{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.942322383498854,
  "eval_steps": 655,
  "global_step": 10472,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 0.4369395077228546,
      "learning_rate": 1e-05,
      "loss": 1.3161,
      "step": 1
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.559309482574463,
      "eval_runtime": 17.7472,
      "eval_samples_per_second": 153.771,
      "eval_steps_per_second": 38.485,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5338654518127441,
      "learning_rate": 2e-05,
      "loss": 1.56,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.517049252986908,
      "learning_rate": 3e-05,
      "loss": 1.519,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6359376907348633,
      "learning_rate": 4e-05,
      "loss": 1.5065,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5600838661193848,
      "learning_rate": 5e-05,
      "loss": 1.3367,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.49170756340026855,
      "learning_rate": 6e-05,
      "loss": 1.4911,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5616886019706726,
      "learning_rate": 7e-05,
      "loss": 1.4518,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5916192531585693,
      "learning_rate": 8e-05,
      "loss": 1.1924,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6707382798194885,
      "learning_rate": 9e-05,
      "loss": 1.281,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8203673362731934,
      "learning_rate": 0.0001,
      "loss": 1.4793,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5819201469421387,
      "learning_rate": 0.00011000000000000002,
      "loss": 1.2622,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6588992476463318,
      "learning_rate": 0.00012,
      "loss": 1.226,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7845558524131775,
      "learning_rate": 0.00013000000000000002,
      "loss": 1.3695,
      "step": 13
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5720924139022827,
      "learning_rate": 0.00014,
      "loss": 1.3338,
      "step": 14
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5074018836021423,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.9675,
      "step": 15
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5563233494758606,
      "learning_rate": 0.00016,
      "loss": 1.3316,
      "step": 16
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5730008482933044,
      "learning_rate": 0.00017,
      "loss": 1.1587,
      "step": 17
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.557260274887085,
      "learning_rate": 0.00018,
      "loss": 1.0954,
      "step": 18
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5220329165458679,
      "learning_rate": 0.00019,
      "loss": 1.2881,
      "step": 19
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6541146039962769,
      "learning_rate": 0.0002,
      "loss": 1.1474,
      "step": 20
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.621552050113678,
      "learning_rate": 0.0001999999954827831,
      "loss": 1.1891,
      "step": 21
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6264808177947998,
      "learning_rate": 0.0001999999819311328,
      "loss": 1.2244,
      "step": 22
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.550438642501831,
      "learning_rate": 0.00019999995934505033,
      "loss": 1.3496,
      "step": 23
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5227052569389343,
      "learning_rate": 0.00019999992772453778,
      "loss": 1.1316,
      "step": 24
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.4750714898109436,
      "learning_rate": 0.0001999998870695979,
      "loss": 1.1112,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5615024566650391,
      "learning_rate": 0.00019999983738023446,
      "loss": 1.2386,
      "step": 26
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.4816851317882538,
      "learning_rate": 0.00019999977865645186,
      "loss": 1.0982,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5965765118598938,
      "learning_rate": 0.00019999971089825549,
      "loss": 1.2331,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5031258463859558,
      "learning_rate": 0.00019999963410565142,
      "loss": 1.2095,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5152090787887573,
      "learning_rate": 0.00019999954827864663,
      "loss": 1.1875,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5212087035179138,
      "learning_rate": 0.0001999994534172488,
      "loss": 1.1729,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.4427972137928009,
      "learning_rate": 0.0001999993495214666,
      "loss": 1.129,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5280154943466187,
      "learning_rate": 0.00019999923659130934,
      "loss": 1.1871,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.437427282333374,
      "learning_rate": 0.00019999911462678728,
      "loss": 1.0607,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.4341500699520111,
      "learning_rate": 0.00019999898362791139,
      "loss": 1.1404,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.45423945784568787,
      "learning_rate": 0.0001999988435946935,
      "loss": 1.1451,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.47884103655815125,
      "learning_rate": 0.0001999986945271463,
      "loss": 1.1615,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.4309810996055603,
      "learning_rate": 0.00019999853642528327,
      "loss": 1.0011,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.4934332072734833,
      "learning_rate": 0.0001999983692891186,
      "loss": 1.2428,
      "step": 39
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5232778787612915,
      "learning_rate": 0.0001999981931186675,
      "loss": 1.2445,
      "step": 40
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4906657934188843,
      "learning_rate": 0.00019999800791394582,
      "loss": 1.112,
      "step": 41
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4859429895877838,
      "learning_rate": 0.00019999781367497033,
      "loss": 1.2359,
      "step": 42
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5400615930557251,
      "learning_rate": 0.00019999761040175856,
      "loss": 1.1782,
      "step": 43
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.48445501923561096,
      "learning_rate": 0.00019999739809432887,
      "loss": 1.1459,
      "step": 44
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6529949307441711,
      "learning_rate": 0.0001999971767527004,
      "loss": 1.2262,
      "step": 45
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5127531886100769,
      "learning_rate": 0.00019999694637689327,
      "loss": 1.1846,
      "step": 46
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5761195421218872,
      "learning_rate": 0.00019999670696692817,
      "loss": 1.009,
      "step": 47
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.469578355550766,
      "learning_rate": 0.0001999964585228268,
      "loss": 1.0883,
      "step": 48
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5488829612731934,
      "learning_rate": 0.00019999620104461157,
      "loss": 1.1774,
      "step": 49
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5317593216896057,
      "learning_rate": 0.00019999593453230575,
      "loss": 1.2767,
      "step": 50
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4968358278274536,
      "learning_rate": 0.0001999956589859334,
      "loss": 1.2343,
      "step": 51
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5535591244697571,
      "learning_rate": 0.00019999537440551946,
      "loss": 1.18,
      "step": 52
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4718022644519806,
      "learning_rate": 0.00019999508079108962,
      "loss": 1.1936,
      "step": 53
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5101296305656433,
      "learning_rate": 0.0001999947781426704,
      "loss": 1.3481,
      "step": 54
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4822118282318115,
      "learning_rate": 0.00019999446646028912,
      "loss": 1.1209,
      "step": 55
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5444576144218445,
      "learning_rate": 0.00019999414574397396,
      "loss": 1.2257,
      "step": 56
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5353034734725952,
      "learning_rate": 0.00019999381599375394,
      "loss": 1.2426,
      "step": 57
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.65821373462677,
      "learning_rate": 0.0001999934772096588,
      "loss": 1.1748,
      "step": 58
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5470272898674011,
      "learning_rate": 0.00019999312939171914,
      "loss": 1.1972,
      "step": 59
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5184686779975891,
      "learning_rate": 0.00019999277253996637,
      "loss": 1.1509,
      "step": 60
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5672779083251953,
      "learning_rate": 0.0001999924066544328,
      "loss": 1.1963,
      "step": 61
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5974107980728149,
      "learning_rate": 0.00019999203173515142,
      "loss": 1.1709,
      "step": 62
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5848557949066162,
      "learning_rate": 0.00019999164778215613,
      "loss": 1.1453,
      "step": 63
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.48387718200683594,
      "learning_rate": 0.00019999125479548165,
      "loss": 0.9851,
      "step": 64
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.44351813197135925,
      "learning_rate": 0.0001999908527751634,
      "loss": 1.105,
      "step": 65
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.500687301158905,
      "learning_rate": 0.00019999044172123775,
      "loss": 1.1271,
      "step": 66
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5383696556091309,
      "learning_rate": 0.00019999002163374188,
      "loss": 1.2929,
      "step": 67
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4801323115825653,
      "learning_rate": 0.00019998959251271367,
      "loss": 0.9591,
      "step": 68
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5076848864555359,
      "learning_rate": 0.0001999891543581919,
      "loss": 1.1723,
      "step": 69
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4907875955104828,
      "learning_rate": 0.0001999887071702162,
      "loss": 1.126,
      "step": 70
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6520359516143799,
      "learning_rate": 0.00019998825094882694,
      "loss": 1.2344,
      "step": 71
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5223540663719177,
      "learning_rate": 0.0001999877856940653,
      "loss": 1.2312,
      "step": 72
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5751238465309143,
      "learning_rate": 0.00019998731140597338,
      "loss": 1.2276,
      "step": 73
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.540924608707428,
      "learning_rate": 0.000199986828084594,
      "loss": 1.1759,
      "step": 74
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4565652310848236,
      "learning_rate": 0.00019998633572997085,
      "loss": 1.0503,
      "step": 75
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46219968795776367,
      "learning_rate": 0.00019998583434214834,
      "loss": 1.1385,
      "step": 76
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5129724740982056,
      "learning_rate": 0.00019998532392117183,
      "loss": 1.1054,
      "step": 77
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.576178252696991,
      "learning_rate": 0.00019998480446708746,
      "loss": 1.2987,
      "step": 78
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5178436040878296,
      "learning_rate": 0.00019998427597994208,
      "loss": 1.0454,
      "step": 79
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5002140998840332,
      "learning_rate": 0.0001999837384597835,
      "loss": 1.1912,
      "step": 80
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5395256280899048,
      "learning_rate": 0.00019998319190666023,
      "loss": 1.1172,
      "step": 81
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5830578207969666,
      "learning_rate": 0.00019998263632062168,
      "loss": 1.15,
      "step": 82
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46478208899497986,
      "learning_rate": 0.00019998207170171805,
      "loss": 1.1597,
      "step": 83
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5071977376937866,
      "learning_rate": 0.00019998149805000034,
      "loss": 1.1847,
      "step": 84
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.455512672662735,
      "learning_rate": 0.00019998091536552035,
      "loss": 1.0309,
      "step": 85
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48724740743637085,
      "learning_rate": 0.00019998032364833076,
      "loss": 1.2756,
      "step": 86
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48692628741264343,
      "learning_rate": 0.00019997972289848503,
      "loss": 1.1126,
      "step": 87
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5381304621696472,
      "learning_rate": 0.00019997911311603737,
      "loss": 1.2491,
      "step": 88
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44811877608299255,
      "learning_rate": 0.00019997849430104298,
      "loss": 1.0797,
      "step": 89
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5321200489997864,
      "learning_rate": 0.0001999778664535577,
      "loss": 1.2419,
      "step": 90
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5130752325057983,
      "learning_rate": 0.00019997722957363823,
      "loss": 1.1859,
      "step": 91
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49227434396743774,
      "learning_rate": 0.00019997658366134217,
      "loss": 1.2368,
      "step": 92
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5907085537910461,
      "learning_rate": 0.0001999759287167278,
      "loss": 1.0933,
      "step": 93
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5108592510223389,
      "learning_rate": 0.00019997526473985438,
      "loss": 1.214,
      "step": 94
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5565717816352844,
      "learning_rate": 0.0001999745917307818,
      "loss": 1.3146,
      "step": 95
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5041968822479248,
      "learning_rate": 0.00019997390968957098,
      "loss": 1.2215,
      "step": 96
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49414822459220886,
      "learning_rate": 0.0001999732186162834,
      "loss": 1.1545,
      "step": 97
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5626671314239502,
      "learning_rate": 0.0001999725185109816,
      "loss": 1.3104,
      "step": 98
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.459627628326416,
      "learning_rate": 0.00019997180937372878,
      "loss": 1.1461,
      "step": 99
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.484120637178421,
      "learning_rate": 0.00019997109120458903,
      "loss": 1.0694,
      "step": 100
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5494018197059631,
      "learning_rate": 0.00019997036400362723,
      "loss": 1.1665,
      "step": 101
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4950861632823944,
      "learning_rate": 0.00019996962777090905,
      "loss": 1.1021,
      "step": 102
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4946760833263397,
      "learning_rate": 0.00019996888250650108,
      "loss": 1.2554,
      "step": 103
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5194969773292542,
      "learning_rate": 0.00019996812821047055,
      "loss": 1.0449,
      "step": 104
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6520335078239441,
      "learning_rate": 0.00019996736488288568,
      "loss": 0.9141,
      "step": 105
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5012494921684265,
      "learning_rate": 0.0001999665925238154,
      "loss": 1.2185,
      "step": 106
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47467753291130066,
      "learning_rate": 0.00019996581113332948,
      "loss": 1.106,
      "step": 107
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45975348353385925,
      "learning_rate": 0.00019996502071149854,
      "loss": 1.1995,
      "step": 108
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5025244355201721,
      "learning_rate": 0.00019996422125839397,
      "loss": 1.0846,
      "step": 109
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5010673999786377,
      "learning_rate": 0.00019996341277408802,
      "loss": 1.1118,
      "step": 110
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5463888645172119,
      "learning_rate": 0.0001999625952586537,
      "loss": 1.3421,
      "step": 111
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4908903241157532,
      "learning_rate": 0.0001999617687121649,
      "loss": 1.2373,
      "step": 112
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4427969753742218,
      "learning_rate": 0.00019996093313469628,
      "loss": 1.0984,
      "step": 113
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45442789793014526,
      "learning_rate": 0.00019996008852632332,
      "loss": 1.1715,
      "step": 114
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4985520541667938,
      "learning_rate": 0.00019995923488712233,
      "loss": 1.0816,
      "step": 115
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5188296437263489,
      "learning_rate": 0.00019995837221717044,
      "loss": 1.1667,
      "step": 116
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.547581136226654,
      "learning_rate": 0.0001999575005165456,
      "loss": 1.1387,
      "step": 117
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5591850876808167,
      "learning_rate": 0.00019995661978532652,
      "loss": 1.2088,
      "step": 118
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5121439695358276,
      "learning_rate": 0.0001999557300235928,
      "loss": 1.2097,
      "step": 119
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4894709885120392,
      "learning_rate": 0.00019995483123142483,
      "loss": 1.1892,
      "step": 120
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5119152665138245,
      "learning_rate": 0.00019995392340890379,
      "loss": 1.257,
      "step": 121
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4950897991657257,
      "learning_rate": 0.00019995300655611172,
      "loss": 1.247,
      "step": 122
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5071575045585632,
      "learning_rate": 0.00019995208067313144,
      "loss": 1.219,
      "step": 123
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48433318734169006,
      "learning_rate": 0.0001999511457600466,
      "loss": 1.1201,
      "step": 124
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.507840096950531,
      "learning_rate": 0.00019995020181694167,
      "loss": 1.0267,
      "step": 125
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.604081392288208,
      "learning_rate": 0.00019994924884390188,
      "loss": 1.0828,
      "step": 126
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5288147926330566,
      "learning_rate": 0.0001999482868410134,
      "loss": 1.1112,
      "step": 127
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5043405294418335,
      "learning_rate": 0.00019994731580836312,
      "loss": 1.1041,
      "step": 128
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5322749614715576,
      "learning_rate": 0.00019994633574603877,
      "loss": 0.8943,
      "step": 129
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5540361404418945,
      "learning_rate": 0.00019994534665412883,
      "loss": 1.0218,
      "step": 130
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5248005390167236,
      "learning_rate": 0.00019994434853272272,
      "loss": 1.0896,
      "step": 131
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5261930227279663,
      "learning_rate": 0.00019994334138191065,
      "loss": 0.9431,
      "step": 132
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5371242761611938,
      "learning_rate": 0.00019994232520178352,
      "loss": 1.2457,
      "step": 133
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4499124586582184,
      "learning_rate": 0.0001999412999924332,
      "loss": 1.1453,
      "step": 134
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4927183985710144,
      "learning_rate": 0.00019994026575395225,
      "loss": 0.9934,
      "step": 135
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4928085505962372,
      "learning_rate": 0.00019993922248643417,
      "loss": 1.1859,
      "step": 136
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5777192711830139,
      "learning_rate": 0.00019993817018997323,
      "loss": 1.124,
      "step": 137
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4621788561344147,
      "learning_rate": 0.00019993710886466443,
      "loss": 1.0674,
      "step": 138
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5149490237236023,
      "learning_rate": 0.00019993603851060367,
      "loss": 1.2451,
      "step": 139
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46570640802383423,
      "learning_rate": 0.0001999349591278877,
      "loss": 1.0202,
      "step": 140
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48668521642684937,
      "learning_rate": 0.000199933870716614,
      "loss": 1.1878,
      "step": 141
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5824366211891174,
      "learning_rate": 0.0001999327732768809,
      "loss": 1.0239,
      "step": 142
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.447409451007843,
      "learning_rate": 0.00019993166680878756,
      "loss": 1.0978,
      "step": 143
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42665061354637146,
      "learning_rate": 0.00019993055131243393,
      "loss": 1.0087,
      "step": 144
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4215356409549713,
      "learning_rate": 0.00019992942678792076,
      "loss": 0.9732,
      "step": 145
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5241891145706177,
      "learning_rate": 0.00019992829323534973,
      "loss": 1.096,
      "step": 146
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.483650803565979,
      "learning_rate": 0.00019992715065482317,
      "loss": 1.1588,
      "step": 147
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5333243608474731,
      "learning_rate": 0.00019992599904644433,
      "loss": 1.0622,
      "step": 148
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4784759283065796,
      "learning_rate": 0.00019992483841031725,
      "loss": 1.1042,
      "step": 149
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47385480999946594,
      "learning_rate": 0.00019992366874654684,
      "loss": 1.1554,
      "step": 150
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.49824684858322144,
      "learning_rate": 0.00019992249005523869,
      "loss": 1.1757,
      "step": 151
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5586369633674622,
      "learning_rate": 0.00019992130233649933,
      "loss": 1.0107,
      "step": 152
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5922762751579285,
      "learning_rate": 0.00019992010559043605,
      "loss": 1.1978,
      "step": 153
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4725801646709442,
      "learning_rate": 0.00019991889981715698,
      "loss": 0.993,
      "step": 154
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5265770554542542,
      "learning_rate": 0.00019991768501677104,
      "loss": 1.0756,
      "step": 155
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5356742143630981,
      "learning_rate": 0.00019991646118938802,
      "loss": 1.1144,
      "step": 156
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45678016543388367,
      "learning_rate": 0.00019991522833511842,
      "loss": 1.0504,
      "step": 157
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48851096630096436,
      "learning_rate": 0.0001999139864540737,
      "loss": 1.2359,
      "step": 158
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4795781373977661,
      "learning_rate": 0.00019991273554636595,
      "loss": 1.1277,
      "step": 159
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4898346960544586,
      "learning_rate": 0.00019991147561210834,
      "loss": 1.1664,
      "step": 160
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5645068883895874,
      "learning_rate": 0.00019991020665141454,
      "loss": 1.2044,
      "step": 161
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.488219290971756,
      "learning_rate": 0.00019990892866439928,
      "loss": 1.1247,
      "step": 162
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5142199397087097,
      "learning_rate": 0.000199907641651178,
      "loss": 1.178,
      "step": 163
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5282805562019348,
      "learning_rate": 0.00019990634561186695,
      "loss": 1.1834,
      "step": 164
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5011789798736572,
      "learning_rate": 0.00019990504054658327,
      "loss": 1.3229,
      "step": 165
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4901934266090393,
      "learning_rate": 0.00019990372645544482,
      "loss": 1.1035,
      "step": 166
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.49366825819015503,
      "learning_rate": 0.00019990240333857035,
      "loss": 1.3065,
      "step": 167
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5546338558197021,
      "learning_rate": 0.00019990107119607938,
      "loss": 1.2996,
      "step": 168
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4404127895832062,
      "learning_rate": 0.00019989973002809225,
      "loss": 1.1766,
      "step": 169
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6145757436752319,
      "learning_rate": 0.00019989837983473016,
      "loss": 1.3385,
      "step": 170
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5417330265045166,
      "learning_rate": 0.00019989702061611506,
      "loss": 1.1634,
      "step": 171
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4397343397140503,
      "learning_rate": 0.00019989565237236978,
      "loss": 1.0259,
      "step": 172
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48129138350486755,
      "learning_rate": 0.0001998942751036179,
      "loss": 1.0549,
      "step": 173
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4494941234588623,
      "learning_rate": 0.00019989288880998388,
      "loss": 0.9936,
      "step": 174
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4411003291606903,
      "learning_rate": 0.00019989149349159294,
      "loss": 1.2235,
      "step": 175
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4981510043144226,
      "learning_rate": 0.00019989008914857116,
      "loss": 1.0567,
      "step": 176
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.43061527609825134,
      "learning_rate": 0.0001998886757810454,
      "loss": 0.8618,
      "step": 177
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4655401408672333,
      "learning_rate": 0.00019988725338914335,
      "loss": 1.2116,
      "step": 178
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.43923428654670715,
      "learning_rate": 0.00019988582197299348,
      "loss": 1.102,
      "step": 179
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5278385877609253,
      "learning_rate": 0.0001998843815327252,
      "loss": 1.1409,
      "step": 180
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47227001190185547,
      "learning_rate": 0.00019988293206846854,
      "loss": 0.9435,
      "step": 181
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.46571439504623413,
      "learning_rate": 0.00019988147358035454,
      "loss": 1.0508,
      "step": 182
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.44509732723236084,
      "learning_rate": 0.0001998800060685149,
      "loss": 1.0575,
      "step": 183
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4418330192565918,
      "learning_rate": 0.00019987852953308227,
      "loss": 0.9238,
      "step": 184
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5942162275314331,
      "learning_rate": 0.00019987704397418998,
      "loss": 1.3404,
      "step": 185
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.49123358726501465,
      "learning_rate": 0.0001998755493919723,
      "loss": 1.2518,
      "step": 186
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48855024576187134,
      "learning_rate": 0.00019987404578656421,
      "loss": 1.1701,
      "step": 187
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.450970321893692,
      "learning_rate": 0.00019987253315810154,
