "learning_rate": 5.875994705394794e-08,
      "loss": 1.1438,
      "step": 2516
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.229830265045166,
      "learning_rate": 5.7418193655861545e-08,
      "loss": 1.029,
      "step": 2517
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.054973840713501,
      "learning_rate": 5.609189272320237e-08,
      "loss": 1.0411,
      "step": 2518
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.059713125228882,
      "learning_rate": 5.4781046317267103e-08,
      "loss": 1.1116,
      "step": 2519
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.131804943084717,
      "learning_rate": 5.348565647533388e-08,
      "loss": 1.0682,
      "step": 2520
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0404107570648193,
      "learning_rate": 5.220572521066003e-08,
      "loss": 1.0843,
      "step": 2521
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.020953416824341,
      "learning_rate": 5.094125451247656e-08,
      "loss": 1.0885,
      "step": 2522
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9062414169311523,
      "learning_rate": 4.9692246345985905e-08,
      "loss": 1.0399,
      "step": 2523
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.029109239578247,
      "learning_rate": 4.8458702652359744e-08,
      "loss": 1.086,
      "step": 2524
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7486499547958374,
      "learning_rate": 4.7240625348735636e-08,
      "loss": 1.1269,
      "step": 2525
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8982162475585938,
      "learning_rate": 4.603801632821148e-08,
      "loss": 1.105,
      "step": 2526
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.903022050857544,
      "learning_rate": 4.485087745984884e-08,
      "loss": 1.0737,
      "step": 2527
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7210338115692139,
      "learning_rate": 4.367921058866187e-08,
      "loss": 1.0088,
      "step": 2528
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.241347551345825,
      "learning_rate": 4.252301753562171e-08,
      "loss": 1.0598,
      "step": 2529
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.960512399673462,
      "learning_rate": 4.1382300097647655e-08,
      "loss": 1.0745,
      "step": 2530
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2497718334198,
      "learning_rate": 4.025706004760932e-08,
      "loss": 1.0728,
      "step": 2531
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.172863006591797,
      "learning_rate": 3.914729913432336e-08,
      "loss": 1.0534,
      "step": 2532
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.040266990661621,
      "learning_rate": 3.805301908254455e-08,
      "loss": 1.0314,
      "step": 2533
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.103468894958496,
      "learning_rate": 3.697422159297248e-08,
      "loss": 1.05,
      "step": 2534
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.111663341522217,
      "learning_rate": 3.591090834224153e-08,
      "loss": 1.0741,
      "step": 2535
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9815165996551514,
      "learning_rate": 3.4863080982919794e-08,
      "loss": 1.0453,
      "step": 2536
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3565168380737305,
      "learning_rate": 3.383074114351237e-08,
      "loss": 1.1072,
      "step": 2537
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1248133182525635,
      "learning_rate": 3.281389042844918e-08,
      "loss": 1.0283,
      "step": 2538
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9816641807556152,
      "learning_rate": 3.181253041809052e-08,
      "loss": 1.0716,
      "step": 2539
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1975905895233154,
      "learning_rate": 3.082666266872036e-08,
      "loss": 1.1003,
      "step": 2540
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.950649619102478,
      "learning_rate": 2.9856288712544204e-08,
      "loss": 1.0767,
      "step": 2541
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9536888599395752,
      "learning_rate": 2.8901410057688982e-08,
      "loss": 1.0356,
      "step": 2542
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.220674991607666,
      "learning_rate": 2.796202818819871e-08,
      "loss": 1.1099,
      "step": 2543
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.056966781616211,
      "learning_rate": 2.7038144564033307e-08,
      "loss": 1.0299,
      "step": 2544
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0472664833068848,
      "learning_rate": 2.6129760621063095e-08,
      "loss": 1.1414,
      "step": 2545
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0530166625976562,
      "learning_rate": 2.5236877771074308e-08,
      "loss": 1.0585,
      "step": 2546
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0605247020721436,
      "learning_rate": 2.4359497401758026e-08,
      "loss": 1.0723,
      "step": 2547
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0177693367004395,
      "learning_rate": 2.349762087671126e-08,
      "loss": 1.0698,
      "step": 2548
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7948215007781982,
      "learning_rate": 2.265124953543918e-08,
      "loss": 1.0007,
      "step": 2549
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.061897039413452,
      "learning_rate": 2.182038469334513e-08,
      "loss": 1.0895,
      "step": 2550
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.034893274307251,
      "learning_rate": 2.1005027641736176e-08,
      "loss": 1.0168,
      "step": 2551
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.843342900276184,
      "learning_rate": 2.020517964781532e-08,
      "loss": 1.1367,
      "step": 2552
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7608646154403687,
      "learning_rate": 1.9420841954681525e-08,
      "loss": 0.9844,
      "step": 2553
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9821500778198242,
      "learning_rate": 1.86520157813308e-08,
      "loss": 1.0697,
      "step": 2554
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.0149457454681396,
      "learning_rate": 1.7898702322648453e-08,
      "loss": 1.1024,
      "step": 2555
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.2242860794067383,
      "learning_rate": 1.716090274941351e-08,
      "loss": 1.0704,
      "step": 2556
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.3519678115844727,
      "learning_rate": 1.6438618208290957e-08,
      "loss": 1.0307,
      "step": 2557
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.1405584812164307,
      "learning_rate": 1.5731849821833955e-08,
      "loss": 1.0304,
      "step": 2558