      "loss": 1.0354,
      "step": 188
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4818516671657562,
      "learning_rate": 0.00019987101150672102,
      "loss": 0.9924,
      "step": 189
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5119654536247253,
      "learning_rate": 0.00019986948083256003,
      "loss": 1.157,
      "step": 190
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5616949200630188,
      "learning_rate": 0.00019986794113575695,
      "loss": 1.2006,
      "step": 191
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5410425066947937,
      "learning_rate": 0.00019986639241645083,
      "loss": 1.0805,
      "step": 192
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47965753078460693,
      "learning_rate": 0.00019986483467478158,
      "loss": 1.0842,
      "step": 193
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.45503634214401245,
      "learning_rate": 0.00019986326791088997,
      "loss": 1.1245,
      "step": 194
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5413264632225037,
      "learning_rate": 0.0001998616921249175,
      "loss": 1.3534,
      "step": 195
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.45204904675483704,
      "learning_rate": 0.0001998601073170066,
      "loss": 1.0567,
      "step": 196
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.44439467787742615,
      "learning_rate": 0.0001998585134873004,
      "loss": 1.1653,
      "step": 197
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5485827326774597,
      "learning_rate": 0.00019985691063594287,
      "loss": 1.3116,
      "step": 198
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.466046541929245,
      "learning_rate": 0.0001998552987630789,
      "loss": 0.9963,
      "step": 199
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5031853914260864,
      "learning_rate": 0.00019985367786885404,
      "loss": 1.184,
      "step": 200
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5012284517288208,
      "learning_rate": 0.00019985204795341477,
      "loss": 1.0757,
      "step": 201
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48965683579444885,
      "learning_rate": 0.0001998504090169083,
      "loss": 1.2179,
      "step": 202
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5679877996444702,
      "learning_rate": 0.00019984876105948274,
      "loss": 1.109,
      "step": 203
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.559474766254425,
      "learning_rate": 0.00019984710408128695,
      "loss": 1.2473,
      "step": 204
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4734433591365814,
      "learning_rate": 0.00019984543808247067,
      "loss": 1.1077,
      "step": 205
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4242611229419708,
      "learning_rate": 0.00019984376306318437,
      "loss": 1.1687,
      "step": 206
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48564302921295166,
      "learning_rate": 0.00019984207902357936,
      "loss": 1.1441,
      "step": 207
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4814341068267822,
      "learning_rate": 0.00019984038596380788,
      "loss": 1.0971,
      "step": 208
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.506041407585144,
      "learning_rate": 0.00019983868388402277,
      "loss": 1.1421,
      "step": 209
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4765606224536896,
      "learning_rate": 0.00019983697278437785,
      "loss": 0.9763,
      "step": 210
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4573558568954468,
      "learning_rate": 0.00019983525266502775,
      "loss": 1.0432,
      "step": 211
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49923500418663025,
      "learning_rate": 0.0001998335235261278,
      "loss": 0.9541,
      "step": 212
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8542819619178772,
      "learning_rate": 0.0001998317853678343,
      "loss": 1.2511,
      "step": 213
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48192134499549866,
      "learning_rate": 0.0001998300381903042,
      "loss": 1.037,
      "step": 214
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5000719428062439,
      "learning_rate": 0.00019982828199369541,
      "loss": 1.2035,
      "step": 215
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5127101540565491,
      "learning_rate": 0.00019982651677816657,
      "loss": 1.0139,
      "step": 216
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5559672713279724,
      "learning_rate": 0.00019982474254387716,
      "loss": 1.123,
      "step": 217
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.47846996784210205,
      "learning_rate": 0.00019982295929098744,
      "loss": 1.1112,
      "step": 218
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5220378041267395,
      "learning_rate": 0.00019982116701965853,
      "loss": 1.1898,
      "step": 219
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4617263376712799,
      "learning_rate": 0.0001998193657300524,
      "loss": 1.0178,
      "step": 220
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5010495781898499,
      "learning_rate": 0.00019981755542233177,
      "loss": 1.0739,
      "step": 221
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.47030919790267944,
      "learning_rate": 0.00019981573609666013,
      "loss": 0.943,
      "step": 222
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.48498865962028503,
      "learning_rate": 0.00019981390775320192,
      "loss": 1.0353,
      "step": 223
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.48837897181510925,
      "learning_rate": 0.00019981207039212226,
      "loss": 1.0481,
      "step": 224
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6528398394584656,
      "learning_rate": 0.0001998102240135872,
      "loss": 1.1928,
      "step": 225
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46565020084381104,
      "learning_rate": 0.0001998083686177635,
      "loss": 1.1219,
      "step": 226
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.48760753870010376,
      "learning_rate": 0.00019980650420481884,
      "loss": 1.1069,
      "step": 227
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.45117929577827454,
      "learning_rate": 0.0001998046307749216,
      "loss": 1.1878,
      "step": 228
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5408883690834045,
      "learning_rate": 0.00019980274832824107,
      "loss": 1.2435,
      "step": 229
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5185354351997375,
      "learning_rate": 0.00019980085686494733,
      "loss": 1.1932,
      "step": 230
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4900776445865631,
      "learning_rate": 0.00019979895638521123,
      "loss": 1.2784,
      "step": 231
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47951623797416687,
      "learning_rate": 0.00019979704688920447,
      "loss": 1.1653,
      "step": 232
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47839614748954773,
      "learning_rate": 0.00019979512837709957,
      "loss": 1.1518,
      "step": 233
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5080764293670654,
      "learning_rate": 0.00019979320084906988,
      "loss": 1.2718,
      "step": 234
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4477003812789917,
      "learning_rate": 0.00019979126430528949,
      "loss": 1.1909,
      "step": 235
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5127712488174438,
      "learning_rate": 0.00019978931874593342,
      "loss": 1.0702,
      "step": 236
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4719958007335663,
      "learning_rate": 0.0001997873641711774,
      "loss": 1.1527,
      "step": 237
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.571456789970398,
      "learning_rate": 0.000199785400581198,
      "loss": 1.1834,
      "step": 238
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5032457113265991,
      "learning_rate": 0.00019978342797617266,
      "loss": 1.0674,
      "step": 239
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.48442238569259644,
      "learning_rate": 0.00019978144635627958,
      "loss": 1.0571,
      "step": 240
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4788631498813629,
      "learning_rate": 0.0001997794557216978,
      "loss": 1.1184,
      "step": 241
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.598183274269104,
      "learning_rate": 0.00019977745607260708,
      "loss": 1.0265,
      "step": 242
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5404005646705627,
      "learning_rate": 0.0001997754474091882,
      "loss": 1.2397,
      "step": 243
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4596356749534607,
      "learning_rate": 0.00019977342973162253,
      "loss": 1.1055,
      "step": 244
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5855688452720642,
      "learning_rate": 0.00019977140304009244,
      "loss": 1.2011,
      "step": 245
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5521079897880554,
      "learning_rate": 0.00019976936733478098,
      "loss": 1.2093,
      "step": 246
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.623518168926239,
      "learning_rate": 0.00019976732261587206,
      "loss": 1.1398,
      "step": 247
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4470531940460205,
      "learning_rate": 0.00019976526888355043,
      "loss": 1.0628,
      "step": 248
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5066975951194763,
      "learning_rate": 0.00019976320613800162,
      "loss": 1.0332,
      "step": 249
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5067146420478821,
      "learning_rate": 0.00019976113437941202,
      "loss": 1.2509,
      "step": 250
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.475010484457016,
      "learning_rate": 0.00019975905360796877,
      "loss": 1.1128,
      "step": 251
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5198237895965576,
      "learning_rate": 0.00019975696382385985,
      "loss": 0.9448,
      "step": 252
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.42087745666503906,
      "learning_rate": 0.00019975486502727405,
      "loss": 1.0128,
      "step": 253
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45363667607307434,
      "learning_rate": 0.00019975275721840103,
      "loss": 1.1145,
      "step": 254
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5228469967842102,
      "learning_rate": 0.0001997506403974312,
      "loss": 1.0619,
      "step": 255
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5582649111747742,
      "learning_rate": 0.00019974851456455576,
      "loss": 1.2683,
      "step": 256
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.507866382598877,
      "learning_rate": 0.00019974637971996684,
      "loss": 1.1629,
      "step": 257
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4675002694129944,
      "learning_rate": 0.00019974423586385724,
      "loss": 1.172,
      "step": 258
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.508798360824585,
      "learning_rate": 0.00019974208299642073,
      "loss": 1.2427,
      "step": 259
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5536615252494812,
      "learning_rate": 0.00019973992111785172,
      "loss": 0.9203,
      "step": 260
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5231752991676331,
      "learning_rate": 0.00019973775022834555,
      "loss": 1.0862,
      "step": 261
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45515871047973633,
      "learning_rate": 0.0001997355703280984,
      "loss": 1.1443,
      "step": 262
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6466176509857178,
      "learning_rate": 0.00019973338141730714,
      "loss": 1.3494,
      "step": 263
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4798348844051361,
      "learning_rate": 0.0001997311834961696,
      "loss": 1.065,
      "step": 264
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5107787251472473,
      "learning_rate": 0.00019972897656488428,
      "loss": 1.153,
      "step": 265
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44574907422065735,
      "learning_rate": 0.00019972676062365058,
      "loss": 0.8468,
      "step": 266
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5033632516860962,
      "learning_rate": 0.00019972453567266874,
      "loss": 1.2483,
      "step": 267
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5271995663642883,
      "learning_rate": 0.00019972230171213972,
      "loss": 0.977,
      "step": 268
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.48854905366897583,
      "learning_rate": 0.00019972005874226536,
      "loss": 1.2457,
      "step": 269
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5067196488380432,
      "learning_rate": 0.00019971780676324833,
      "loss": 1.0715,
      "step": 270
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4790126383304596,
      "learning_rate": 0.00019971554577529203,
      "loss": 1.1117,
      "step": 271
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4900168478488922,
      "learning_rate": 0.00019971327577860077,
      "loss": 1.1876,
      "step": 272
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.654327392578125,
      "learning_rate": 0.00019971099677337963,
      "loss": 1.2282,
      "step": 273
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4578112065792084,
      "learning_rate": 0.00019970870875983448,
      "loss": 1.1386,
      "step": 274
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5260718464851379,
      "learning_rate": 0.00019970641173817203,
      "loss": 1.1163,
      "step": 275
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5048907995223999,
      "learning_rate": 0.00019970410570859984,
      "loss": 1.0529,
      "step": 276
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4529321789741516,
      "learning_rate": 0.00019970179067132623,
      "loss": 1.188,
      "step": 277
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.490434467792511,
      "learning_rate": 0.00019969946662656032,
      "loss": 1.127,
      "step": 278
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4846895635128021,
      "learning_rate": 0.00019969713357451208,
      "loss": 1.2213,
      "step": 279
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4657130837440491,
      "learning_rate": 0.00019969479151539236,
      "loss": 1.1398,
      "step": 280
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44968271255493164,
      "learning_rate": 0.00019969244044941266,
      "loss": 1.0703,
      "step": 281
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.49765828251838684,
      "learning_rate": 0.00019969008037678542,
      "loss": 0.955,
      "step": 282
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4985436797142029,
      "learning_rate": 0.0001996877112977239,
      "loss": 1.1547,
      "step": 283
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4842534363269806,
      "learning_rate": 0.00019968533321244208,
      "loss": 1.1838,
      "step": 284
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4974116086959839,
      "learning_rate": 0.0001996829461211548,
      "loss": 1.0741,
      "step": 285
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5230697989463806,
      "learning_rate": 0.00019968055002407776,
      "loss": 1.2932,
      "step": 286
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4365893304347992,
      "learning_rate": 0.00019967814492142738,
      "loss": 1.0626,
      "step": 287
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5909698009490967,
      "learning_rate": 0.00019967573081342103,
      "loss": 1.1187,
      "step": 288
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48496195673942566,
      "learning_rate": 0.00019967330770027673,
      "loss": 1.3553,
      "step": 289
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46772319078445435,
      "learning_rate": 0.0001996708755822134,
      "loss": 0.9889,
      "step": 290
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5091903805732727,
      "learning_rate": 0.00019966843445945088,
      "loss": 1.1878,
      "step": 291
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.500097930431366,
      "learning_rate": 0.00019966598433220957,
      "loss": 1.107,
      "step": 292
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3994102478027344,
      "learning_rate": 0.0001996635252007109,
      "loss": 0.861,
      "step": 293
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5007700324058533,
      "learning_rate": 0.000199661057065177,
      "loss": 1.0658,
      "step": 294
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48105570673942566,
      "learning_rate": 0.0001996585799258309,
      "loss": 1.2075,
      "step": 295
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4781433343887329,
      "learning_rate": 0.00019965609378289637,
      "loss": 1.2312,
      "step": 296
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4458458721637726,
      "learning_rate": 0.000199653598636598,
      "loss": 0.9215,
      "step": 297
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5045272707939148,
      "learning_rate": 0.00019965109448716125,
      "loss": 1.2269,
      "step": 298
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.606421947479248,
      "learning_rate": 0.0001996485813348123,
      "loss": 1.074,
      "step": 299
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4829641580581665,
      "learning_rate": 0.00019964605917977828,
      "loss": 1.0225,
      "step": 300
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5274330377578735,
      "learning_rate": 0.00019964352802228697,
      "loss": 1.1151,
      "step": 301
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5542184710502625,
      "learning_rate": 0.0001996409878625671,
      "loss": 1.1317,
      "step": 302
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5314948558807373,
      "learning_rate": 0.00019963843870084814,
      "loss": 1.004,
      "step": 303
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5083348751068115,
      "learning_rate": 0.0001996358805373604,
      "loss": 1.0661,
      "step": 304
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4306085407733917,
      "learning_rate": 0.00019963331337233498,
      "loss": 0.8742,
      "step": 305
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4597809314727783,
      "learning_rate": 0.00019963073720600383,
      "loss": 1.0761,
      "step": 306
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4719105362892151,
      "learning_rate": 0.00019962815203859968,
      "loss": 0.8865,
      "step": 307
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45550569891929626,
      "learning_rate": 0.00019962555787035607,
      "loss": 0.9979,
      "step": 308
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4660949409008026,
      "learning_rate": 0.00019962295470150743,
      "loss": 1.1004,
      "step": 309
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.48242247104644775,
      "learning_rate": 0.00019962034253228885,
      "loss": 1.0824,
      "step": 310
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5680294036865234,
      "learning_rate": 0.0001996177213629364,
      "loss": 1.2334,
      "step": 311
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.47598588466644287,
      "learning_rate": 0.00019961509119368684,
      "loss": 0.9472,
      "step": 312
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4807590842247009,
      "learning_rate": 0.0001996124520247778,
      "loss": 1.0742,
      "step": 313
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45941096544265747,
      "learning_rate": 0.00019960980385644775,
      "loss": 1.1298,
      "step": 314
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.49606096744537354,
      "learning_rate": 0.0001996071466889359,
      "loss": 1.1194,
      "step": 315
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5158990621566772,
      "learning_rate": 0.00019960448052248233,
      "loss": 1.0912,
      "step": 316
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5971901416778564,
      "learning_rate": 0.00019960180535732788,
      "loss": 1.1296,
      "step": 317
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5912845730781555,
      "learning_rate": 0.00019959912119371426,
      "loss": 1.1678,
      "step": 318
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.47174957394599915,
      "learning_rate": 0.00019959642803188402,
      "loss": 1.1544,
      "step": 319
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5198231935501099,
      "learning_rate": 0.00019959372587208035,
      "loss": 1.1442,
      "step": 320
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.49466976523399353,
      "learning_rate": 0.0001995910147145475,
      "loss": 1.1183,
      "step": 321
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5019547343254089,
      "learning_rate": 0.00019958829455953033,
      "loss": 1.2528,
      "step": 322
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4621056914329529,
      "learning_rate": 0.0001995855654072746,
      "loss": 0.9278,
      "step": 323
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4333496689796448,
      "learning_rate": 0.0001995828272580269,
      "loss": 1.1729,
      "step": 324
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45768311619758606,
      "learning_rate": 0.0001995800801120346,
      "loss": 1.1136,
      "step": 325
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4838506281375885,
      "learning_rate": 0.0001995773239695459,
      "loss": 1.17,
      "step": 326
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5374295115470886,
      "learning_rate": 0.00019957455883080978,
      "loss": 1.2547,
      "step": 327
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4597136378288269,
      "learning_rate": 0.00019957178469607602,
      "loss": 1.1141,
      "step": 328
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4528232514858246,
      "learning_rate": 0.00019956900156559534,
      "loss": 1.0884,
      "step": 329
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4871821403503418,
      "learning_rate": 0.00019956620943961908,
      "loss": 1.2074,
      "step": 330
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4409606456756592,
      "learning_rate": 0.00019956340831839956,
      "loss": 0.9346,
      "step": 331
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.498367041349411,
      "learning_rate": 0.00019956059820218982,
      "loss": 1.1141,
      "step": 332
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5050720572471619,
      "learning_rate": 0.00019955777909124372,
      "loss": 1.2472,
      "step": 333
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5095576047897339,
      "learning_rate": 0.00019955495098581604,
      "loss": 0.9069,
      "step": 334
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5386484265327454,
      "learning_rate": 0.00019955211388616214,
      "loss": 1.087,
      "step": 335
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6390095353126526,
      "learning_rate": 0.00019954926779253848,
      "loss": 1.3301,
      "step": 336
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5223362445831299,
      "learning_rate": 0.00019954641270520206,
      "loss": 1.3201,
      "step": 337
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5296879410743713,
      "learning_rate": 0.00019954354862441093,
      "loss": 1.2484,
      "step": 338
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4566580653190613,
      "learning_rate": 0.0001995406755504238,
      "loss": 0.9785,
      "step": 339
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5283248424530029,
      "learning_rate": 0.0001995377934835002,
      "loss": 1.1248,
      "step": 340
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4475520849227905,
      "learning_rate": 0.00019953490242390057,
      "loss": 1.0337,
      "step": 341
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5099059343338013,
      "learning_rate": 0.00019953200237188604,
      "loss": 1.059,
      "step": 342
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4882862865924835,
      "learning_rate": 0.00019952909332771865,
      "loss": 1.1237,
      "step": 343
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4969572424888611,
      "learning_rate": 0.0001995261752916612,
      "loss": 1.194,
      "step": 344
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4979659914970398,
      "learning_rate": 0.00019952324826397736,
      "loss": 1.22,
      "step": 345
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5434485077857971,
      "learning_rate": 0.00019952031224493153,
      "loss": 1.3075,
      "step": 346
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4933587908744812,
      "learning_rate": 0.00019951736723478896,
      "loss": 1.0579,
      "step": 347
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4828581213951111,
      "learning_rate": 0.0001995144132338157,
      "loss": 1.2079,
      "step": 348
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5049371123313904,
      "learning_rate": 0.00019951145024227872,
      "loss": 1.115,
      "step": 349
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.47797635197639465,
      "learning_rate": 0.0001995084782604456,
      "loss": 0.9859,
      "step": 350
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5296621322631836,
      "learning_rate": 0.0001995054972885849,
      "loss": 1.0991,
      "step": 351
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4989735186100006,
      "learning_rate": 0.0001995025073269659,
      "loss": 1.0342,
      "step": 352
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5142637491226196,
      "learning_rate": 0.00019949950837585875,
      "loss": 1.2478,
      "step": 353
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.48040661215782166,
      "learning_rate": 0.0001994965004355344,
      "loss": 1.0811,
      "step": 354
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4938506484031677,
      "learning_rate": 0.00019949348350626456,
      "loss": 1.279,
      "step": 355
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5298872590065002,
      "learning_rate": 0.00019949045758832186,
      "loss": 1.0503,
      "step": 356
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5765920281410217,
      "learning_rate": 0.00019948742268197957,
      "loss": 1.2411,
      "step": 357
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5472105145454407,
      "learning_rate": 0.00019948437878751198,
      "loss": 1.296,
      "step": 358
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5123063921928406,
      "learning_rate": 0.00019948132590519402,
      "loss": 1.0406,
      "step": 359
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.512622058391571,
      "learning_rate": 0.00019947826403530155,
      "loss": 1.0446,
      "step": 360
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5731431841850281,
      "learning_rate": 0.00019947519317811113,
      "loss": 1.1033,
      "step": 361
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5514195561408997,
      "learning_rate": 0.00019947211333390028,
      "loss": 1.2534,
      "step": 362
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.48427775502204895,
      "learning_rate": 0.00019946902450294717,
      "loss": 1.2724,
      "step": 363
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4817625880241394,
      "learning_rate": 0.0001994659266855309,
      "loss": 0.935,
      "step": 364
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5593578219413757,
      "learning_rate": 0.00019946281988193133,
      "loss": 1.1755,
      "step": 365
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.48790112137794495,
      "learning_rate": 0.00019945970409242914,
      "loss": 1.2172,
      "step": 366
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5469039082527161,
      "learning_rate": 0.0001994565793173058,
      "loss": 1.2309,
      "step": 367
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5210196375846863,
      "learning_rate": 0.00019945344555684366,
      "loss": 1.2343,
      "step": 368
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5936497449874878,
      "learning_rate": 0.00019945030281132584,
      "loss": 1.3347,
      "step": 369
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5588001608848572,
      "learning_rate": 0.00019944715108103622,
      "loss": 1.226,
      "step": 370
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4977017343044281,
      "learning_rate": 0.0001994439903662596,
      "loss": 1.1088,
      "step": 371
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5470244884490967,
      "learning_rate": 0.00019944082066728145,
      "loss": 1.2318,
      "step": 372
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.48215651512145996,
      "learning_rate": 0.0001994376419843882,
      "loss": 1.0203,
      "step": 373
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.473414808511734,
      "learning_rate": 0.00019943445431786706,
      "loss": 0.9304,
      "step": 374
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.47687965631484985,
      "learning_rate": 0.00019943125766800593,
      "loss": 1.1128,
      "step": 375
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5194618105888367,
      "learning_rate": 0.00019942805203509363,
      "loss": 1.0648,
      "step": 376
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6038587689399719,
      "learning_rate": 0.0001994248374194198,
      "loss": 0.9894,
      "step": 377
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6201353073120117,
      "learning_rate": 0.00019942161382127487,
      "loss": 1.1983,
      "step": 378
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5684635639190674,
      "learning_rate": 0.00019941838124095003,
      "loss": 1.1072,
      "step": 379
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.518798828125,
      "learning_rate": 0.00019941513967873737,
      "loss": 1.0759,
      "step": 380
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5163238644599915,
      "learning_rate": 0.00019941188913492972,
      "loss": 1.1764,
      "step": 381
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.47596409916877747,
      "learning_rate": 0.00019940862960982075,
      "loss": 1.0686,
      "step": 382
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5515662431716919,
      "learning_rate": 0.00019940536110370495,
      "loss": 1.274,
      "step": 383
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4495806396007538,
      "learning_rate": 0.0001994020836168776,
      "loss": 0.9764,
      "step": 384
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5222442150115967,
      "learning_rate": 0.00019939879714963484,
      "loss": 1.108,
      "step": 385
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.47913652658462524,
      "learning_rate": 0.0001993955017022735,
      "loss": 1.1575,
      "step": 386
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4735942482948303,
      "learning_rate": 0.0001993921972750914,
      "loss": 1.1684,
      "step": 387
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44809603691101074,
      "learning_rate": 0.000199388883868387,
      "loss": 1.0301,
      "step": 388
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5486351847648621,
      "learning_rate": 0.0001993855614824597,
      "loss": 1.2701,
      "step": 389
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4721758961677551,
      "learning_rate": 0.00019938223011760963,
      "loss": 1.1417,
      "step": 390
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.561992883682251,
      "learning_rate": 0.0001993788897741378,
      "loss": 1.1319,
      "step": 391
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5158048272132874,
      "learning_rate": 0.00019937554045234594,
      "loss": 0.9899,
      "step": 392
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5148165225982666,
      "learning_rate": 0.00019937218215253667,
      "loss": 1.2607,
      "step": 393
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4346942901611328,
      "learning_rate": 0.00019936881487501337,
      "loss": 0.8729,
      "step": 394
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5264867544174194,
      "learning_rate": 0.0001993654386200803,
      "loss": 1.0844,
      "step": 395
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5341372489929199,
      "learning_rate": 0.0001993620533880424,
      "loss": 1.055,
      "step": 396
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5435947775840759,
      "learning_rate": 0.0001993586591792056,
      "loss": 1.109,
      "step": 397
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4812977612018585,
      "learning_rate": 0.0001993552559938765,
      "loss": 1.1092,
      "step": 398
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5616835951805115,
      "learning_rate": 0.00019935184383236258,
      "loss": 1.1535,
      "step": 399
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4366762638092041,
      "learning_rate": 0.00019934842269497209,
      "loss": 0.9751,
      "step": 400
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5996217131614685,
      "learning_rate": 0.00019934499258201415,
      "loss": 1.2012,
      "step": 401
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.49099141359329224,
      "learning_rate": 0.00019934155349379858,
      "loss": 1.0873,
      "step": 402
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4674723744392395,
      "learning_rate": 0.00019933810543063617,
      "loss": 1.1367,
      "step": 403
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5626590847969055,
      "learning_rate": 0.00019933464839283832,
      "loss": 1.3648,
      "step": 404
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5492469072341919,
      "learning_rate": 0.00019933118238071743,
      "loss": 1.249,
      "step": 405
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5024725198745728,
      "learning_rate": 0.00019932770739458668,
      "loss": 1.0049,
      "step": 406
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.48954451084136963,
      "learning_rate": 0.00019932422343475992,
      "loss": 1.1437,
      "step": 407
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5946091413497925,
      "learning_rate": 0.00019932073050155194,
      "loss": 1.2738,
      "step": 408
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.504777729511261,
      "learning_rate": 0.00019931722859527833,
      "loss": 1.2256,
      "step": 409
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5623805522918701,
      "learning_rate": 0.00019931371771625544,
      "loss": 1.0886,
      "step": 410
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.545048177242279,
      "learning_rate": 0.00019931019786480046,
      "loss": 1.0154,
      "step": 411
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4837050139904022,
      "learning_rate": 0.00019930666904123138,
      "loss": 1.1486,
      "step": 412
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5394037365913391,
      "learning_rate": 0.00019930313124586705,
      "loss": 1.211,
      "step": 413
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45422598719596863,
      "learning_rate": 0.00019929958447902708,
      "loss": 1.1847,
      "step": 414
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5187952518463135,
      "learning_rate": 0.00019929602874103186,
      "loss": 1.1617,
      "step": 415
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5281522870063782,
      "learning_rate": 0.00019929246403220267,
      "loss": 1.1458,
      "step": 416
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5519776344299316,
      "learning_rate": 0.00019928889035286153,
      "loss": 1.0696,
      "step": 417
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4418583810329437,
      "learning_rate": 0.00019928530770333135,
      "loss": 0.8771,
      "step": 418
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4728032648563385,
      "learning_rate": 0.00019928171608393574,
      "loss": 0.9559,
      "step": 419
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.49322378635406494,
      "learning_rate": 0.0001992781154949992,
      "loss": 1.1025,
      "step": 420
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5575322508811951,
      "learning_rate": 0.00019927450593684707,
      "loss": 1.2714,
      "step": 421
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46672430634498596,
      "learning_rate": 0.0001992708874098054,
      "loss": 1.1076,
      "step": 422
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5349050164222717,
      "learning_rate": 0.00019926725991420115,
      "loss": 1.1568,
      "step": 423
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6394595503807068,
      "learning_rate": 0.00019926362345036198,
      "loss": 1.1381,
      "step": 424
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5520142316818237,
      "learning_rate": 0.00019925997801861648,
      "loss": 0.9789,
      "step": 425
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.48990902304649353,
      "learning_rate": 0.000199256323619294,
      "loss": 1.0684,
      "step": 426
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4744899570941925,
      "learning_rate": 0.00019925266025272461,
      "loss": 1.0202,
      "step": 427
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5531217455863953,
      "learning_rate": 0.00019924898791923935,
      "loss": 1.135,
      "step": 428
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4650912880897522,
      "learning_rate": 0.00019924530661917,
      "loss": 1.3101,
      "step": 429
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4258342385292053,
      "learning_rate": 0.00019924161635284912,
      "loss": 1.0714,
      "step": 430
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4997932016849518,
      "learning_rate": 0.00019923791712061012,
      "loss": 1.1918,
      "step": 431
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4654412269592285,
      "learning_rate": 0.00019923420892278718,
      "loss": 0.8791,
      "step": 432
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5107722282409668,
      "learning_rate": 0.0001992304917597153,
      "loss": 1.199,
      "step": 433
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5670316815376282,
      "learning_rate": 0.00019922676563173037,
      "loss": 1.1498,
      "step": 434
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5506672263145447,
      "learning_rate": 0.00019922303053916896,
      "loss": 1.0911,
      "step": 435
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5242930054664612,
      "learning_rate": 0.00019921928648236853,
      "loss": 1.0317,
      "step": 436
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5948449373245239,
      "learning_rate": 0.00019921553346166737,
      "loss": 1.2627,
      "step": 437
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5378965139389038,
      "learning_rate": 0.0001992117714774045,
      "loss": 1.1651,
      "step": 438
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.614194929599762,
      "learning_rate": 0.00019920800052991984,
      "loss": 1.1672,
      "step": 439
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5187473893165588,
      "learning_rate": 0.00019920422061955404,
      "loss": 1.1762,
      "step": 440
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47316837310791016,
      "learning_rate": 0.00019920043174664857,
      "loss": 1.0654,
      "step": 441
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5558268427848816,
      "learning_rate": 0.00019919663391154577,
      "loss": 1.0374,
      "step": 442
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5064995884895325,
      "learning_rate": 0.0001991928271145888,
      "loss": 1.205,
      "step": 443
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5428379774093628,
      "learning_rate": 0.00019918901135612145,
      "loss": 1.0972,
      "step": 444
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.500322163105011,
      "learning_rate": 0.00019918518663648857,
      "loss": 1.1213,
      "step": 445
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47069665789604187,
      "learning_rate": 0.00019918135295603567,
      "loss": 1.0122,
      "step": 446
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5628748536109924,
      "learning_rate": 0.00019917751031510908,
      "loss": 1.1794,
      "step": 447
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47065627574920654,
      "learning_rate": 0.00019917365871405598,
      "loss": 1.0724,
      "step": 448
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5735258460044861,
      "learning_rate": 0.00019916979815322433,
      "loss": 1.0173,
      "step": 449
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6087215542793274,
      "learning_rate": 0.00019916592863296293,
      "loss": 1.1557,
      "step": 450
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4948890209197998,
      "learning_rate": 0.00019916205015362132,
      "loss": 1.1302,
      "step": 451
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5734955668449402,
      "learning_rate": 0.00019915816271554997,
      "loss": 1.0741,
      "step": 452
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5332045555114746,
      "learning_rate": 0.00019915426631910004,
      "loss": 1.1055,
      "step": 453
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5224483609199524,
      "learning_rate": 0.00019915036096462355,
      "loss": 1.1148,
      "step": 454
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5145097374916077,
      "learning_rate": 0.00019914644665247332,
      "loss": 1.1198,
      "step": 455
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5348233580589294,
      "learning_rate": 0.000199142523383003,
      "loss": 1.2597,
      "step": 456
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5970399975776672,
      "learning_rate": 0.00019913859115656707,
      "loss": 1.3203,
      "step": 457
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.514270544052124,
      "learning_rate": 0.00019913464997352073,
      "loss": 1.2183,
      "step": 458
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5271024107933044,
      "learning_rate": 0.0001991306998342201,
      "loss": 0.9667,
      "step": 459
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.507999837398529,
      "learning_rate": 0.00019912674073902195,
      "loss": 1.1014,
      "step": 460
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.48636719584465027,
      "learning_rate": 0.00019912277268828408,
      "loss": 1.1854,
      "step": 461
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5034084916114807,
      "learning_rate": 0.0001991187956823649,
      "loss": 0.9449,
      "step": 462
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5856225490570068,
      "learning_rate": 0.00019911480972162378,
      "loss": 1.2222,
      "step": 463
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.516998827457428,
      "learning_rate": 0.00019911081480642072,
      "loss": 1.0252,
      "step": 464
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4869126081466675,
      "learning_rate": 0.00019910681093711675,
      "loss": 1.1473,
      "step": 465
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5495402812957764,
      "learning_rate": 0.00019910279811407357,
      "loss": 1.131,
      "step": 466
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4950717091560364,
      "learning_rate": 0.00019909877633765368,
      "loss": 1.0872,
      "step": 467
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.609627366065979,
      "learning_rate": 0.00019909474560822043,
      "loss": 1.1313,
      "step": 468
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4924461841583252,
      "learning_rate": 0.00019909070592613799,
      "loss": 1.1424,
      "step": 469
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.48707911372184753,
      "learning_rate": 0.00019908665729177134,
      "loss": 1.1168,
      "step": 470
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5481781363487244,
      "learning_rate": 0.00019908259970548623,
      "loss": 1.2636,
      "step": 471
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4931468963623047,
      "learning_rate": 0.0001990785331676492,
      "loss": 0.9433,
      "step": 472
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5407423377037048,
      "learning_rate": 0.00019907445767862773,
      "loss": 1.0628,
      "step": 473
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4662988483905792,
      "learning_rate": 0.00019907037323878995,
      "loss": 1.0792,
      "step": 474
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4906497299671173,
      "learning_rate": 0.00019906627984850488,
      "loss": 1.0281,
      "step": 475
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4651735723018646,
      "learning_rate": 0.00019906217750814234,
      "loss": 0.9659,
      "step": 476
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.485735684633255,
      "learning_rate": 0.00019905806621807294,
      "loss": 1.1931,
      "step": 477
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4991046190261841,
      "learning_rate": 0.00019905394597866813,
      "loss": 1.2245,
      "step": 478
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5731810927391052,
      "learning_rate": 0.00019904981679030016,
      "loss": 1.185,
      "step": 479
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5500648021697998,
      "learning_rate": 0.00019904567865334204,
      "loss": 1.0929,
      "step": 480
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.47128331661224365,
      "learning_rate": 0.00019904153156816764,
      "loss": 0.7726,
      "step": 481
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5505924820899963,
      "learning_rate": 0.00019903737553515164,
      "loss": 1.1731,
      "step": 482
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5098860263824463,
      "learning_rate": 0.00019903321055466953,
      "loss": 1.0776,
      "step": 483
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.54964280128479,
      "learning_rate": 0.00019902903662709755,
      "loss": 1.1415,
      "step": 484
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5043061971664429,
      "learning_rate": 0.00019902485375281281,
      "loss": 1.0319,
      "step": 485
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5406997203826904,
      "learning_rate": 0.00019902066193219322,
      "loss": 1.1259,
      "step": 486
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6080590486526489,
      "learning_rate": 0.0001990164611656175,
      "loss": 1.0707,
      "step": 487
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.46147873997688293,
      "learning_rate": 0.0001990122514534651,
      "loss": 0.8235,
      "step": 488
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5430011749267578,
      "learning_rate": 0.0001990080327961164,
      "loss": 1.1288,
      "step": 489
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4875696003437042,
      "learning_rate": 0.00019900380519395253,
      "loss": 1.0249,
      "step": 490
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.46323084831237793,
      "learning_rate": 0.00019899956864735544,
      "loss": 0.9812,
      "step": 491
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4611752927303314,
      "learning_rate": 0.00019899532315670782,
      "loss": 0.8467,
      "step": 492
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5283900499343872,
      "learning_rate": 0.00019899106872239326,
      "loss": 1.2849,
      "step": 493
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.47522106766700745,
      "learning_rate": 0.00019898680534479618,
      "loss": 0.9322,
      "step": 494
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4241219758987427,
      "learning_rate": 0.00019898253302430164,
      "loss": 1.1607,
      "step": 495
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5026375651359558,
      "learning_rate": 0.0001989782517612957,
      "loss": 1.1405,
      "step": 496
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5204796195030212,
      "learning_rate": 0.00019897396155616512,
      "loss": 1.1251,
      "step": 497
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.612561821937561,
      "learning_rate": 0.0001989696624092975,
      "loss": 0.9414,
      "step": 498
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6452274322509766,
      "learning_rate": 0.00019896535432108127,
      "loss": 1.1394,
      "step": 499
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5177410244941711,
      "learning_rate": 0.00019896103729190562,
      "loss": 1.1647,
      "step": 500
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5609557628631592,
      "learning_rate": 0.0001989567113221605,
      "loss": 1.1083,
      "step": 501
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4865688383579254,
      "learning_rate": 0.00019895237641223687,
      "loss": 0.8964,
      "step": 502
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5484577417373657,
      "learning_rate": 0.00019894803256252628,
      "loss": 1.1321,
      "step": 503
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5816463232040405,
      "learning_rate": 0.0001989436797734212,
      "loss": 1.1356,
      "step": 504
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5487164855003357,
      "learning_rate": 0.00019893931804531488,
      "loss": 1.3929,
      "step": 505
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5809823870658875,
      "learning_rate": 0.00019893494737860135,
      "loss": 1.2339,
      "step": 506
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.601434051990509,
      "learning_rate": 0.00019893056777367548,
      "loss": 1.1482,
      "step": 507
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.48192596435546875,
      "learning_rate": 0.00019892617923093298,
      "loss": 1.0692,
      "step": 508
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6076588034629822,
      "learning_rate": 0.00019892178175077028,
      "loss": 1.229,
      "step": 509
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5927091836929321,
      "learning_rate": 0.0001989173753335847,
      "loss": 1.0472,
      "step": 510
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.48782506585121155,
      "learning_rate": 0.00019891295997977434,
      "loss": 1.1462,
      "step": 511
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.48741236329078674,
      "learning_rate": 0.00019890853568973807,
      "loss": 1.1603,
      "step": 512
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5580458045005798,
      "learning_rate": 0.00019890410246387562,
      "loss": 1.1697,
      "step": 513
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5367544889450073,
      "learning_rate": 0.00019889966030258753,
      "loss": 1.118,
      "step": 514
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4435345232486725,
      "learning_rate": 0.00019889520920627506,
      "loss": 1.0305,
      "step": 515
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5683806538581848,
      "learning_rate": 0.0001988907491753404,
      "loss": 1.0101,
      "step": 516
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5692442059516907,
      "learning_rate": 0.00019888628021018652,
      "loss": 1.0924,
      "step": 517
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5492920279502869,
      "learning_rate": 0.00019888180231121707,
      "loss": 1.1479,
      "step": 518
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5364216566085815,
      "learning_rate": 0.00019887731547883663,
      "loss": 1.0438,
      "step": 519
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4900088310241699,
      "learning_rate": 0.0001988728197134506,
      "loss": 1.1337,
      "step": 520
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5172410011291504,
      "learning_rate": 0.00019886831501546512,
      "loss": 1.2197,
      "step": 521
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5062763690948486,
      "learning_rate": 0.00019886380138528717,
      "loss": 1.1406,
      "step": 522
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4935275912284851,
      "learning_rate": 0.00019885927882332455,
      "loss": 1.1764,
      "step": 523
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5174946188926697,
      "learning_rate": 0.0001988547473299858,
      "loss": 1.131,
      "step": 524
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.45522478222846985,
      "learning_rate": 0.00019885020690568034,
      "loss": 0.9077,
      "step": 525
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5257841944694519,
      "learning_rate": 0.0001988456575508184,
      "loss": 0.9448,
      "step": 526
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.484298974275589,
      "learning_rate": 0.00019884109926581096,
      "loss": 1.1791,
      "step": 527
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.536929190158844,
      "learning_rate": 0.00019883653205106985,
      "loss": 1.2124,
      "step": 528
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5551151037216187,
      "learning_rate": 0.00019883195590700764,
      "loss": 1.1383,
      "step": 529
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.49238866567611694,
      "learning_rate": 0.00019882737083403785,
      "loss": 1.0841,
      "step": 530
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.49737808108329773,
      "learning_rate": 0.0001988227768325746,
      "loss": 1.0843,
      "step": 531
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.529952883720398,
      "learning_rate": 0.00019881817390303303,
      "loss": 1.1796,
      "step": 532
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5074145197868347,
      "learning_rate": 0.00019881356204582895,
      "loss": 1.1394,
      "step": 533
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.553284227848053,
      "learning_rate": 0.00019880894126137903,
      "loss": 1.1341,
      "step": 534
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4974454641342163,
      "learning_rate": 0.00019880431155010073,
      "loss": 1.0177,
      "step": 535
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5199590921401978,
      "learning_rate": 0.00019879967291241228,
      "loss": 1.2168,
      "step": 536
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5003868341445923,
      "learning_rate": 0.00019879502534873283,
      "loss": 1.0574,
      "step": 537
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.49978846311569214,
      "learning_rate": 0.00019879036885948217,
      "loss": 1.268,
      "step": 538
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5444263219833374,
      "learning_rate": 0.00019878570344508106,
      "loss": 1.147,
      "step": 539
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.540688693523407,
      "learning_rate": 0.00019878102910595095,
      "loss": 1.2374,
      "step": 540
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4839583933353424,
      "learning_rate": 0.00019877634584251417,
      "loss": 0.9215,
      "step": 541
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5708199143409729,
      "learning_rate": 0.00019877165365519379,
      "loss": 1.2743,
      "step": 542
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.496097594499588,
      "learning_rate": 0.00019876695254441378,
      "loss": 1.1455,
      "step": 543
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4960477948188782,
      "learning_rate": 0.0001987622425105988,
      "loss": 1.0126,
      "step": 544
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5258330702781677,
      "learning_rate": 0.00019875752355417441,
      "loss": 1.1997,
      "step": 545
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5457305312156677,
      "learning_rate": 0.00019875279567556692,
      "loss": 1.2039,
      "step": 546
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5998676419258118,
      "learning_rate": 0.0001987480588752035,
      "loss": 0.9432,
      "step": 547
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5208805799484253,
      "learning_rate": 0.00019874331315351205,
      "loss": 1.0473,
      "step": 548
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6455830335617065,
      "learning_rate": 0.00019873855851092133,
      "loss": 1.0942,
      "step": 549
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5795512199401855,
      "learning_rate": 0.00019873379494786094,
      "loss": 1.1694,
      "step": 550
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5360243320465088,
      "learning_rate": 0.0001987290224647612,
      "loss": 0.9945,
      "step": 551
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.48501020669937134,
      "learning_rate": 0.00019872424106205324,
      "loss": 1.1037,
      "step": 552
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5583041310310364,
      "learning_rate": 0.0001987194507401691,
      "loss": 0.9302,
      "step": 553
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.539480447769165,
      "learning_rate": 0.00019871465149954155,
      "loss": 1.0732,
      "step": 554
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4358266294002533,
      "learning_rate": 0.00019870984334060415,
      "loss": 0.8591,
      "step": 555
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5443989634513855,
      "learning_rate": 0.00019870502626379127,
      "loss": 1.0836,
      "step": 556
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5532322525978088,
      "learning_rate": 0.00019870020026953818,