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.090714931488037,
      "learning_rate": 1.5040598688482732e-08,
      "loss": 1.0093,
      "step": 2559
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0778794288635254,
      "learning_rate": 1.436486588255681e-08,
      "loss": 1.0848,
      "step": 2560
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0867276191711426,
      "learning_rate": 1.370465245426167e-08,
      "loss": 1.0406,
      "step": 2561
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0437560081481934,
      "learning_rate": 1.3059959429679859e-08,
      "loss": 1.0722,
      "step": 2562
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0525994300842285,
      "learning_rate": 1.2430787810776556e-08,
      "loss": 1.1118,
      "step": 2563
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1362595558166504,
      "learning_rate": 1.1817138575389576e-08,
      "loss": 1.1266,
      "step": 2564
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.2187294960021973,
      "learning_rate": 1.1219012677234908e-08,
      "loss": 1.0558,
      "step": 2565
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.028306245803833,
      "learning_rate": 1.0636411045902296e-08,
      "loss": 1.1192,
      "step": 2566
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9894956350326538,
      "learning_rate": 1.0069334586854106e-08,
      "loss": 1.1483,
      "step": 2567
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.063035726547241,
      "learning_rate": 9.517784181422018e-09,
      "loss": 1.0788,
      "step": 2568
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1766254901885986,
      "learning_rate": 8.981760686811448e-09,
      "loss": 1.0418,
      "step": 2569
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0792109966278076,
      "learning_rate": 8.461264936093783e-09,
      "loss": 1.0708,
      "step": 2570
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.181647539138794,
      "learning_rate": 7.956297738207496e-09,
      "loss": 1.0896,
      "step": 2571
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0690011978149414,
      "learning_rate": 7.466859877958143e-09,
      "loss": 1.0564,
      "step": 2572
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1614553928375244,
      "learning_rate": 6.992952116013918e-09,
      "loss": 1.0632,
      "step": 2573
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0442588329315186,
      "learning_rate": 6.534575188911208e-09,
      "loss": 1.0799,
      "step": 2574
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9397488832473755,
      "learning_rate": 6.091729809042379e-09,
      "loss": 1.1116,
      "step": 2575
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9455980062484741,
      "learning_rate": 5.664416664666883e-09,
      "loss": 1.0148,
      "step": 2576
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.057600498199463,
      "learning_rate": 5.252636419902368e-09,
      "loss": 1.1327,
      "step": 2577
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9807806015014648,
      "learning_rate": 4.856389714723575e-09,
      "loss": 1.1269,
      "step": 2578
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1234798431396484,
      "learning_rate": 4.475677164966774e-09,
      "loss": 1.0241,
      "step": 2579
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8234140872955322,
      "learning_rate": 4.110499362323106e-09,
      "loss": 1.0341,
      "step": 2580
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.19174861907959,
      "learning_rate": 3.760856874341912e-09,
      "loss": 0.9994,
      "step": 2581
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1175832748413086,
      "learning_rate": 3.4267502444274013e-09,
      "loss": 1.0479,
      "step": 2582
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.1661698818206787,
      "learning_rate": 3.1081799918375454e-09,
      "loss": 1.1193,
      "step": 2583
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8538522720336914,
      "learning_rate": 2.805146611684073e-09,
      "loss": 1.118,
      "step": 2584
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8933985233306885,
      "learning_rate": 2.5176505749346937e-09,
      "loss": 1.0648,
      "step": 2585
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0827386379241943,
      "learning_rate": 2.245692328404214e-09,
      "loss": 1.1075,
      "step": 2586
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0926945209503174,
      "learning_rate": 1.9892722947645328e-09,
      "loss": 1.1722,
      "step": 2587
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.1068105697631836,
      "learning_rate": 1.7483908725357546e-09,
      "loss": 1.0713,
      "step": 2588
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9252822399139404,
      "learning_rate": 1.5230484360873043e-09,
      "loss": 1.0551,
      "step": 2589
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.285743236541748,
      "learning_rate": 1.3132453356412556e-09,
      "loss": 1.0933,
      "step": 2590
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0660808086395264,
      "learning_rate": 1.1189818972656697e-09,
      "loss": 1.0759,
      "step": 2591
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0082414150238037,
      "learning_rate": 9.40258422880147e-10,
      "loss": 1.1154,
      "step": 2592
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.082523822784424,
      "learning_rate": 7.770751902513862e-10,
      "loss": 1.14,
      "step": 2593
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0849084854125977,
      "learning_rate": 6.294324529942942e-10,
      "loss": 1.0109,
      "step": 2594
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.100006103515625,
      "learning_rate": 4.973304405697654e-10,
      "loss": 1.0643,
      "step": 2595
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.871573805809021,
      "learning_rate": 3.807693582869032e-10,
      "loss": 1.0228,
      "step": 2596
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0291032791137695,
      "learning_rate": 2.797493873019086e-10,
      "loss": 1.0246,
      "step": 2597
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8526897430419922,
      "learning_rate": 1.9427068461808086e-10,
      "loss": 1.1333,
      "step": 2598
    },
    {
      "epoch": 1.0,
      "step": 2598,
      "total_flos": 1.61739710955887e+18,
      "train_loss": 1.1385140157232292,
      "train_runtime": 9220.9,
      "train_samples_per_second": 72.151,
      "train_steps_per_second": 0.282
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2598,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50000,
  "total_flos": 1.61739710955887e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}