      "loss": 1.0821,
      "step": 557
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5223169922828674,
      "learning_rate": 0.00019869536535828083,
      "loss": 1.0802,
      "step": 558
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4703574478626251,
      "learning_rate": 0.00019869052153045598,
      "loss": 0.9764,
      "step": 559
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.542907178401947,
      "learning_rate": 0.00019868566878650134,
      "loss": 1.2281,
      "step": 560
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5914562344551086,
      "learning_rate": 0.00019868080712685528,
      "loss": 1.2142,
      "step": 561
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5440067052841187,
      "learning_rate": 0.000198675936551957,
      "loss": 1.177,
      "step": 562
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4753040075302124,
      "learning_rate": 0.0001986710570622466,
      "loss": 1.1859,
      "step": 563
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4693201780319214,
      "learning_rate": 0.00019866616865816484,
      "loss": 0.9495,
      "step": 564
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.48725131154060364,
      "learning_rate": 0.00019866127134015335,
      "loss": 1.1822,
      "step": 565
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4840669631958008,
      "learning_rate": 0.00019865636510865467,
      "loss": 1.0614,
      "step": 566
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4818721413612366,
      "learning_rate": 0.00019865144996411197,
      "loss": 1.0566,
      "step": 567
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5293144583702087,
      "learning_rate": 0.00019864652590696934,
      "loss": 1.1449,
      "step": 568
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5687733292579651,
      "learning_rate": 0.00019864159293767162,
      "loss": 1.2512,
      "step": 569
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5350620150566101,
      "learning_rate": 0.00019863665105666447,
      "loss": 1.176,
      "step": 570
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5288693904876709,
      "learning_rate": 0.0001986317002643944,
      "loss": 0.9136,
      "step": 571
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4852888584136963,
      "learning_rate": 0.00019862674056130865,
      "loss": 1.0506,
      "step": 572
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5876291394233704,
      "learning_rate": 0.0001986217719478553,
      "loss": 1.2772,
      "step": 573
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5367861390113831,
      "learning_rate": 0.00019861679442448327,
      "loss": 1.1672,
      "step": 574
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5490912199020386,
      "learning_rate": 0.00019861180799164222,
      "loss": 1.0972,
      "step": 575
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5793784260749817,
      "learning_rate": 0.0001986068126497827,
      "loss": 1.1316,
      "step": 576
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.502665638923645,
      "learning_rate": 0.0001986018083993559,
      "loss": 1.1683,
      "step": 577
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5572852492332458,
      "learning_rate": 0.00019859679524081406,
      "loss": 1.3337,
      "step": 578
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.49824976921081543,
      "learning_rate": 0.00019859177317460994,
      "loss": 1.2149,
      "step": 579
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5626625418663025,
      "learning_rate": 0.00019858674220119742,
      "loss": 1.1765,
      "step": 580
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5086746215820312,
      "learning_rate": 0.00019858170232103087,
      "loss": 1.293,
      "step": 581
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5596989989280701,
      "learning_rate": 0.00019857665353456577,
      "loss": 1.084,
      "step": 582
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5050382018089294,
      "learning_rate": 0.00019857159584225808,
      "loss": 1.0266,
      "step": 583
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4970535635948181,
      "learning_rate": 0.00019856652924456488,
      "loss": 1.2084,
      "step": 584
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5188227891921997,
      "learning_rate": 0.0001985614537419438,
      "loss": 1.2503,
      "step": 585
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5552204251289368,
      "learning_rate": 0.00019855636933485343,
      "loss": 1.2152,
      "step": 586
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7056576013565063,
      "learning_rate": 0.0001985512760237531,
      "loss": 1.0611,
      "step": 587
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.499446839094162,
      "learning_rate": 0.00019854617380910303,
      "loss": 1.0997,
      "step": 588
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6023534536361694,
      "learning_rate": 0.0001985410626913641,
      "loss": 1.102,
      "step": 589
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5384407043457031,
      "learning_rate": 0.0001985359426709981,
      "loss": 1.2433,
      "step": 590
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5116797685623169,
      "learning_rate": 0.00019853081374846755,
      "loss": 0.9768,
      "step": 591
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5307533144950867,
      "learning_rate": 0.0001985256759242359,
      "loss": 1.1518,
      "step": 592
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4904177784919739,
      "learning_rate": 0.00019852052919876726,
      "loss": 1.0781,
      "step": 593
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.48693302273750305,
      "learning_rate": 0.00019851537357252667,
      "loss": 1.1335,
      "step": 594
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.557604193687439,
      "learning_rate": 0.00019851020904597982,
      "loss": 1.0688,
      "step": 595
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5276851058006287,
      "learning_rate": 0.00019850503561959335,
      "loss": 1.0065,
      "step": 596
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.46584829688072205,
      "learning_rate": 0.00019849985329383466,
      "loss": 1.0136,
      "step": 597
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5401173830032349,
      "learning_rate": 0.00019849466206917193,
      "loss": 1.1503,
      "step": 598
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5231667160987854,
      "learning_rate": 0.00019848946194607416,
      "loss": 1.079,
      "step": 599
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.46920567750930786,
      "learning_rate": 0.00019848425292501114,
      "loss": 1.0505,
      "step": 600
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4620935022830963,
      "learning_rate": 0.00019847903500645351,
      "loss": 1.1187,
      "step": 601
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5359715223312378,
      "learning_rate": 0.00019847380819087263,
      "loss": 0.9545,
      "step": 602
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5216434001922607,
      "learning_rate": 0.00019846857247874072,
      "loss": 0.958,
      "step": 603
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4918813109397888,
      "learning_rate": 0.00019846332787053084,
      "loss": 1.1468,
      "step": 604
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5258176922798157,
      "learning_rate": 0.0001984580743667168,
      "loss": 0.9893,
      "step": 605
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5433478355407715,
      "learning_rate": 0.00019845281196777319,
      "loss": 1.365,
      "step": 606
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5165151357650757,
      "learning_rate": 0.00019844754067417544,
      "loss": 0.9692,
      "step": 607
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.562602698802948,
      "learning_rate": 0.00019844226048639982,
      "loss": 1.0073,
      "step": 608
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.47087547183036804,
      "learning_rate": 0.00019843697140492332,
      "loss": 1.178,
      "step": 609
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.633965790271759,
      "learning_rate": 0.00019843167343022383,
      "loss": 1.0784,
      "step": 610
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5291125774383545,
      "learning_rate": 0.00019842636656277994,
      "loss": 1.164,
      "step": 611
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.46963217854499817,
      "learning_rate": 0.00019842105080307117,
      "loss": 1.1386,
      "step": 612
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4947594404220581,
      "learning_rate": 0.0001984157261515777,
      "loss": 1.1135,
      "step": 613
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.47310054302215576,
      "learning_rate": 0.00019841039260878055,
      "loss": 1.0398,
      "step": 614
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5538829565048218,
      "learning_rate": 0.00019840505017516166,
      "loss": 1.31,
      "step": 615
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5245307683944702,
      "learning_rate": 0.00019839969885120367,
      "loss": 1.1665,
      "step": 616
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.500907301902771,
      "learning_rate": 0.00019839433863739,
      "loss": 1.0136,
      "step": 617
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5003906488418579,
      "learning_rate": 0.00019838896953420493,
      "loss": 1.2467,
      "step": 618
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48780718445777893,
      "learning_rate": 0.00019838359154213358,
      "loss": 0.8645,
      "step": 619
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.520158588886261,
      "learning_rate": 0.00019837820466166175,
      "loss": 0.8612,
      "step": 620
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5978065133094788,
      "learning_rate": 0.0001983728088932762,
      "loss": 1.1102,
      "step": 621
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5158056020736694,
      "learning_rate": 0.00019836740423746427,
      "loss": 1.0781,
      "step": 622
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5026460886001587,
      "learning_rate": 0.00019836199069471437,
      "loss": 1.0797,
      "step": 623
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5634730458259583,
      "learning_rate": 0.00019835656826551552,
      "loss": 0.9508,
      "step": 624
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6410563588142395,
      "learning_rate": 0.00019835113695035764,
      "loss": 0.931,
      "step": 625
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.539527952671051,
      "learning_rate": 0.00019834569674973137,
      "loss": 1.0981,
      "step": 626
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6172979474067688,
      "learning_rate": 0.00019834024766412825,
      "loss": 1.131,
      "step": 627
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5615272521972656,
      "learning_rate": 0.00019833478969404054,
      "loss": 1.246,
      "step": 628
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6126092076301575,
      "learning_rate": 0.00019832932283996138,
      "loss": 1.0175,
      "step": 629
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4889993369579315,
      "learning_rate": 0.00019832384710238464,
      "loss": 1.0162,
      "step": 630
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.49586522579193115,
      "learning_rate": 0.00019831836248180497,
      "loss": 1.0558,
      "step": 631
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5401427745819092,
      "learning_rate": 0.00019831286897871797,
      "loss": 1.1391,
      "step": 632
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6986876130104065,
      "learning_rate": 0.00019830736659361988,
      "loss": 1.2645,
      "step": 633
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.49258509278297424,
      "learning_rate": 0.00019830185532700788,
      "loss": 1.0886,
      "step": 634
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4943321943283081,
      "learning_rate": 0.0001982963351793798,
      "loss": 1.2191,
      "step": 635
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.46854934096336365,
      "learning_rate": 0.0001982908061512344,
      "loss": 1.0435,
      "step": 636
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5639446377754211,
      "learning_rate": 0.00019828526824307115,
      "loss": 1.132,
      "step": 637
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5033563375473022,
      "learning_rate": 0.00019827972145539043,
      "loss": 1.1469,
      "step": 638
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5321240425109863,
      "learning_rate": 0.00019827416578869334,
      "loss": 1.2356,
      "step": 639
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5140905380249023,
      "learning_rate": 0.0001982686012434818,
      "loss": 1.0041,
      "step": 640
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5813565850257874,
      "learning_rate": 0.00019826302782025853,
      "loss": 1.2285,
      "step": 641
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4850314259529114,
      "learning_rate": 0.00019825744551952706,
      "loss": 1.1132,
      "step": 642
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5613282322883606,
      "learning_rate": 0.0001982518543417917,
      "loss": 1.1257,
      "step": 643
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5611051917076111,
      "learning_rate": 0.0001982462542875576,
      "loss": 1.1526,
      "step": 644
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.48500093817710876,
      "learning_rate": 0.00019824064535733073,
      "loss": 1.0102,
      "step": 645
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5718590617179871,
      "learning_rate": 0.00019823502755161778,
      "loss": 1.2613,
      "step": 646
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5485739707946777,
      "learning_rate": 0.00019822940087092629,
      "loss": 0.9491,
      "step": 647
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4833703339099884,
      "learning_rate": 0.0001982237653157646,
      "loss": 0.9716,
      "step": 648
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5101345777511597,
      "learning_rate": 0.00019821812088664185,
      "loss": 1.0698,
      "step": 649
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5286256074905396,
      "learning_rate": 0.000198212467584068,
      "loss": 1.019,
      "step": 650
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5590866208076477,
      "learning_rate": 0.00019820680540855377,
      "loss": 1.2636,
      "step": 651
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5393713712692261,
      "learning_rate": 0.00019820113436061073,
      "loss": 1.2037,
      "step": 652
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.495000958442688,
      "learning_rate": 0.0001981954544407512,
      "loss": 1.0951,
      "step": 653
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5339246988296509,
      "learning_rate": 0.00019818976564948835,
      "loss": 1.0815,
      "step": 654
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5487880110740662,
      "learning_rate": 0.00019818406798733612,
      "loss": 1.1396,
      "step": 655
    },
    {
      "epoch": 0.25,
      "eval_loss": 1.0153948068618774,
      "eval_runtime": 17.9627,
      "eval_samples_per_second": 151.926,
      "eval_steps_per_second": 38.023,
      "step": 655
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.52154541015625,
      "learning_rate": 0.00019817836145480925,
      "loss": 1.1395,
      "step": 656
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4636710286140442,
      "learning_rate": 0.00019817264605242332,
      "loss": 1.071,
      "step": 657
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5892826914787292,
      "learning_rate": 0.0001981669217806947,
      "loss": 1.0462,
      "step": 658
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5976261496543884,
      "learning_rate": 0.00019816118864014048,
      "loss": 1.2249,
      "step": 659
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6425426602363586,
      "learning_rate": 0.00019815544663127866,
      "loss": 1.2933,
      "step": 660
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6284853219985962,
      "learning_rate": 0.000198149695754628,
      "loss": 0.9808,
      "step": 661
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5053337216377258,
      "learning_rate": 0.00019814393601070803,
      "loss": 0.9719,
      "step": 662
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5163286924362183,
      "learning_rate": 0.00019813816740003914,
      "loss": 1.1396,
      "step": 663
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.54429692029953,
      "learning_rate": 0.00019813238992314253,
      "loss": 1.0073,
      "step": 664
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5420398116111755,
      "learning_rate": 0.00019812660358054006,
      "loss": 1.1474,
      "step": 665
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5848464369773865,
      "learning_rate": 0.00019812080837275454,
      "loss": 1.0693,
      "step": 666
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5412850379943848,
      "learning_rate": 0.00019811500430030958,
      "loss": 1.2033,
      "step": 667
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5231608152389526,
      "learning_rate": 0.0001981091913637295,
      "loss": 0.9086,
      "step": 668
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.517072319984436,
      "learning_rate": 0.00019810336956353942,
      "loss": 1.1337,
      "step": 669
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6114885210990906,
      "learning_rate": 0.00019809753890026542,
      "loss": 1.1743,
      "step": 670
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.48399075865745544,
      "learning_rate": 0.0001980916993744342,
      "loss": 0.9902,
      "step": 671
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5298494696617126,
      "learning_rate": 0.00019808585098657333,
      "loss": 1.0834,
      "step": 672
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5229631662368774,
      "learning_rate": 0.00019807999373721118,
      "loss": 1.1433,
      "step": 673
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5524067282676697,
      "learning_rate": 0.0001980741276268769,
      "loss": 1.1291,
      "step": 674
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4828319847583771,
      "learning_rate": 0.00019806825265610052,
      "loss": 1.0705,
      "step": 675
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5148788690567017,
      "learning_rate": 0.00019806236882541275,
      "loss": 1.0661,
      "step": 676
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5624847412109375,
      "learning_rate": 0.0001980564761353452,
      "loss": 1.2352,
      "step": 677
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4777878522872925,
      "learning_rate": 0.00019805057458643024,
      "loss": 1.0932,
      "step": 678
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5480254888534546,
      "learning_rate": 0.000198044664179201,
      "loss": 1.1319,
      "step": 679
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5098835229873657,
      "learning_rate": 0.00019803874491419146,
      "loss": 1.0556,
      "step": 680
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5364381074905396,
      "learning_rate": 0.00019803281679193645,
      "loss": 1.2167,
      "step": 681
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5941902995109558,
      "learning_rate": 0.0001980268798129715,
      "loss": 1.4031,
      "step": 682
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.47746074199676514,
      "learning_rate": 0.00019802093397783296,
      "loss": 0.9778,
      "step": 683
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5172235369682312,
      "learning_rate": 0.00019801497928705804,
      "loss": 1.1598,
      "step": 684
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6679323315620422,
      "learning_rate": 0.00019800901574118472,
      "loss": 1.2064,
      "step": 685
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5311746597290039,
      "learning_rate": 0.00019800304334075173,
      "loss": 1.1862,
      "step": 686
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5657793283462524,
      "learning_rate": 0.00019799706208629866,
      "loss": 1.1143,
      "step": 687
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.47132763266563416,
      "learning_rate": 0.00019799107197836593,
      "loss": 1.1225,
      "step": 688
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5872971415519714,
      "learning_rate": 0.00019798507301749463,
      "loss": 1.1489,
      "step": 689
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.511029839515686,
      "learning_rate": 0.00019797906520422677,
      "loss": 1.0488,
      "step": 690
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5370973944664001,
      "learning_rate": 0.00019797304853910513,
      "loss": 1.1974,
      "step": 691
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5594046711921692,
      "learning_rate": 0.0001979670230226733,
      "loss": 1.0964,
      "step": 692
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6002780795097351,
      "learning_rate": 0.00019796098865547564,
      "loss": 1.0355,
      "step": 693
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5207370519638062,
      "learning_rate": 0.00019795494543805727,
      "loss": 1.0801,
      "step": 694
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6425591707229614,
      "learning_rate": 0.0001979488933709642,
      "loss": 1.0904,
      "step": 695
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5756493210792542,
      "learning_rate": 0.0001979428324547432,
      "loss": 1.0362,
      "step": 696
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6055238842964172,
      "learning_rate": 0.00019793676268994187,
      "loss": 1.2432,
      "step": 697
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.48325762152671814,
      "learning_rate": 0.00019793068407710851,
      "loss": 1.0837,
      "step": 698
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5550029873847961,
      "learning_rate": 0.00019792459661679236,
      "loss": 1.2134,
      "step": 699
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.49292463064193726,
      "learning_rate": 0.00019791850030954333,
      "loss": 1.0565,
      "step": 700
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.48105907440185547,
      "learning_rate": 0.00019791239515591224,
      "loss": 1.0579,
      "step": 701
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5209401845932007,
      "learning_rate": 0.00019790628115645058,
      "loss": 1.1013,
      "step": 702
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5278286337852478,
      "learning_rate": 0.00019790015831171079,
      "loss": 1.0529,
      "step": 703
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5473965406417847,
      "learning_rate": 0.000197894026622246,
      "loss": 1.121,
      "step": 704
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5992923974990845,
      "learning_rate": 0.00019788788608861017,
      "loss": 1.3566,
      "step": 705
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6117520928382874,
      "learning_rate": 0.00019788173671135808,
      "loss": 1.1498,
      "step": 706
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5273234248161316,
      "learning_rate": 0.00019787557849104527,
      "loss": 1.0953,
      "step": 707
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.49864324927330017,
      "learning_rate": 0.00019786941142822814,
      "loss": 1.0837,
      "step": 708
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6347833871841431,
      "learning_rate": 0.0001978632355234638,
      "loss": 0.9678,
      "step": 709
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5046628713607788,
      "learning_rate": 0.00019785705077731024,
      "loss": 1.0924,
      "step": 710
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4987553656101227,
      "learning_rate": 0.00019785085719032622,
      "loss": 0.953,
      "step": 711
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5188940167427063,
      "learning_rate": 0.0001978446547630713,
      "loss": 0.9737,
      "step": 712
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.585609495639801,
      "learning_rate": 0.00019783844349610576,
      "loss": 0.9819,
      "step": 713
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5059577226638794,
      "learning_rate": 0.00019783222338999083,
      "loss": 1.0045,
      "step": 714
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5237625241279602,
      "learning_rate": 0.00019782599444528846,
      "loss": 0.8314,
      "step": 715
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.49594178795814514,
      "learning_rate": 0.00019781975666256137,
      "loss": 0.9754,
      "step": 716
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6107966899871826,
      "learning_rate": 0.0001978135100423731,
      "loss": 1.2708,
      "step": 717
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5323823094367981,
      "learning_rate": 0.00019780725458528803,
      "loss": 1.1506,
      "step": 718
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5096708536148071,
      "learning_rate": 0.0001978009902918713,
      "loss": 0.9683,
      "step": 719
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.48245328664779663,
      "learning_rate": 0.00019779471716268885,
      "loss": 1.0692,
      "step": 720
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5705918073654175,
      "learning_rate": 0.00019778843519830742,
      "loss": 1.1573,
      "step": 721
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5196608901023865,
      "learning_rate": 0.00019778214439929452,
      "loss": 1.0694,
      "step": 722
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5451333522796631,
      "learning_rate": 0.00019777584476621853,
      "loss": 1.1756,
      "step": 723
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6256030797958374,
      "learning_rate": 0.00019776953629964855,
      "loss": 1.1507,
      "step": 724
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5393909811973572,
      "learning_rate": 0.00019776321900015457,
      "loss": 0.992,
      "step": 725
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.563827395439148,
      "learning_rate": 0.0001977568928683073,
      "loss": 1.111,
      "step": 726
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.55351722240448,
      "learning_rate": 0.0001977505579046782,
      "loss": 1.1057,
      "step": 727
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5464900135993958,
      "learning_rate": 0.0001977442141098397,
      "loss": 1.1258,
      "step": 728
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6698980927467346,
      "learning_rate": 0.00019773786148436485,
      "loss": 1.1193,
      "step": 729
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5508808493614197,
      "learning_rate": 0.00019773150002882762,
      "loss": 0.9323,
      "step": 730
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5508869886398315,
      "learning_rate": 0.00019772512974380275,
      "loss": 1.0364,
      "step": 731
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5076518654823303,
      "learning_rate": 0.00019771875062986566,
      "loss": 1.0878,
      "step": 732
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4498327374458313,
      "learning_rate": 0.0001977123626875928,
      "loss": 1.0504,
      "step": 733
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45657438039779663,
      "learning_rate": 0.00019770596591756116,
      "loss": 0.9122,
      "step": 734
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5719609260559082,
      "learning_rate": 0.00019769956032034873,
      "loss": 1.3045,
      "step": 735
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6293899416923523,
      "learning_rate": 0.00019769314589653423,
      "loss": 1.1325,
      "step": 736
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5307220220565796,
      "learning_rate": 0.0001976867226466971,
      "loss": 1.0942,
      "step": 737
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5085244178771973,
      "learning_rate": 0.0001976802905714177,
      "loss": 1.1687,
      "step": 738
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6277353167533875,
      "learning_rate": 0.00019767384967127707,
      "loss": 1.0866,
      "step": 739
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5884594917297363,
      "learning_rate": 0.00019766739994685718,
      "loss": 1.175,
      "step": 740
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5401700139045715,
      "learning_rate": 0.0001976609413987407,
      "loss": 1.1499,
      "step": 741
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5686537623405457,
      "learning_rate": 0.0001976544740275111,
      "loss": 1.0965,
      "step": 742
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5871611833572388,
      "learning_rate": 0.00019764799783375273,
      "loss": 1.2375,
      "step": 743
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5987391471862793,
      "learning_rate": 0.0001976415128180506,
      "loss": 1.2463,
      "step": 744
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4921867251396179,
      "learning_rate": 0.00019763501898099065,
      "loss": 1.0619,
      "step": 745
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5578277707099915,
      "learning_rate": 0.00019762851632315953,
      "loss": 1.0181,
      "step": 746
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5684673190116882,
      "learning_rate": 0.00019762200484514474,
      "loss": 1.2507,
      "step": 747
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5005682110786438,
      "learning_rate": 0.00019761548454753453,
      "loss": 1.1111,
      "step": 748
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5176371335983276,
      "learning_rate": 0.000197608955430918,
      "loss": 1.0868,
      "step": 749
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6516520977020264,
      "learning_rate": 0.00019760241749588501,
      "loss": 0.8422,
      "step": 750
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5634924173355103,
      "learning_rate": 0.00019759587074302625,
      "loss": 1.184,
      "step": 751
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4741862416267395,
      "learning_rate": 0.0001975893151729331,
      "loss": 1.0444,
      "step": 752
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5085967183113098,
      "learning_rate": 0.0001975827507861979,
      "loss": 0.9326,
      "step": 753
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5261965394020081,
      "learning_rate": 0.0001975761775834137,
      "loss": 1.0514,
      "step": 754
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.558593213558197,
      "learning_rate": 0.0001975695955651743,
      "loss": 0.9558,
      "step": 755
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4968149960041046,
      "learning_rate": 0.0001975630047320744,
      "loss": 0.9564,
      "step": 756
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5512811541557312,
      "learning_rate": 0.00019755640508470942,
      "loss": 1.0315,
      "step": 757
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5796666145324707,
      "learning_rate": 0.00019754979662367557,
      "loss": 1.1686,
      "step": 758
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.48021426796913147,
      "learning_rate": 0.00019754317934956995,
      "loss": 0.9551,
      "step": 759
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5672682523727417,
      "learning_rate": 0.00019753655326299035,
      "loss": 1.0394,
      "step": 760
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5781568288803101,
      "learning_rate": 0.00019752991836453543,
      "loss": 1.0976,
      "step": 761
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.566172182559967,
      "learning_rate": 0.00019752327465480458,
      "loss": 1.0958,
      "step": 762
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5664011240005493,
      "learning_rate": 0.00019751662213439805,
      "loss": 1.28,
      "step": 763
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5427090525627136,
      "learning_rate": 0.00019750996080391685,
      "loss": 1.1989,
      "step": 764
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5100749731063843,
      "learning_rate": 0.0001975032906639628,
      "loss": 0.9632,
      "step": 765
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5142964720726013,
      "learning_rate": 0.0001974966117151385,
      "loss": 1.0154,
      "step": 766
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.48683270812034607,
      "learning_rate": 0.00019748992395804733,
      "loss": 1.1774,
      "step": 767
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4964374601840973,
      "learning_rate": 0.00019748322739329352,
      "loss": 1.0984,
      "step": 768
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5703089833259583,
      "learning_rate": 0.0001974765220214821,
      "loss": 1.263,
      "step": 769
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5038565993309021,
      "learning_rate": 0.0001974698078432188,
      "loss": 0.9689,
      "step": 770
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5699312090873718,
      "learning_rate": 0.00019746308485911023,
      "loss": 1.2384,
      "step": 771
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.47143155336380005,
      "learning_rate": 0.0001974563530697638,
      "loss": 1.0539,
      "step": 772
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5221191644668579,
      "learning_rate": 0.00019744961247578763,
      "loss": 1.0732,
      "step": 773
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5209429264068604,
      "learning_rate": 0.00019744286307779075,
      "loss": 1.0056,
      "step": 774
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5356734395027161,
      "learning_rate": 0.0001974361048763829,
      "loss": 0.9635,
      "step": 775
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.512253999710083,
      "learning_rate": 0.00019742933787217468,
      "loss": 1.1979,
      "step": 776
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5092757344245911,
      "learning_rate": 0.0001974225620657774,
      "loss": 1.1399,
      "step": 777
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6500785946846008,
      "learning_rate": 0.00019741577745780326,
      "loss": 1.0413,
      "step": 778
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5108067393302917,
      "learning_rate": 0.0001974089840488652,
      "loss": 0.8789,
      "step": 779
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5151072144508362,
      "learning_rate": 0.00019740218183957696,
      "loss": 1.0146,
      "step": 780
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.49716851115226746,
      "learning_rate": 0.00019739537083055306,
      "loss": 1.0114,
      "step": 781
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48336997628211975,
      "learning_rate": 0.00019738855102240885,
      "loss": 0.8086,
      "step": 782
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5495067238807678,
      "learning_rate": 0.00019738172241576049,
      "loss": 1.1508,
      "step": 783
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6326724886894226,
      "learning_rate": 0.00019737488501122487,
      "loss": 1.0295,
      "step": 784
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5587719082832336,
      "learning_rate": 0.00019736803880941977,
      "loss": 1.0741,
      "step": 785
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6244709491729736,
      "learning_rate": 0.0001973611838109636,
      "loss": 1.1393,
      "step": 786
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5473697185516357,
      "learning_rate": 0.00019735432001647576,
      "loss": 1.0927,
      "step": 787
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.531061053276062,
      "learning_rate": 0.00019734744742657632,
      "loss": 1.1514,
      "step": 788
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48232513666152954,
      "learning_rate": 0.00019734056604188616,
      "loss": 1.1516,
      "step": 789
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6108174324035645,
      "learning_rate": 0.00019733367586302703,
      "loss": 1.1386,
      "step": 790
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6374059319496155,
      "learning_rate": 0.0001973267768906214,
      "loss": 0.9623,
      "step": 791
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5877413153648376,
      "learning_rate": 0.0001973198691252925,
      "loss": 1.0466,
      "step": 792
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5468994975090027,
      "learning_rate": 0.00019731295256766447,
      "loss": 1.1942,
      "step": 793
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5458383560180664,
      "learning_rate": 0.00019730602721836218,
      "loss": 1.1622,
      "step": 794
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4972197115421295,
      "learning_rate": 0.00019729909307801124,
      "loss": 1.027,
      "step": 795
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5629715919494629,
      "learning_rate": 0.00019729215014723816,
      "loss": 1.3745,
      "step": 796
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5757743120193481,
      "learning_rate": 0.0001972851984266702,
      "loss": 1.0184,
      "step": 797
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.572532594203949,
      "learning_rate": 0.00019727823791693537,
      "loss": 1.0389,
      "step": 798
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5449635982513428,
      "learning_rate": 0.00019727126861866253,
      "loss": 1.0972,
      "step": 799
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5162309408187866,
      "learning_rate": 0.0001972642905324813,
      "loss": 1.167,
      "step": 800
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.611807644367218,
      "learning_rate": 0.0001972573036590222,
      "loss": 1.3319,
      "step": 801
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5340973138809204,
      "learning_rate": 0.0001972503079989163,
      "loss": 1.1356,
      "step": 802
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.56681889295578,
      "learning_rate": 0.00019724330355279574,
      "loss": 1.2657,
      "step": 803
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.49857208132743835,
      "learning_rate": 0.00019723629032129329,
      "loss": 1.1107,
      "step": 804
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5250954627990723,
      "learning_rate": 0.00019722926830504256,
      "loss": 1.1725,
      "step": 805
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5133506655693054,
      "learning_rate": 0.00019722223750467794,
      "loss": 0.9294,
      "step": 806
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5176982879638672,
      "learning_rate": 0.00019721519792083468,
      "loss": 1.0949,
      "step": 807
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5122258067131042,
      "learning_rate": 0.00019720814955414865,
      "loss": 1.0317,
      "step": 808
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6034652590751648,
      "learning_rate": 0.00019720109240525674,
      "loss": 0.9448,
      "step": 809
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5805848240852356,
      "learning_rate": 0.00019719402647479649,
      "loss": 1.1572,
      "step": 810
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5414599180221558,
      "learning_rate": 0.00019718695176340623,
      "loss": 1.0706,
      "step": 811
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5518433451652527,
      "learning_rate": 0.00019717986827172515,
      "loss": 1.1705,
      "step": 812
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5469940304756165,
      "learning_rate": 0.00019717277600039325,
      "loss": 1.1299,
      "step": 813
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5368144512176514,
      "learning_rate": 0.00019716567495005116,
      "loss": 1.1,
      "step": 814
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5745230317115784,
      "learning_rate": 0.00019715856512134052,
      "loss": 1.052,
      "step": 815
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.466328501701355,
      "learning_rate": 0.00019715144651490362,
      "loss": 0.9903,
      "step": 816
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6098025441169739,
      "learning_rate": 0.00019714431913138363,
      "loss": 1.0933,
      "step": 817
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5240216851234436,
      "learning_rate": 0.0001971371829714244,
      "loss": 1.1298,
      "step": 818
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.510958731174469,
      "learning_rate": 0.00019713003803567073,
      "loss": 1.1109,
      "step": 819
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.505074679851532,
      "learning_rate": 0.00019712288432476802,
      "loss": 1.1258,
      "step": 820
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5658806562423706,
      "learning_rate": 0.00019711572183936266,
      "loss": 1.0631,
      "step": 821
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4546443521976471,
      "learning_rate": 0.00019710855058010168,
      "loss": 0.916,
      "step": 822
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.526270866394043,
      "learning_rate": 0.00019710137054763295,
      "loss": 1.1089,
      "step": 823
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5158497095108032,
      "learning_rate": 0.0001970941817426052,
      "loss": 1.0486,
      "step": 824
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5480288863182068,
      "learning_rate": 0.0001970869841656679,
      "loss": 1.0551,
      "step": 825
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.46271073818206787,
      "learning_rate": 0.00019707977781747125,
      "loss": 1.0294,
      "step": 826
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5485731363296509,
      "learning_rate": 0.00019707256269866638,
      "loss": 1.1623,
      "step": 827
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5255643725395203,
      "learning_rate": 0.00019706533880990506,
      "loss": 0.9108,
      "step": 828
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5513526201248169,
      "learning_rate": 0.00019705810615183996,
      "loss": 1.1178,
      "step": 829
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5589191317558289,
      "learning_rate": 0.00019705086472512453,
      "loss": 1.2384,
      "step": 830
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.48324573040008545,
      "learning_rate": 0.00019704361453041296,
      "loss": 0.9521,
      "step": 831
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5757077932357788,
      "learning_rate": 0.00019703635556836028,
      "loss": 1.1034,
      "step": 832
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6256049871444702,
      "learning_rate": 0.00019702908783962228,
      "loss": 1.2153,
      "step": 833
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5107579827308655,
      "learning_rate": 0.0001970218113448556,
      "loss": 1.0975,
      "step": 834
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.47654515504837036,
      "learning_rate": 0.00019701452608471759,
      "loss": 1.0845,
      "step": 835
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.557918906211853,
      "learning_rate": 0.00019700723205986643,
      "loss": 1.1108,
      "step": 836
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5833942294120789,
      "learning_rate": 0.00019699992927096113,
      "loss": 1.124,
      "step": 837
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5856698751449585,
      "learning_rate": 0.0001969926177186614,
      "loss": 1.2941,
      "step": 838
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6649505496025085,
      "learning_rate": 0.00019698529740362785,
      "loss": 1.1847,
      "step": 839
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5869125723838806,
      "learning_rate": 0.0001969779683265218,
      "loss": 1.1469,
      "step": 840
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.47669145464897156,
      "learning_rate": 0.00019697063048800542,
      "loss": 1.0475,
      "step": 841
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5173715949058533,
      "learning_rate": 0.00019696328388874162,
      "loss": 1.1664,
      "step": 842
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4949846863746643,
      "learning_rate": 0.0001969559285293941,
      "loss": 1.2023,
      "step": 843
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5999616384506226,
      "learning_rate": 0.0001969485644106274,
      "loss": 1.1492,
      "step": 844
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5564090013504028,
      "learning_rate": 0.00019694119153310685,
      "loss": 0.9729,
      "step": 845
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5422832369804382,
      "learning_rate": 0.00019693380989749853,
      "loss": 1.1642,
      "step": 846
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5115847587585449,
      "learning_rate": 0.0001969264195044693,
      "loss": 1.1102,
      "step": 847
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5365723967552185,
      "learning_rate": 0.00019691902035468687,
      "loss": 1.1249,
      "step": 848
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5202025175094604,
      "learning_rate": 0.0001969116124488197,
      "loss": 1.1992,
      "step": 849
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5052026510238647,
      "learning_rate": 0.00019690419578753706,
      "loss": 0.9967,
      "step": 850
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.552277147769928,
      "learning_rate": 0.00019689677037150902,
      "loss": 1.161,
      "step": 851
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5004220604896545,
      "learning_rate": 0.00019688933620140637,
      "loss": 1.2874,
      "step": 852
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.596388578414917,
      "learning_rate": 0.0001968818932779008,
      "loss": 1.1453,
      "step": 853
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5080670714378357,
      "learning_rate": 0.00019687444160166473,
      "loss": 0.9901,
      "step": 854
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5395559072494507,
      "learning_rate": 0.00019686698117337133,
      "loss": 1.2054,
      "step": 855
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.592228353023529,
      "learning_rate": 0.00019685951199369467,
      "loss": 1.1201,
      "step": 856
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5213476419448853,
      "learning_rate": 0.0001968520340633095,
      "loss": 1.1273,
      "step": 857
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5358402729034424,
      "learning_rate": 0.00019684454738289143,
      "loss": 1.1751,
      "step": 858
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6148902177810669,
      "learning_rate": 0.00019683705195311686,
      "loss": 1.2059,
      "step": 859
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6230486035346985,
      "learning_rate": 0.00019682954777466293,
      "loss": 1.1166,
      "step": 860
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4527413547039032,
      "learning_rate": 0.00019682203484820758,
      "loss": 1.0781,
      "step": 861
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5335724949836731,
      "learning_rate": 0.00019681451317442962,
      "loss": 1.1133,
      "step": 862
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5677633881568909,
      "learning_rate": 0.00019680698275400854,
      "loss": 1.0835,
      "step": 863
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5778499245643616,
      "learning_rate": 0.00019679944358762466,
      "loss": 1.2469,
      "step": 864
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5224189162254333,
      "learning_rate": 0.00019679189567595916,
      "loss": 1.2731,
      "step": 865
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6404327154159546,
      "learning_rate": 0.00019678433901969391,
      "loss": 1.2088,
      "step": 866
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5300662517547607,
      "learning_rate": 0.00019677677361951165,
      "loss": 1.3882,
      "step": 867
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.47733646631240845,
      "learning_rate": 0.0001967691994760958,
      "loss": 0.9524,
      "step": 868
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5740486979484558,
      "learning_rate": 0.0001967616165901307,
      "loss": 1.1168,
      "step": 869
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5322950482368469,
      "learning_rate": 0.0001967540249623014,
      "loss": 1.176,
      "step": 870
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4464534819126129,
      "learning_rate": 0.00019674642459329377,
      "loss": 0.888,
      "step": 871
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5296953916549683,
      "learning_rate": 0.00019673881548379444,
      "loss": 1.2192,
      "step": 872
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5391892194747925,
      "learning_rate": 0.0001967311976344909,
      "loss": 1.1632,
      "step": 873
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.47413426637649536,
      "learning_rate": 0.00019672357104607132,
      "loss": 1.0121,
      "step": 874
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5288782119750977,
      "learning_rate": 0.00019671593571922476,
      "loss": 0.895,
      "step": 875
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5563179850578308,
      "learning_rate": 0.000196708291654641,
      "loss": 1.306,
      "step": 876
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6332594752311707,
      "learning_rate": 0.00019670063885301065,
      "loss": 1.2006,
      "step": 877
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5800545811653137,
      "learning_rate": 0.00019669297731502507,
      "loss": 1.1349,
      "step": 878
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.553862452507019,
      "learning_rate": 0.0001966853070413765,
      "loss": 1.1195,
      "step": 879
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5305919647216797,
      "learning_rate": 0.00019667762803275785,
      "loss": 1.1277,
      "step": 880
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6028212308883667,
      "learning_rate": 0.0001966699402898629,
      "loss": 1.3267,
      "step": 881
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5439375042915344,
      "learning_rate": 0.00019666224381338618,
      "loss": 1.0961,
      "step": 882
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5040217638015747,
      "learning_rate": 0.00019665453860402305,
      "loss": 0.9993,
      "step": 883
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6264601945877075,
      "learning_rate": 0.00019664682466246963,
      "loss": 1.2248,
      "step": 884
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.48470181226730347,
      "learning_rate": 0.00019663910198942277,
      "loss": 1.0676,
      "step": 885
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4559854567050934,
      "learning_rate": 0.00019663137058558024,
      "loss": 0.9664,
      "step": 886
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6404067277908325,
      "learning_rate": 0.00019662363045164048,
      "loss": 1.1033,
      "step": 887
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5953955054283142,
      "learning_rate": 0.00019661588158830285,
      "loss": 1.2204,
      "step": 888
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5667288899421692,
      "learning_rate": 0.0001966081239962673,
      "loss": 0.9428,
      "step": 889
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5570257902145386,
      "learning_rate": 0.00019660035767623475,
      "loss": 1.1523,
      "step": 890
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6536972522735596,
      "learning_rate": 0.00019659258262890683,
      "loss": 1.2999,
      "step": 891
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5308727622032166,
      "learning_rate": 0.000196584798854986,
      "loss": 1.1709,
      "step": 892
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4792231321334839,
      "learning_rate": 0.00019657700635517542,
      "loss": 0.9741,
      "step": 893
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5758273005485535,
      "learning_rate": 0.00019656920513017915,
      "loss": 1.2811,
      "step": 894
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5072007179260254,
      "learning_rate": 0.00019656139518070198,
      "loss": 1.0564,
      "step": 895
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5229262709617615,
      "learning_rate": 0.0001965535765074495,
      "loss": 0.9786,
      "step": 896
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46316370368003845,
      "learning_rate": 0.00019654574911112804,
      "loss": 1.1107,
      "step": 897
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5957727432250977,
      "learning_rate": 0.00019653791299244477,
      "loss": 1.1218,
      "step": 898
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.534303605556488,
      "learning_rate": 0.0001965300681521077,
      "loss": 1.2024,
      "step": 899
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5468682050704956,
      "learning_rate": 0.00019652221459082552,
      "loss": 0.936,
      "step": 900
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5230331420898438,
      "learning_rate": 0.0001965143523093077,
      "loss": 1.0716,
      "step": 901
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4797626733779907,
      "learning_rate": 0.00019650648130826467,
      "loss": 1.002,
      "step": 902
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5666486024856567,
      "learning_rate": 0.00019649860158840748,
      "loss": 1.0306,
      "step": 903
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5790895223617554,
      "learning_rate": 0.00019649071315044797,
      "loss": 1.0271,
      "step": 904
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5381376147270203,
      "learning_rate": 0.00019648281599509888,
      "loss": 1.1928,
      "step": 905
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5734236836433411,
      "learning_rate": 0.00019647491012307364,
      "loss": 1.2453,
      "step": 906
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4927200376987457,
      "learning_rate": 0.0001964669955350865,
      "loss": 0.98,
      "step": 907
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5980042219161987,
      "learning_rate": 0.00019645907223185254,
      "loss": 1.0942,
      "step": 908
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4530969560146332,
      "learning_rate": 0.0001964511402140875,
      "loss": 0.9979,
      "step": 909
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6606910228729248,
      "learning_rate": 0.00019644319948250812,
      "loss": 1.1149,
      "step": 910
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5825684666633606,
      "learning_rate": 0.0001964352500378317,
      "loss": 1.094,
      "step": 911
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5432309508323669,
      "learning_rate": 0.00019642729188077643,
      "loss": 1.0084,
      "step": 912
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5725202560424805,
      "learning_rate": 0.0001964193250120613,
      "loss": 1.1022,
      "step": 913
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6069723963737488,
      "learning_rate": 0.00019641134943240612,
      "loss": 1.1993,
      "step": 914
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5283558964729309,
      "learning_rate": 0.00019640336514253136,
      "loss": 0.8801,
      "step": 915
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6143441796302795,
      "learning_rate": 0.00019639537214315844,
      "loss": 1.0242,
      "step": 916
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5614748597145081,
      "learning_rate": 0.0001963873704350094,
      "loss": 1.1471,
      "step": 917
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5619175434112549,
      "learning_rate": 0.0001963793600188072,
      "loss": 1.0781,
      "step": 918
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5835575461387634,
      "learning_rate": 0.0001963713408952755,
      "loss": 1.0833,
      "step": 919
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6437380909919739,
      "learning_rate": 0.00019636331306513883,
      "loss": 1.0159,
      "step": 920
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5471232533454895,
      "learning_rate": 0.00019635527652912244,
      "loss": 0.9727,
      "step": 921
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5375336408615112,
      "learning_rate": 0.00019634723128795231,
      "loss": 1.15,
      "step": 922
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5501574277877808,
      "learning_rate": 0.00019633917734235542,
      "loss": 1.0332,
      "step": 923
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5777191519737244,
      "learning_rate": 0.0001963311146930593,
      "loss": 1.0671,
      "step": 924
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.558753490447998,
      "learning_rate": 0.00019632304334079242,
      "loss": 1.0862,
      "step": 925
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.49828171730041504,
      "learning_rate": 0.00019631496328628391,
      "loss": 1.1407,
      "step": 926
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.538738489151001,
      "learning_rate": 0.0001963068745302638,
      "loss": 1.0637,
      "step": 927
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.609881579875946,
      "learning_rate": 0.00019629877707346292,
      "loss": 1.1749,
      "step": 928
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5410135388374329,
      "learning_rate": 0.00019629067091661272,
      "loss": 1.0109,
      "step": 929
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4779724180698395,
      "learning_rate": 0.00019628255606044564,
      "loss": 1.1007,
      "step": 930
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.603225827217102,
      "learning_rate": 0.00019627443250569475,
      "loss": 1.0987,
      "step": 931
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5400119423866272,
      "learning_rate": 0.000196266300253094,
      "loss": 1.1858,
      "step": 932
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5057092905044556,
      "learning_rate": 0.00019625815930337804,
      "loss": 1.1432,
      "step": 933
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.545569658279419,
      "learning_rate": 0.00019625000965728243,
      "loss": 1.2376,
      "step": 934
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.47009384632110596,
      "learning_rate": 0.00019624185131554342,
      "loss": 0.9036,
      "step": 935
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5348869562149048,
      "learning_rate": 0.00019623368427889806,
      "loss": 1.052,
      "step": 936
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.45622172951698303,
      "learning_rate": 0.0001962255085480842,
      "loss": 0.8479,
      "step": 937
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5504438281059265,
      "learning_rate": 0.00019621732412384045,
      "loss": 1.1362,
      "step": 938
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.48980462551116943,
      "learning_rate": 0.00019620913100690626,
      "loss": 1.009,
      "step": 939
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5361881256103516,
      "learning_rate": 0.00019620092919802183,
      "loss": 1.1752,
      "step": 940
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.54579758644104,
      "learning_rate": 0.00019619271869792814,
      "loss": 1.0558,
      "step": 941
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.549178421497345,
      "learning_rate": 0.0001961844995073669,
      "loss": 1.0308,
      "step": 942
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5896941423416138,
      "learning_rate": 0.0001961762716270808,
      "loss": 1.2261,
      "step": 943
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5419191122055054,
      "learning_rate": 0.00019616803505781305,
      "loss": 1.152,
      "step": 944
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5686753392219543,
      "learning_rate": 0.00019615978980030784,
      "loss": 1.0636,
      "step": 945
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5907273888587952,
      "learning_rate": 0.0001961515358553101,
      "loss": 1.0999,
      "step": 946
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5179197788238525,
      "learning_rate": 0.0001961432732235655,
      "loss": 1.0976,
      "step": 947
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5165078639984131,
      "learning_rate": 0.00019613500190582053,
      "loss": 1.1532,
      "step": 948
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5007051229476929,
      "learning_rate": 0.00019612672190282243,
      "loss": 1.0341,
      "step": 949
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5154484510421753,
      "learning_rate": 0.0001961184332153193,
      "loss": 1.0931,
      "step": 950
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6172532439231873,
      "learning_rate": 0.00019611013584405995,
      "loss": 1.2503,
      "step": 951
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5256145000457764,
      "learning_rate": 0.000196101829789794,
      "loss": 1.1124,
      "step": 952
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.49367621541023254,
      "learning_rate": 0.00019609351505327184,
      "loss": 1.0179,
      "step": 953
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5787032246589661,
      "learning_rate": 0.00019608519163524468,
      "loss": 1.0986,
      "step": 954
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5206260085105896,
      "learning_rate": 0.00019607685953646452,
      "loss": 0.9906,
      "step": 955
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5005407929420471,
      "learning_rate": 0.000196068518757684,
      "loss": 1.0735,
      "step": 956
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5073145031929016,
      "learning_rate": 0.00019606016929965683,
      "loss": 1.0736,
      "step": 957
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5678821206092834,
      "learning_rate": 0.00019605181116313724,
      "loss": 1.1991,
      "step": 958
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5657750964164734,
      "learning_rate": 0.00019604344434888033,
      "loss": 1.0046,
      "step": 959
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5822930335998535,
      "learning_rate": 0.00019603506885764206,
      "loss": 1.2042,
      "step": 960
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5306090116500854,
      "learning_rate": 0.00019602668469017903,
      "loss": 1.0183,
      "step": 961
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6258050799369812,
      "learning_rate": 0.00019601829184724877,
      "loss": 1.2029,
      "step": 962
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5240448713302612,
      "learning_rate": 0.0001960098903296095,
      "loss": 0.9021,
      "step": 963
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5050828456878662,
      "learning_rate": 0.0001960014801380202,
      "loss": 0.9737,
      "step": 964
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4429282546043396,
      "learning_rate": 0.00019599306127324076,
      "loss": 0.8778,
      "step": 965
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5414524078369141,
      "learning_rate": 0.00019598463373603174,
      "loss": 1.1454,
      "step": 966
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5213294625282288,
      "learning_rate": 0.0001959761975271545,
      "loss": 1.171,
      "step": 967
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5557194352149963,
      "learning_rate": 0.00019596775264737126,
      "loss": 1.2047,
      "step": 968
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5759382843971252,
      "learning_rate": 0.00019595929909744494,
      "loss": 1.1048,
      "step": 969
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5198878645896912,
      "learning_rate": 0.00019595083687813926,
      "loss": 1.044,
      "step": 970
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5648888349533081,
      "learning_rate": 0.00019594236599021874,
      "loss": 1.1299,
      "step": 971
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5540794730186462,
      "learning_rate": 0.00019593388643444865,
      "loss": 1.1495,
      "step": 972
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5652869939804077,
      "learning_rate": 0.00019592539821159513,
      "loss": 0.9845,
      "step": 973
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6016077399253845,
      "learning_rate": 0.000195916901322425,
      "loss": 1.0299,
      "step": 974
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6304303407669067,
      "learning_rate": 0.00019590839576770592,
      "loss": 1.1451,
      "step": 975
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5486611127853394,
      "learning_rate": 0.00019589988154820632,
      "loss": 1.1388,
      "step": 976
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5753059983253479,
      "learning_rate": 0.0001958913586646954,
      "loss": 1.0921,
      "step": 977
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.49512913823127747,
      "learning_rate": 0.00019588282711794315,
      "loss": 1.0276,
      "step": 978
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5401121377944946,
      "learning_rate": 0.00019587428690872035,
      "loss": 0.9792,
      "step": 979
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5886608362197876,
      "learning_rate": 0.00019586573803779858,
      "loss": 1.0626,
      "step": 980
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4967743158340454,
      "learning_rate": 0.00019585718050595019,
      "loss": 1.1254,
      "step": 981
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.49494069814682007,
      "learning_rate": 0.00019584861431394822,
      "loss": 1.1964,
      "step": 982
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5207960605621338,
      "learning_rate": 0.0001958400394625667,
      "loss": 1.1442,
      "step": 983
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5062996745109558,
      "learning_rate": 0.00019583145595258023,
      "loss": 1.1793,
      "step": 984
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5631035566329956,
      "learning_rate": 0.00019582286378476432,
      "loss": 1.4415,
      "step": 985
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.568055272102356,
      "learning_rate": 0.00019581426295989525,
      "loss": 1.2234,
      "step": 986
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5561823844909668,
      "learning_rate": 0.00019580565347874997,
      "loss": 1.2309,
      "step": 987
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5685033202171326,
      "learning_rate": 0.00019579703534210635,
      "loss": 1.3636,
      "step": 988
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5274617671966553,
      "learning_rate": 0.00019578840855074303,
      "loss": 1.0681,
      "step": 989
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5587525963783264,
      "learning_rate": 0.00019577977310543932,
      "loss": 1.1667,
      "step": 990
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5266395211219788,
      "learning_rate": 0.00019577112900697542,
      "loss": 1.2668,
      "step": 991
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5371447801589966,
      "learning_rate": 0.00019576247625613226,
      "loss": 1.2036,
      "step": 992
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6547426581382751,
      "learning_rate": 0.00019575381485369158,
      "loss": 1.1855,
      "step": 993
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4874076843261719,
      "learning_rate": 0.00019574514480043587,
      "loss": 1.1199,
      "step": 994
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5545238852500916,
      "learning_rate": 0.0001957364660971485,
      "loss": 1.1382,
      "step": 995
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5526460409164429,
      "learning_rate": 0.00019572777874461342,
      "loss": 1.0697,
      "step": 996
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6076409816741943,
      "learning_rate": 0.00019571908274361555,
      "loss": 1.3189,
      "step": 997
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5631090998649597,
      "learning_rate": 0.00019571037809494048,
      "loss": 1.2412,
      "step": 998
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5299128890037537,
      "learning_rate": 0.0001957016647993747,
      "loss": 1.0527,
      "step": 999
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5710862874984741,
      "learning_rate": 0.00019569294285770537,
      "loss": 1.0326,
      "step": 1000
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5347865223884583,
      "learning_rate": 0.00019568421227072046,
      "loss": 1.0002,
      "step": 1001
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6362932324409485,
      "learning_rate": 0.0001956754730392087,
      "loss": 1.1981,
      "step": 1002
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5342056155204773,
      "learning_rate": 0.00019566672516395972,
      "loss": 1.1775,
      "step": 1003
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6062530279159546,
      "learning_rate": 0.00019565796864576377,
      "loss": 1.1539,
      "step": 1004
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6102681159973145,
      "learning_rate": 0.00019564920348541192,
      "loss": 1.0816,
      "step": 1005
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5745230913162231,
      "learning_rate": 0.00019564042968369614,
      "loss": 0.9836,
      "step": 1006
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5198560953140259,
      "learning_rate": 0.00019563164724140903,
      "loss": 1.003,
      "step": 1007
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6608436703681946,
      "learning_rate": 0.0001956228561593441,
      "loss": 1.2068,
      "step": 1008
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6578373908996582,
      "learning_rate": 0.00019561405643829549,
      "loss": 1.2482,
      "step": 1009
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5678254961967468,
      "learning_rate": 0.00019560524807905828,
      "loss": 1.0239,
      "step": 1010
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5507716536521912,
      "learning_rate": 0.00019559643108242817,
      "loss": 1.0791,
      "step": 1011
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5153545141220093,
      "learning_rate": 0.00019558760544920184,
      "loss": 1.1079,
      "step": 1012
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5246258974075317,
      "learning_rate": 0.0001955787711801765,
      "loss": 1.0968,
      "step": 1013
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5086669325828552,
      "learning_rate": 0.00019556992827615043,
      "loss": 1.0082,
      "step": 1014
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5282449722290039,
      "learning_rate": 0.00019556107673792238,
      "loss": 1.04,
      "step": 1015
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5100098848342896,
      "learning_rate": 0.00019555221656629218,
      "loss": 1.106,
      "step": 1016
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5582842826843262,
      "learning_rate": 0.0001955433477620602,
      "loss": 1.1453,
      "step": 1017
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5469565391540527,
      "learning_rate": 0.0001955344703260277,
      "loss": 1.0879,
      "step": 1018
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5286270976066589,
      "learning_rate": 0.00019552558425899673,
      "loss": 1.0567,
      "step": 1019
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5991113185882568,
      "learning_rate": 0.0001955166895617701,
      "loss": 1.1594,
      "step": 1020
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.514499843120575,
      "learning_rate": 0.00019550778623515135,
      "loss": 1.0381,
      "step": 1021
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5724852085113525,
      "learning_rate": 0.00019549887427994494,
      "loss": 1.1264,
      "step": 1022
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5247836112976074,
      "learning_rate": 0.00019548995369695588,
      "loss": 1.0393,
      "step": 1023
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5243775248527527,
      "learning_rate": 0.0001954810244869902,
      "loss": 0.9849,
      "step": 1024
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5602267980575562,
      "learning_rate": 0.00019547208665085457,
      "loss": 1.1284,
      "step": 1025
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5384296178817749,
      "learning_rate": 0.00019546314018935648,
      "loss": 1.1361,
      "step": 1026
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5097274780273438,
      "learning_rate": 0.00019545418510330417,
      "loss": 0.7718,
      "step": 1027
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.529334306716919,
      "learning_rate": 0.0001954452213935067,
      "loss": 0.9913,
      "step": 1028
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6140390634536743,
      "learning_rate": 0.00019543624906077388,
      "loss": 1.1762,
      "step": 1029
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5565348863601685,
      "learning_rate": 0.00019542726810591634,
      "loss": 1.0107,
      "step": 1030
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6203579306602478,
      "learning_rate": 0.0001954182785297454,
      "loss": 0.9787,
      "step": 1031
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.541043758392334,
      "learning_rate": 0.00019540928033307328,
      "loss": 1.0197,
      "step": 1032
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5186806917190552,
      "learning_rate": 0.00019540027351671287,
      "loss": 1.0029,
      "step": 1033
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5269144177436829,
      "learning_rate": 0.0001953912580814779,
      "loss": 1.1005,
      "step": 1034
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5349538326263428,
      "learning_rate": 0.00019538223402818287,
      "loss": 1.1029,
      "step": 1035
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5993475914001465,
      "learning_rate": 0.00019537320135764308,
      "loss": 1.1197,
      "step": 1036
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.525951623916626,
      "learning_rate": 0.0001953641600706745,
      "loss": 0.9929,
      "step": 1037
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6319278478622437,
      "learning_rate": 0.00019535511016809405,
      "loss": 1.053,
      "step": 1038
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5484718680381775,
      "learning_rate": 0.0001953460516507193,
      "loss": 1.0309,
      "step": 1039
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.649655818939209,
      "learning_rate": 0.00019533698451936856,
      "loss": 1.1151,
      "step": 1040
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5016194581985474,
      "learning_rate": 0.00019532790877486113,
      "loss": 1.0153,
      "step": 1041
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6027005910873413,
      "learning_rate": 0.0001953188244180169,
      "loss": 0.9985,
      "step": 1042
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.526719331741333,
      "learning_rate": 0.00019530973144965652,
      "loss": 0.9503,
      "step": 1043
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6107224822044373,
      "learning_rate": 0.0001953006298706016,
      "loss": 1.0747,
      "step": 1044
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5502063035964966,
      "learning_rate": 0.0001952915196816743,
      "loss": 1.0741,
      "step": 1045
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5427907109260559,
      "learning_rate": 0.00019528240088369778,
      "loss": 0.8883,
      "step": 1046
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5658277869224548,
      "learning_rate": 0.00019527327347749582,
      "loss": 1.1569,
      "step": 1047
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5371255874633789,
      "learning_rate": 0.00019526413746389306,
      "loss": 1.4064,
      "step": 1048
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.572701632976532,
      "learning_rate": 0.00019525499284371483,
      "loss": 1.3042,
      "step": 1049
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6183245182037354,
      "learning_rate": 0.00019524583961778733,
      "loss": 1.0634,
      "step": 1050
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5503761172294617,
      "learning_rate": 0.00019523667778693752,
      "loss": 0.9387,
      "step": 1051
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.553078293800354,
      "learning_rate": 0.0001952275073519931,
      "loss": 0.9584,
      "step": 1052
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6011327505111694,
      "learning_rate": 0.00019521832831378256,
      "loss": 1.0545,
      "step": 1053
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5534068942070007,
      "learning_rate": 0.00019520914067313515,
      "loss": 0.9852,
      "step": 1054
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5138425230979919,
      "learning_rate": 0.000195199944430881,
      "loss": 0.9858,
      "step": 1055
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5652523040771484,
      "learning_rate": 0.00019519073958785088,
      "loss": 1.0748,
      "step": 1056
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5925012230873108,
      "learning_rate": 0.0001951815261448764,
      "loss": 1.0663,
      "step": 1057
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6615341305732727,
      "learning_rate": 0.00019517230410278995,
      "loss": 1.2507,
      "step": 1058
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5900757908821106,
      "learning_rate": 0.0001951630734624247,
      "loss": 1.1842,
      "step": 1059
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6076449155807495,
      "learning_rate": 0.00019515383422461454,
      "loss": 1.0514,
      "step": 1060
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5401352643966675,
      "learning_rate": 0.00019514458639019426,
      "loss": 1.036,
      "step": 1061
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5284707546234131,
      "learning_rate": 0.00019513532995999927,
      "loss": 1.1178,
      "step": 1062
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.606998860836029,
      "learning_rate": 0.0001951260649348659,
      "loss": 1.1377,
      "step": 1063
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5934026837348938,
      "learning_rate": 0.00019511679131563114,
      "loss": 1.1728,
      "step": 1064
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5732269883155823,
      "learning_rate": 0.00019510750910313286,
      "loss": 0.9816,
      "step": 1065
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.492980033159256,
      "learning_rate": 0.00019509821829820964,
      "loss": 1.1011,
      "step": 1066
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6220079064369202,
      "learning_rate": 0.0001950889189017008,
      "loss": 1.1738,
      "step": 1067
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5315749645233154,
      "learning_rate": 0.00019507961091444655,
      "loss": 1.1742,
      "step": 1068
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5375043153762817,
      "learning_rate": 0.00019507029433728782,
      "loss": 0.9013,
      "step": 1069
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5389807820320129,
      "learning_rate": 0.00019506096917106626,
      "loss": 1.1154,
      "step": 1070
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5907564759254456,
      "learning_rate": 0.00019505163541662433,
      "loss": 1.2198,
      "step": 1071
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5677747130393982,
      "learning_rate": 0.0001950422930748054,
      "loss": 1.0026,
      "step": 1072
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4908526539802551,
      "learning_rate": 0.00019503294214645337,
      "loss": 1.1098,
      "step": 1073
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6064789295196533,
      "learning_rate": 0.00019502358263241308,
      "loss": 1.1671,
      "step": 1074
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5308496356010437,
      "learning_rate": 0.00019501421453353016,
      "loss": 1.1727,
      "step": 1075
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6154651045799255,
      "learning_rate": 0.0001950048378506509,
      "loss": 1.2673,
      "step": 1076
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5188403725624084,
      "learning_rate": 0.00019499545258462245,
      "loss": 1.0785,
      "step": 1077
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5489109754562378,
      "learning_rate": 0.00019498605873629273,
      "loss": 1.0982,
      "step": 1078
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5472975373268127,
      "learning_rate": 0.00019497665630651044,
      "loss": 0.9751,
      "step": 1079
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5919507145881653,
      "learning_rate": 0.00019496724529612498,
      "loss": 1.0141,
      "step": 1080
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6334194540977478,
      "learning_rate": 0.00019495782570598663,
      "loss": 1.381,
      "step": 1081
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5649259090423584,
      "learning_rate": 0.00019494839753694633,
      "loss": 1.2031,
      "step": 1082
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.532267153263092,
      "learning_rate": 0.00019493896078985596,
      "loss": 1.1422,
      "step": 1083
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5439337491989136,
      "learning_rate": 0.000194929515465568,
      "loss": 1.0401,
      "step": 1084
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.59439617395401,
      "learning_rate": 0.00019492006156493583,
      "loss": 1.0084,
      "step": 1085
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5659463405609131,
      "learning_rate": 0.0001949105990888135,
      "loss": 1.194,
      "step": 1086
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.504042387008667,
      "learning_rate": 0.00019490112803805597,
      "loss": 1.1048,
      "step": 1087
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4883994460105896,
      "learning_rate": 0.0001948916484135188,
      "loss": 0.938,
      "step": 1088
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5720723867416382,
      "learning_rate": 0.00019488216021605853,
      "loss": 1.1989,
      "step": 1089
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5613930821418762,
      "learning_rate": 0.00019487266344653224,
      "loss": 1.0445,
      "step": 1090
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5611872673034668,
      "learning_rate": 0.00019486315810579802,
      "loss": 1.0526,
      "step": 1091
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5774508714675903,
      "learning_rate": 0.00019485364419471454,
      "loss": 1.299,
      "step": 1092
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5790034532546997,
      "learning_rate": 0.0001948441217141414,
      "loss": 1.0856,
      "step": 1093
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6689309477806091,
      "learning_rate": 0.00019483459066493887,
      "loss": 1.1007,
      "step": 1094
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5516353249549866,
      "learning_rate": 0.00019482505104796802,
      "loss": 1.1348,
      "step": 1095
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5340696573257446,
      "learning_rate": 0.00019481550286409068,
      "loss": 1.1177,
      "step": 1096
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5577419400215149,
      "learning_rate": 0.00019480594611416953,
      "loss": 1.0439,
      "step": 1097
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.536791980266571,
      "learning_rate": 0.00019479638079906793,
      "loss": 1.0458,
      "step": 1098
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6502873301506042,
      "learning_rate": 0.0001947868069196501,
      "loss": 1.0867,
      "step": 1099
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5620083212852478,
      "learning_rate": 0.0001947772244767809,
      "loss": 1.0497,
      "step": 1100
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5127768516540527,
      "learning_rate": 0.00019476763347132608,
      "loss": 1.0542,
      "step": 1101
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.659235417842865,
      "learning_rate": 0.0001947580339041522,
      "loss": 1.2275,
      "step": 1102
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6155188083648682,
      "learning_rate": 0.00019474842577612645,
      "loss": 1.0007,
      "step": 1103
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5258010029792786,
      "learning_rate": 0.0001947388090881169,
      "loss": 1.0907,
      "step": 1104
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5436065793037415,
      "learning_rate": 0.00019472918384099233,
      "loss": 0.8972,
      "step": 1105
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5574349164962769,
      "learning_rate": 0.0001947195500356224,
      "loss": 1.0799,
      "step": 1106
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5468463897705078,
      "learning_rate": 0.00019470990767287737,
      "loss": 1.0913,
      "step": 1107
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5177785754203796,
      "learning_rate": 0.00019470025675362847,
      "loss": 1.0303,
      "step": 1108
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4689900279045105,
      "learning_rate": 0.00019469059727874756,
      "loss": 1.0365,
      "step": 1109
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5466490983963013,
      "learning_rate": 0.00019468092924910732,
      "loss": 1.1891,
      "step": 1110
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5339245796203613,
      "learning_rate": 0.00019467125266558119,
      "loss": 1.0907,
      "step": 1111
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5850642323493958,
      "learning_rate": 0.00019466156752904343,
      "loss": 1.1258,
      "step": 1112
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5640799403190613,
      "learning_rate": 0.000194651873840369,
      "loss": 1.0536,
      "step": 1113
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5763959884643555,
      "learning_rate": 0.0001946421716004337,
      "loss": 1.1633,
      "step": 1114
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5479110479354858,
      "learning_rate": 0.00019463246081011403,
      "loss": 1.1711,
      "step": 1115
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5160706043243408,
      "learning_rate": 0.00019462274147028738,
      "loss": 1.0581,
      "step": 1116
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.620171844959259,
      "learning_rate": 0.00019461301358183176,
      "loss": 1.0797,
      "step": 1117
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6158525347709656,
      "learning_rate": 0.00019460327714562606,
      "loss": 1.1704,
      "step": 1118
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5333216190338135,
      "learning_rate": 0.0001945935321625499,
      "loss": 1.1178,
      "step": 1119
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4954620599746704,
      "learning_rate": 0.00019458377863348374,
      "loss": 1.2292,
      "step": 1120
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.573711633682251,
      "learning_rate": 0.00019457401655930868,
      "loss": 1.0373,
      "step": 1121
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.513664186000824,
      "learning_rate": 0.0001945642459409067,
      "loss": 0.8973,
      "step": 1122
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5909994840621948,
      "learning_rate": 0.0001945544667791605,
      "loss": 1.196,
      "step": 1123
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6800118088722229,
      "learning_rate": 0.00019454467907495363,
      "loss": 1.1782,
      "step": 1124
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6476559042930603,
      "learning_rate": 0.00019453488282917031,
      "loss": 1.1332,
      "step": 1125
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5529746413230896,
      "learning_rate": 0.00019452507804269557,
      "loss": 1.1508,
      "step": 1126
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5596948266029358,
      "learning_rate": 0.00019451526471641527,
      "loss": 1.1156,
      "step": 1127
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6035382151603699,
      "learning_rate": 0.0001945054428512159,
      "loss": 1.1644,
      "step": 1128
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5915977954864502,
      "learning_rate": 0.00019449561244798486,
      "loss": 1.1096,
      "step": 1129
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.511471688747406,
      "learning_rate": 0.00019448577350761028,
      "loss": 0.9777,
      "step": 1130
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5073311924934387,
      "learning_rate": 0.00019447592603098103,
      "loss": 0.9382,
      "step": 1131
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5543911457061768,
      "learning_rate": 0.0001944660700189868,
      "loss": 0.9666,
      "step": 1132
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5058267712593079,
      "learning_rate": 0.00019445620547251798,
      "loss": 1.109,
      "step": 1133
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5585498809814453,
      "learning_rate": 0.00019444633239246585,
      "loss": 1.2121,
      "step": 1134
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5616839528083801,
      "learning_rate": 0.0001944364507797223,
      "loss": 1.052,
      "step": 1135
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5528122782707214,
      "learning_rate": 0.00019442656063518014,
      "loss": 1.1647,
      "step": 1136
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5317437648773193,
      "learning_rate": 0.00019441666195973287,
      "loss": 1.2549,
      "step": 1137
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5605677366256714,
      "learning_rate": 0.00019440675475427478,
      "loss": 1.1905,
      "step": 1138
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6127129197120667,
      "learning_rate": 0.00019439683901970094,
      "loss": 1.1418,
      "step": 1139
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6533770561218262,
      "learning_rate": 0.00019438691475690714,
      "loss": 1.1403,
      "step": 1140
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5641628503799438,
      "learning_rate": 0.00019437698196679,
      "loss": 1.134,
      "step": 1141
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6538822054862976,
      "learning_rate": 0.00019436704065024693,
      "loss": 1.1597,
      "step": 1142
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5727140307426453,
      "learning_rate": 0.00019435709080817604,
      "loss": 0.9952,
      "step": 1143
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5956428647041321,
      "learning_rate": 0.00019434713244147623,
      "loss": 1.0015,
      "step": 1144
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6197925806045532,
      "learning_rate": 0.00019433716555104723,
      "loss": 1.1185,
      "step": 1145
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5413385033607483,
      "learning_rate": 0.00019432719013778938,
      "loss": 1.1774,
      "step": 1146
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5331758260726929,
      "learning_rate": 0.00019431720620260404,
      "loss": 1.222,
      "step": 1147
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5929687023162842,
      "learning_rate": 0.00019430721374639312,
      "loss": 0.971,
      "step": 1148
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6004317402839661,
      "learning_rate": 0.00019429721277005942,
      "loss": 1.0906,
      "step": 1149
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5036609768867493,
      "learning_rate": 0.00019428720327450649,
      "loss": 1.0425,
      "step": 1150
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5406267046928406,
      "learning_rate": 0.00019427718526063856,
      "loss": 1.0313,
      "step": 1151
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5303625464439392,
      "learning_rate": 0.0001942671587293607,
      "loss": 0.9847,
      "step": 1152
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4866414964199066,
      "learning_rate": 0.00019425712368157885,
      "loss": 1.0592,
      "step": 1153
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4721148908138275,
      "learning_rate": 0.00019424708011819952,
      "loss": 0.9647,
      "step": 1154
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.551730215549469,
      "learning_rate": 0.00019423702804013013,
      "loss": 1.1571,
      "step": 1155
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5729138851165771,
      "learning_rate": 0.00019422696744827883,
      "loss": 0.969,
      "step": 1156
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5567776560783386,
      "learning_rate": 0.00019421689834355456,
      "loss": 0.9652,
      "step": 1157
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5598708391189575,
      "learning_rate": 0.00019420682072686696,
      "loss": 1.1677,
      "step": 1158
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5344229340553284,
      "learning_rate": 0.0001941967345991265,
      "loss": 1.0504,
      "step": 1159
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5083239078521729,
      "learning_rate": 0.00019418663996124443,
      "loss": 1.1487,
      "step": 1160
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5367764234542847,
      "learning_rate": 0.00019417653681413267,
      "loss": 1.0572,
      "step": 1161
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5423426032066345,
      "learning_rate": 0.0001941664251587041,
      "loss": 1.1104,
      "step": 1162
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6547486782073975,
      "learning_rate": 0.0001941563049958721,
      "loss": 1.0926,
      "step": 1163
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.535735011100769,
      "learning_rate": 0.00019414617632655115,
      "loss": 0.9657,
      "step": 1164
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5491278171539307,
      "learning_rate": 0.0001941360391516562,
      "loss": 1.0038,
      "step": 1165
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6106647253036499,
      "learning_rate": 0.0001941258934721031,
      "loss": 1.1907,
      "step": 1166
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5045748949050903,
      "learning_rate": 0.00019411573928880844,
      "loss": 0.9276,
      "step": 1167
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5820029377937317,
      "learning_rate": 0.0001941055766026896,
      "loss": 1.0474,
      "step": 1168
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5021175146102905,
      "learning_rate": 0.0001940954054146648,
      "loss": 0.9492,
      "step": 1169
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5358083844184875,
      "learning_rate": 0.00019408522572565286,
      "loss": 0.9982,
      "step": 1170
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.667941153049469,
      "learning_rate": 0.00019407503753657345,
      "loss": 1.1492,
      "step": 1171
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5163488984107971,
      "learning_rate": 0.00019406484084834712,
      "loss": 0.9993,
      "step": 1172
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5500897169113159,
      "learning_rate": 0.00019405463566189496,
      "loss": 0.9547,
      "step": 1173
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4756630063056946,
      "learning_rate": 0.00019404442197813902,
      "loss": 0.778,
      "step": 1174
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6951075196266174,
      "learning_rate": 0.00019403419979800203,
      "loss": 1.1922,
      "step": 1175
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6551975011825562,
      "learning_rate": 0.00019402396912240752,
      "loss": 1.141,
      "step": 1176
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5312485694885254,
      "learning_rate": 0.00019401372995227976,
      "loss": 1.0688,
      "step": 1177
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.572547197341919,
      "learning_rate": 0.00019400348228854378,
      "loss": 1.0993,
      "step": 1178
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5445220470428467,
      "learning_rate": 0.00019399322613212545,
      "loss": 1.0595,
      "step": 1179
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.49804481863975525,
      "learning_rate": 0.0001939829614839513,
      "loss": 1.1169,
      "step": 1180
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5678784251213074,
      "learning_rate": 0.00019397268834494874,
      "loss": 1.1402,
      "step": 1181
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.511607825756073,
      "learning_rate": 0.00019396240671604585,
      "loss": 1.1218,
      "step": 1182
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5115561485290527,
      "learning_rate": 0.00019395211659817152,
      "loss": 0.9857,
      "step": 1183
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4995981454849243,
      "learning_rate": 0.0001939418179922554,
      "loss": 1.1512,
      "step": 1184
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5507586598396301,
      "learning_rate": 0.00019393151089922795,
      "loss": 1.0724,
      "step": 1185
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5785311460494995,
      "learning_rate": 0.0001939211953200203,
      "loss": 1.1254,
      "step": 1186
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5976409316062927,
      "learning_rate": 0.00019391087125556446,
      "loss": 1.191,
      "step": 1187
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6236246824264526,
      "learning_rate": 0.00019390053870679312,
      "loss": 1.2168,
      "step": 1188
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5489307641983032,
      "learning_rate": 0.00019389019767463976,
      "loss": 1.1693,
      "step": 1189
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5228361487388611,
      "learning_rate": 0.00019387984816003867,
      "loss": 0.9666,
      "step": 1190
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6135544180870056,
      "learning_rate": 0.0001938694901639248,
      "loss": 0.9748,
      "step": 1191
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5711922645568848,
      "learning_rate": 0.00019385912368723401,
      "loss": 1.129,
      "step": 1192
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5178403258323669,
      "learning_rate": 0.00019384874873090282,
      "loss": 1.0801,
      "step": 1193
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5309805870056152,
      "learning_rate": 0.00019383836529586853,
      "loss": 1.2445,
      "step": 1194
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5460125207901001,
      "learning_rate": 0.00019382797338306928,
      "loss": 0.9797,
      "step": 1195
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5426638126373291,
      "learning_rate": 0.00019381757299344386,
      "loss": 1.181,
      "step": 1196
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4527784585952759,
      "learning_rate": 0.00019380716412793194,
      "loss": 0.7509,
      "step": 1197
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.49321427941322327,
      "learning_rate": 0.00019379674678747386,
      "loss": 0.9967,
      "step": 1198
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5492578744888306,
      "learning_rate": 0.0001937863209730108,
      "loss": 0.9134,
      "step": 1199
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5656059384346008,
      "learning_rate": 0.00019377588668548462,
      "loss": 1.2179,
      "step": 1200
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5511481761932373,
      "learning_rate": 0.00019376544392583808,
      "loss": 1.2019,
      "step": 1201
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5833784341812134,
      "learning_rate": 0.00019375499269501457,
      "loss": 1.2725,
      "step": 1202
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5577043890953064,
      "learning_rate": 0.0001937445329939583,
      "loss": 1.0354,
      "step": 1203
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5270592570304871,
      "learning_rate": 0.00019373406482361424,
      "loss": 1.083,
      "step": 1204
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.652744472026825,
      "learning_rate": 0.00019372358818492815,
      "loss": 1.2333,
      "step": 1205
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5374244451522827,
      "learning_rate": 0.00019371310307884654,
      "loss": 1.1342,
      "step": 1206
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5085197687149048,
      "learning_rate": 0.00019370260950631667,
      "loss": 1.0782,
      "step": 1207
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5113154053688049,
      "learning_rate": 0.00019369210746828658,
      "loss": 1.0251,
      "step": 1208
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5512399673461914,
      "learning_rate": 0.00019368159696570505,
      "loss": 1.0595,
      "step": 1209
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.533384382724762,
      "learning_rate": 0.00019367107799952166,
      "loss": 1.2031,
      "step": 1210
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4948982894420624,
      "learning_rate": 0.00019366055057068674,
      "loss": 1.0376,
      "step": 1211
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5904738306999207,
      "learning_rate": 0.00019365001468015137,
      "loss": 1.1022,
      "step": 1212
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6015186309814453,
      "learning_rate": 0.00019363947032886744,
      "loss": 1.2163,
      "step": 1213
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5421590209007263,
      "learning_rate": 0.00019362891751778755,
      "loss": 1.0019,
      "step": 1214
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5972522497177124,
      "learning_rate": 0.00019361835624786506,
      "loss": 1.0744,
      "step": 1215
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5512943267822266,
      "learning_rate": 0.00019360778652005416,
      "loss": 1.0013,
      "step": 1216
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6016600728034973,
      "learning_rate": 0.00019359720833530977,
      "loss": 1.1387,
      "step": 1217
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5378842353820801,
      "learning_rate": 0.00019358662169458753,
      "loss": 1.0681,
      "step": 1218
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5843987464904785,
      "learning_rate": 0.00019357602659884394,
      "loss": 1.0471,
      "step": 1219
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5850900411605835,
      "learning_rate": 0.00019356542304903613,
      "loss": 1.0736,
      "step": 1220
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5504021644592285,
      "learning_rate": 0.00019355481104612213,
      "loss": 1.06,
      "step": 1221
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7328195571899414,
      "learning_rate": 0.00019354419059106065,
      "loss": 1.2731,
      "step": 1222
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5825977921485901,
      "learning_rate": 0.00019353356168481123,
      "loss": 1.0629,
      "step": 1223
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.572723388671875,
      "learning_rate": 0.00019352292432833407,
      "loss": 0.8973,
      "step": 1224
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.526881217956543,
      "learning_rate": 0.00019351227852259023,
      "loss": 0.9208,
      "step": 1225
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5763203501701355,
      "learning_rate": 0.0001935016242685415,
      "loss": 1.1306,
      "step": 1226
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5341783761978149,
      "learning_rate": 0.00019349096156715039,
      "loss": 1.143,
      "step": 1227
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5604948997497559,
      "learning_rate": 0.0001934802904193803,
      "loss": 1.2042,
      "step": 1228
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5423800349235535,
      "learning_rate": 0.00019346961082619522,
      "loss": 0.9875,
      "step": 1229
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5558487176895142,
      "learning_rate": 0.00019345892278856,
      "loss": 0.8975,
      "step": 1230
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.48725664615631104,
      "learning_rate": 0.0001934482263074403,
      "loss": 0.9847,
      "step": 1231
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7110666036605835,
      "learning_rate": 0.00019343752138380245,
      "loss": 1.2409,
      "step": 1232
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5547020435333252,
      "learning_rate": 0.0001934268080186136,
      "loss": 1.0696,
      "step": 1233
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6257910132408142,
      "learning_rate": 0.00019341608621284163,
      "loss": 1.1821,
      "step": 1234
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4753723740577698,
      "learning_rate": 0.00019340535596745515,
      "loss": 0.9012,
      "step": 1235
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5675914883613586,
      "learning_rate": 0.00019339461728342365,
      "loss": 1.1259,
      "step": 1236
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.47405683994293213,
      "learning_rate": 0.00019338387016171727,
      "loss": 1.0401,
      "step": 1237
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6489463448524475,
      "learning_rate": 0.00019337311460330695,
      "loss": 1.1318,
      "step": 1238
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.610980212688446,
      "learning_rate": 0.0001933623506091644,
      "loss": 1.273,
      "step": 1239
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.622636616230011,
      "learning_rate": 0.00019335157818026213,
      "loss": 0.9558,
      "step": 1240
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5120880603790283,
      "learning_rate": 0.00019334079731757328,
      "loss": 1.0607,
      "step": 1241
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5910044312477112,
      "learning_rate": 0.0001933300080220719,
      "loss": 1.182,
      "step": 1242
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6695639491081238,
      "learning_rate": 0.00019331921029473276,
      "loss": 1.2554,
      "step": 1243
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5685967206954956,
      "learning_rate": 0.00019330840413653131,
      "loss": 1.0963,
      "step": 1244
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5170540809631348,
      "learning_rate": 0.00019329758954844385,
      "loss": 1.0808,
      "step": 1245
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.561914324760437,
      "learning_rate": 0.00019328676653144743,
      "loss": 1.1171,
      "step": 1246
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5721147656440735,
      "learning_rate": 0.00019327593508651985,
      "loss": 1.0033,
      "step": 1247
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.616117000579834,
      "learning_rate": 0.00019326509521463967,
      "loss": 1.2112,
      "step": 1248
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.590916633605957,
      "learning_rate": 0.0001932542469167862,
      "loss": 1.2606,
      "step": 1249
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5014984011650085,
      "learning_rate": 0.00019324339019393948,
      "loss": 1.0264,
      "step": 1250
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.536072313785553,
      "learning_rate": 0.00019323252504708046,
      "loss": 0.925,
      "step": 1251
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6966559290885925,
      "learning_rate": 0.00019322165147719065,
      "loss": 1.2475,
      "step": 1252
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5732153058052063,
      "learning_rate": 0.00019321076948525248,
      "loss": 1.2821,
      "step": 1253
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5714349150657654,
      "learning_rate": 0.00019319987907224899,
      "loss": 0.9969,
      "step": 1254
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.49318331480026245,
      "learning_rate": 0.00019318898023916415,
      "loss": 1.1031,
      "step": 1255
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5395241975784302,
      "learning_rate": 0.00019317807298698257,
      "loss": 1.1954,
      "step": 1256
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5213454961776733,
      "learning_rate": 0.00019316715731668973,
      "loss": 1.0166,
      "step": 1257
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5459577441215515,
      "learning_rate": 0.0001931562332292717,
      "loss": 1.0798,
      "step": 1258
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5936806201934814,
      "learning_rate": 0.0001931453007257154,
      "loss": 1.1566,
      "step": 1259
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5617766976356506,
      "learning_rate": 0.0001931343598070086,
      "loss": 1.2553,
      "step": 1260
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6054962277412415,
      "learning_rate": 0.00019312341047413975,
      "loss": 1.1497,
      "step": 1261
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5415417551994324,
      "learning_rate": 0.000193112452728098,
      "loss": 1.0297,
      "step": 1262
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5991696119308472,
      "learning_rate": 0.00019310148656987333,
      "loss": 1.0435,
      "step": 1263
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6084467768669128,
      "learning_rate": 0.00019309051200045653,
      "loss": 1.1188,
      "step": 1264
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6937548518180847,
      "learning_rate": 0.000193079529020839,
      "loss": 1.1313,
      "step": 1265
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6291505098342896,
      "learning_rate": 0.00019306853763201303,
      "loss": 1.1127,
      "step": 1266
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5899692177772522,
      "learning_rate": 0.0001930575378349717,
      "loss": 1.0954,
      "step": 1267
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5272820591926575,
      "learning_rate": 0.0001930465296307087,
      "loss": 1.2213,
      "step": 1268
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5450857877731323,
      "learning_rate": 0.00019303551302021852,
      "loss": 1.1929,
      "step": 1269
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.527804434299469,
      "learning_rate": 0.00019302448800449654,
      "loss": 1.12,
      "step": 1270
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5344571471214294,
      "learning_rate": 0.00019301345458453878,
      "loss": 1.1377,
      "step": 1271
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5595630407333374,
      "learning_rate": 0.00019300241276134204,
      "loss": 1.1037,
      "step": 1272
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5389091968536377,
      "learning_rate": 0.00019299136253590388,
      "loss": 0.9875,
      "step": 1273
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4910154640674591,
      "learning_rate": 0.00019298030390922264,
      "loss": 1.0903,
      "step": 1274
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5613997578620911,
      "learning_rate": 0.0001929692368822974,
      "loss": 1.0653,
      "step": 1275
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5709872245788574,
      "learning_rate": 0.00019295816145612795,
      "loss": 1.0943,
      "step": 1276
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6014252305030823,
      "learning_rate": 0.000192947077631715,
      "loss": 1.1494,
      "step": 1277
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6745042204856873,
      "learning_rate": 0.00019293598541005982,
      "loss": 1.0209,
      "step": 1278
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5821531414985657,
      "learning_rate": 0.00019292488479216458,
      "loss": 1.1824,
      "step": 1279
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5291196703910828,
      "learning_rate": 0.00019291377577903212,
      "loss": 1.1749,
      "step": 1280
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6035329699516296,
      "learning_rate": 0.00019290265837166608,
      "loss": 1.1746,
      "step": 1281
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.582727313041687,
      "learning_rate": 0.0001928915325710709,
      "loss": 1.1319,
      "step": 1282
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5894931554794312,
      "learning_rate": 0.00019288039837825168,
      "loss": 1.1792,
      "step": 1283
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5747166275978088,
      "learning_rate": 0.00019286925579421434,
      "loss": 1.1439,
      "step": 1284
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5466634631156921,
      "learning_rate": 0.00019285810481996563,
      "loss": 1.0759,
      "step": 1285
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5804960131645203,
      "learning_rate": 0.00019284694545651285,
      "loss": 1.1445,
      "step": 1286
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5688068270683289,
      "learning_rate": 0.00019283577770486425,
      "loss": 1.2065,
      "step": 1287
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5299245715141296,
      "learning_rate": 0.00019282460156602878,
      "loss": 1.2183,
      "step": 1288
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5362908244132996,
      "learning_rate": 0.00019281341704101615,
      "loss": 1.1709,
      "step": 1289
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5553381443023682,
      "learning_rate": 0.0001928022241308368,
      "loss": 1.0166,
      "step": 1290
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5688846111297607,
      "learning_rate": 0.0001927910228365019,
      "loss": 1.1638,
      "step": 1291
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5960063934326172,
      "learning_rate": 0.00019277981315902348,
      "loss": 1.1656,
      "step": 1292
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5050798058509827,
      "learning_rate": 0.0001927685950994143,
      "loss": 1.0812,
      "step": 1293
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6014881730079651,
      "learning_rate": 0.00019275736865868778,
      "loss": 1.0961,
      "step": 1294
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5374804139137268,
      "learning_rate": 0.00019274613383785818,
      "loss": 1.181,
      "step": 1295
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5485164523124695,
      "learning_rate": 0.00019273489063794056,
      "loss": 1.2842,
      "step": 1296
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6008142232894897,
      "learning_rate": 0.00019272363905995058,
      "loss": 0.9858,
      "step": 1297
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5489171743392944,
      "learning_rate": 0.00019271237910490485,
      "loss": 0.9488,
      "step": 1298
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5818604826927185,
      "learning_rate": 0.00019270111077382057,
      "loss": 1.0722,
      "step": 1299
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5716251730918884,
      "learning_rate": 0.0001926898340677158,
      "loss": 1.103,
      "step": 1300
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5115184783935547,
      "learning_rate": 0.0001926785489876094,
      "loss": 1.1094,
      "step": 1301
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.571029007434845,
      "learning_rate": 0.00019266725553452075,
      "loss": 1.0811,
      "step": 1302
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5410216450691223,
      "learning_rate": 0.00019265595370947033,
      "loss": 1.0502,
      "step": 1303
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.567905068397522,
      "learning_rate": 0.00019264464351347903,
      "loss": 1.1384,
      "step": 1304
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5957036018371582,
      "learning_rate": 0.0001926333249475688,
      "loss": 0.9977,
      "step": 1305
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5629892945289612,
      "learning_rate": 0.00019262199801276214,
      "loss": 1.1238,
      "step": 1306
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5770018100738525,
      "learning_rate": 0.00019261066271008235,
      "loss": 1.205,
      "step": 1307
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5584204792976379,
      "learning_rate": 0.00019259931904055358,
      "loss": 1.3062,
      "step": 1308
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5548158288002014,
      "learning_rate": 0.0001925879670052006,
      "loss": 1.1269,
      "step": 1309
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5640186071395874,
      "learning_rate": 0.00019257660660504906,
      "loss": 1.107,
      "step": 1310
    },
    {
      "epoch": 0.5,
      "eval_loss": 0.9923078417778015,
      "eval_runtime": 17.9573,
      "eval_samples_per_second": 151.972,
      "eval_steps_per_second": 38.035,
      "step": 1310
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5832816362380981,
      "learning_rate": 0.00019256523784112525,
      "loss": 1.0619,
      "step": 1311
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5452430844306946,
      "learning_rate": 0.00019255386071445637,
      "loss": 1.253,
      "step": 1312
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5174736976623535,
      "learning_rate": 0.00019254247522607014,
      "loss": 1.0906,
      "step": 1313
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5985592007637024,
      "learning_rate": 0.0001925310813769953,
      "loss": 1.2273,
      "step": 1314
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5438377261161804,
      "learning_rate": 0.00019251967916826112,
      "loss": 1.0868,
      "step": 1315
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5200532674789429,
      "learning_rate": 0.0001925082686008978,
      "loss": 1.142,
      "step": 1316
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5317777395248413,
      "learning_rate": 0.0001924968496759362,
      "loss": 1.0676,
      "step": 1317
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6042464971542358,
      "learning_rate": 0.00019248542239440791,
      "loss": 1.1256,
      "step": 1318
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5626322031021118,
      "learning_rate": 0.00019247398675734537,
      "loss": 1.0569,
      "step": 1319
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.58858323097229,
      "learning_rate": 0.00019246254276578174,
      "loss": 1.2068,
      "step": 1320
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5406404733657837,
      "learning_rate": 0.00019245109042075086,
      "loss": 1.074,
      "step": 1321
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6096088290214539,
      "learning_rate": 0.00019243962972328746,
      "loss": 1.1028,
      "step": 1322
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.606028139591217,
      "learning_rate": 0.00019242816067442688,
      "loss": 1.0781,
      "step": 1323
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5860372185707092,
      "learning_rate": 0.00019241668327520532,
      "loss": 1.0578,
      "step": 1324
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5592902302742004,
      "learning_rate": 0.00019240519752665967,
      "loss": 1.2027,
      "step": 1325
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5435681939125061,
      "learning_rate": 0.00019239370342982764,
      "loss": 0.9271,
      "step": 1326
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5579676032066345,
      "learning_rate": 0.0001923822009857476,
      "loss": 1.0776,
      "step": 1327
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6110450029373169,
      "learning_rate": 0.0001923706901954588,
      "loss": 1.1934,
      "step": 1328
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5760442018508911,
      "learning_rate": 0.00019235917106000114,
      "loss": 1.1367,
      "step": 1329
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6102426648139954,
      "learning_rate": 0.00019234764358041533,
      "loss": 1.161,
      "step": 1330
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5880422592163086,
      "learning_rate": 0.00019233610775774276,
      "loss": 1.2247,
      "step": 1331
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5817332863807678,
      "learning_rate": 0.00019232456359302567,
      "loss": 1.2372,
      "step": 1332
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5220525860786438,
      "learning_rate": 0.000192313011087307,
      "loss": 1.181,
      "step": 1333
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5194857716560364,
      "learning_rate": 0.00019230145024163044,
      "loss": 0.9651,
      "step": 1334
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5529868602752686,
      "learning_rate": 0.00019228988105704048,
      "loss": 1.0857,
      "step": 1335
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5783809423446655,
      "learning_rate": 0.0001922783035345823,
      "loss": 1.1625,
      "step": 1336
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6270204186439514,
      "learning_rate": 0.00019226671767530186,
      "loss": 1.2083,
      "step": 1337
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5166578888893127,
      "learning_rate": 0.00019225512348024592,
      "loss": 1.1125,
      "step": 1338
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5585746765136719,
      "learning_rate": 0.00019224352095046192,
      "loss": 1.2018,
      "step": 1339
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5650426149368286,
      "learning_rate": 0.00019223191008699807,
      "loss": 0.9949,
      "step": 1340
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.536048412322998,
      "learning_rate": 0.00019222029089090335,
      "loss": 1.1063,
      "step": 1341
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5900925993919373,
      "learning_rate": 0.00019220866336322752,
      "loss": 1.1525,
      "step": 1342
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.48170042037963867,
      "learning_rate": 0.000192197027505021,
      "loss": 1.0483,
      "step": 1343
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6003119349479675,
      "learning_rate": 0.00019218538331733514,
      "loss": 1.0138,
      "step": 1344
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5208913683891296,
      "learning_rate": 0.00019217373080122177,
      "loss": 1.0371,
      "step": 1345
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5866532325744629,
      "learning_rate": 0.00019216206995773373,
      "loss": 1.1965,
      "step": 1346
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5789579153060913,
      "learning_rate": 0.00019215040078792453,
      "loss": 1.1518,
      "step": 1347
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5461020469665527,
      "learning_rate": 0.00019213872329284834,
      "loss": 1.1143,
      "step": 1348
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6083654761314392,
      "learning_rate": 0.00019212703747356017,
      "loss": 1.1624,
      "step": 1349
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.48819878697395325,
      "learning_rate": 0.00019211534333111582,
      "loss": 0.8658,
      "step": 1350
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.614565372467041,
      "learning_rate": 0.00019210364086657172,
      "loss": 1.0509,
      "step": 1351
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6060966849327087,
      "learning_rate": 0.0001920919300809852,
      "loss": 1.0878,
      "step": 1352
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4852564036846161,
      "learning_rate": 0.00019208021097541419,
      "loss": 0.9189,
      "step": 1353
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6118038892745972,
      "learning_rate": 0.00019206848355091748,
      "loss": 1.1956,
      "step": 1354
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5745032429695129,
      "learning_rate": 0.0001920567478085546,
      "loss": 1.1221,
      "step": 1355
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6071743965148926,
      "learning_rate": 0.00019204500374938577,
      "loss": 1.0735,
      "step": 1356
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.48076263070106506,
      "learning_rate": 0.000192033251374472,
      "loss": 1.0431,
      "step": 1357
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6090218424797058,
      "learning_rate": 0.00019202149068487504,
      "loss": 1.0099,
      "step": 1358
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5336397886276245,
      "learning_rate": 0.00019200972168165742,
      "loss": 1.0205,
      "step": 1359
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5435861945152283,
      "learning_rate": 0.00019199794436588243,
      "loss": 1.0421,
      "step": 1360
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6408600211143494,
      "learning_rate": 0.00019198615873861405,
      "loss": 1.1427,
      "step": 1361
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5759676098823547,
      "learning_rate": 0.00019197436480091704,
      "loss": 1.0883,
      "step": 1362
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5735880732536316,
      "learning_rate": 0.00019196256255385693,
      "loss": 1.0093,
      "step": 1363
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5333439111709595,
      "learning_rate": 0.0001919507519985,
      "loss": 1.1154,
      "step": 1364
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5859337449073792,
      "learning_rate": 0.00019193893313591324,
      "loss": 1.1154,
      "step": 1365
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5673715472221375,
      "learning_rate": 0.00019192710596716443,
      "loss": 0.9514,
      "step": 1366
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6896955370903015,
      "learning_rate": 0.00019191527049332207,
      "loss": 1.1815,
      "step": 1367
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5384703278541565,
      "learning_rate": 0.0001919034267154555,
      "loss": 0.9545,
      "step": 1368
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5984772443771362,
      "learning_rate": 0.00019189157463463463,
      "loss": 0.9701,
      "step": 1369
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.49715912342071533,
      "learning_rate": 0.00019187971425193028,
      "loss": 1.038,
      "step": 1370
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6104459166526794,
      "learning_rate": 0.00019186784556841398,
      "loss": 0.9182,
      "step": 1371
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5620800256729126,
      "learning_rate": 0.000191855968585158,
      "loss": 1.1721,
      "step": 1372
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5518665909767151,
      "learning_rate": 0.00019184408330323534,
      "loss": 0.9945,
      "step": 1373
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.596023440361023,
      "learning_rate": 0.00019183218972371974,
      "loss": 1.173,
      "step": 1374
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5669458508491516,
      "learning_rate": 0.0001918202878476858,
      "loss": 1.0848,
      "step": 1375
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.543956458568573,
      "learning_rate": 0.00019180837767620868,
      "loss": 1.1878,
      "step": 1376
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.509422242641449,
      "learning_rate": 0.00019179645921036452,
      "loss": 1.0451,
      "step": 1377
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5754774212837219,
      "learning_rate": 0.00019178453245123,
      "loss": 1.0036,
      "step": 1378
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6185034513473511,
      "learning_rate": 0.00019177259739988262,
      "loss": 1.2736,
      "step": 1379
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5642620325088501,
      "learning_rate": 0.0001917606540574007,
      "loss": 1.2801,
      "step": 1380
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5372741222381592,
      "learning_rate": 0.00019174870242486325,
      "loss": 1.2769,
      "step": 1381
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.575202465057373,
      "learning_rate": 0.00019173674250335,
      "loss": 1.1626,
      "step": 1382
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5753564238548279,
      "learning_rate": 0.0001917247742939415,
      "loss": 1.0149,
      "step": 1383
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5552440881729126,
      "learning_rate": 0.00019171279779771897,
      "loss": 1.0113,
      "step": 1384
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6205223798751831,
      "learning_rate": 0.00019170081301576444,
      "loss": 1.2073,
      "step": 1385
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6425143480300903,
      "learning_rate": 0.00019168881994916065,
      "loss": 1.1119,
      "step": 1386
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6084074378013611,
      "learning_rate": 0.00019167681859899114,
      "loss": 1.1926,
      "step": 1387
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4909365475177765,
      "learning_rate": 0.00019166480896634012,
      "loss": 1.0389,
      "step": 1388
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5355455875396729,
      "learning_rate": 0.0001916527910522926,
      "loss": 1.0243,
      "step": 1389
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5883185267448425,
      "learning_rate": 0.00019164076485793437,
      "loss": 1.3136,
      "step": 1390
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.536249577999115,
      "learning_rate": 0.0001916287303843519,
      "loss": 0.9112,
      "step": 1391
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5794864892959595,
      "learning_rate": 0.00019161668763263245,
      "loss": 1.204,
      "step": 1392
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.577343761920929,
      "learning_rate": 0.000191604636603864,
      "loss": 1.1141,
      "step": 1393
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4846179485321045,
      "learning_rate": 0.00019159257729913529,
      "loss": 1.0279,
      "step": 1394
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5625943541526794,
      "learning_rate": 0.00019158050971953582,
      "loss": 0.931,
      "step": 1395
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6180777549743652,
      "learning_rate": 0.00019156843386615582,
      "loss": 1.2183,
      "step": 1396
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5064194798469543,
      "learning_rate": 0.0001915563497400863,
      "loss": 1.0858,
      "step": 1397
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5856994390487671,
      "learning_rate": 0.00019154425734241895,
      "loss": 1.0773,
      "step": 1398
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5765694975852966,
      "learning_rate": 0.00019153215667424628,
      "loss": 1.0104,
      "step": 1399
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6079264283180237,
      "learning_rate": 0.0001915200477366615,
      "loss": 1.0882,
      "step": 1400
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5938848853111267,
      "learning_rate": 0.0001915079305307586,
      "loss": 1.1615,
      "step": 1401
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6206392049789429,
      "learning_rate": 0.00019149580505763226,
      "loss": 1.1099,
      "step": 1402
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5335953831672668,
      "learning_rate": 0.00019148367131837803,
      "loss": 0.9998,
      "step": 1403
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6532598733901978,
      "learning_rate": 0.00019147152931409207,
      "loss": 1.1109,
      "step": 1404
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6051193475723267,
      "learning_rate": 0.00019145937904587132,
      "loss": 1.0338,
      "step": 1405
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6793563365936279,
      "learning_rate": 0.00019144722051481355,
      "loss": 1.1387,
      "step": 1406
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5717383623123169,
      "learning_rate": 0.00019143505372201712,
      "loss": 1.167,
      "step": 1407
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5962218046188354,
      "learning_rate": 0.00019142287866858135,
      "loss": 1.1216,
      "step": 1408
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5622634887695312,
      "learning_rate": 0.00019141069535560612,
      "loss": 0.702,
      "step": 1409
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6235326528549194,
      "learning_rate": 0.0001913985037841921,
      "loss": 1.0167,
      "step": 1410
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5555641055107117,
      "learning_rate": 0.00019138630395544077,
      "loss": 1.146,
      "step": 1411
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.551306962966919,
      "learning_rate": 0.00019137409587045433,
      "loss": 1.0985,
      "step": 1412
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5754269957542419,
      "learning_rate": 0.00019136187953033566,
      "loss": 1.0251,
      "step": 1413
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5977479219436646,
      "learning_rate": 0.00019134965493618848,
      "loss": 1.1365,
      "step": 1414
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6409013867378235,
      "learning_rate": 0.00019133742208911719,
      "loss": 1.2431,
      "step": 1415
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5369458198547363,
      "learning_rate": 0.00019132518099022696,
      "loss": 1.1019,
      "step": 1416
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5461875796318054,
      "learning_rate": 0.00019131293164062368,
      "loss": 0.9547,
      "step": 1417
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5358739495277405,
      "learning_rate": 0.0001913006740414141,
      "loss": 1.0399,
      "step": 1418
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6357903480529785,
      "learning_rate": 0.00019128840819370555,
      "loss": 1.0574,
      "step": 1419
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5420630574226379,
      "learning_rate": 0.00019127613409860615,
      "loss": 1.08,
      "step": 1420
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6549668908119202,
      "learning_rate": 0.00019126385175722487,
      "loss": 1.1344,
      "step": 1421
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5863200426101685,
      "learning_rate": 0.00019125156117067132,
      "loss": 1.02,
      "step": 1422
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6721362471580505,
      "learning_rate": 0.0001912392623400559,
      "loss": 1.1862,
      "step": 1423
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5170490145683289,
      "learning_rate": 0.00019122695526648968,
      "loss": 0.9705,
      "step": 1424
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5132006406784058,
      "learning_rate": 0.00019121463995108463,
      "loss": 1.1416,
      "step": 1425
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5969366431236267,
      "learning_rate": 0.00019120231639495328,
      "loss": 1.2315,
      "step": 1426
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5861543416976929,
      "learning_rate": 0.00019118998459920902,
      "loss": 1.1351,
      "step": 1427
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5891702771186829,
      "learning_rate": 0.000191177644564966,
      "loss": 1.1291,
      "step": 1428
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5410522222518921,
      "learning_rate": 0.00019116529629333902,
      "loss": 0.9958,
      "step": 1429
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5884034633636475,
      "learning_rate": 0.00019115293978544368,
      "loss": 1.0791,
      "step": 1430
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6323250532150269,
      "learning_rate": 0.00019114057504239637,
      "loss": 1.0047,
      "step": 1431
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4916994273662567,
      "learning_rate": 0.00019112820206531412,
      "loss": 1.0319,
      "step": 1432
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5568423271179199,
      "learning_rate": 0.0001911158208553148,
      "loss": 1.2002,
      "step": 1433
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5141173601150513,
      "learning_rate": 0.00019110343141351695,
      "loss": 1.1509,
      "step": 1434
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5884588956832886,
      "learning_rate": 0.00019109103374103986,
      "loss": 1.2095,
      "step": 1435
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6008499264717102,
      "learning_rate": 0.00019107862783900368,
      "loss": 1.2357,
      "step": 1436
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5440402626991272,
      "learning_rate": 0.00019106621370852916,
      "loss": 1.1466,
      "step": 1437
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6503211855888367,
      "learning_rate": 0.00019105379135073784,
      "loss": 1.1827,
      "step": 1438
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6284806132316589,
      "learning_rate": 0.000191041360766752,
      "loss": 1.203,
      "step": 1439
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.563576877117157,
      "learning_rate": 0.00019102892195769472,
      "loss": 1.043,
      "step": 1440
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.629399836063385,
      "learning_rate": 0.0001910164749246897,
      "loss": 0.9591,
      "step": 1441
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5576425194740295,
      "learning_rate": 0.00019100401966886153,
      "loss": 1.1031,
      "step": 1442
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5660919547080994,
      "learning_rate": 0.00019099155619133545,
      "loss": 1.0291,
      "step": 1443
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5721517205238342,
      "learning_rate": 0.00019097908449323744,
      "loss": 0.9636,
      "step": 1444
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5158771276473999,
      "learning_rate": 0.00019096660457569432,
      "loss": 1.0108,
      "step": 1445
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5575356483459473,
      "learning_rate": 0.00019095411643983348,
      "loss": 0.9741,
      "step": 1446
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5737133622169495,
      "learning_rate": 0.0001909416200867832,
      "loss": 1.0725,
      "step": 1447
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5935455560684204,
      "learning_rate": 0.00019092911551767247,
      "loss": 1.2495,
      "step": 1448
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.618424654006958,
      "learning_rate": 0.000190916602733631,
      "loss": 1.0883,
      "step": 1449
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5599175095558167,
      "learning_rate": 0.0001909040817357892,
      "loss": 0.9976,
      "step": 1450
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6387348175048828,
      "learning_rate": 0.0001908915525252784,
      "loss": 1.1933,
      "step": 1451
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.507186233997345,
      "learning_rate": 0.00019087901510323038,
      "loss": 1.1166,
      "step": 1452
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.49330127239227295,
      "learning_rate": 0.00019086646947077794,
      "loss": 0.8951,
      "step": 1453
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5779321789741516,
      "learning_rate": 0.00019085391562905443,
      "loss": 1.1588,
      "step": 1454
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6197642087936401,
      "learning_rate": 0.00019084135357919408,
      "loss": 1.0809,
      "step": 1455
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5197485089302063,
      "learning_rate": 0.00019082878332233176,
      "loss": 1.0183,
      "step": 1456
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6561455130577087,
      "learning_rate": 0.00019081620485960317,
      "loss": 1.1886,
      "step": 1457
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5491870641708374,
      "learning_rate": 0.00019080361819214466,
      "loss": 1.1449,
      "step": 1458
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.48696836829185486,
      "learning_rate": 0.00019079102332109334,
      "loss": 1.0596,
      "step": 1459
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.62408447265625,
      "learning_rate": 0.00019077842024758717,
      "loss": 1.2558,
      "step": 1460
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6218500137329102,
      "learning_rate": 0.0001907658089727647,
      "loss": 1.1698,
      "step": 1461
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.558061420917511,
      "learning_rate": 0.0001907531894977653,
      "loss": 1.2256,
      "step": 1462
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5486068725585938,
      "learning_rate": 0.00019074056182372907,
      "loss": 1.2703,
      "step": 1463
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.501006543636322,
      "learning_rate": 0.00019072792595179684,
      "loss": 1.1317,
      "step": 1464
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.562553346157074,
      "learning_rate": 0.00019071528188311022,
      "loss": 0.9698,
      "step": 1465
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6265679001808167,
      "learning_rate": 0.0001907026296188115,
      "loss": 1.0556,
      "step": 1466
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.554051399230957,
      "learning_rate": 0.00019068996916004377,
      "loss": 1.1407,
      "step": 1467
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5753392577171326,
      "learning_rate": 0.0001906773005079508,
      "loss": 1.0468,
      "step": 1468
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5394361019134521,
      "learning_rate": 0.0001906646236636771,
      "loss": 1.12,
      "step": 1469
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5362756252288818,
      "learning_rate": 0.00019065193862836805,
      "loss": 1.0795,
      "step": 1470
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6187636256217957,
      "learning_rate": 0.00019063924540316956,
      "loss": 1.0331,
      "step": 1471
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5482291579246521,
      "learning_rate": 0.00019062654398922853,
      "loss": 1.0507,
      "step": 1472
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5710556507110596,
      "learning_rate": 0.00019061383438769228,
      "loss": 1.2762,
      "step": 1473
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6045145988464355,
      "learning_rate": 0.0001906011165997092,
      "loss": 1.1701,
      "step": 1474
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6226998567581177,
      "learning_rate": 0.00019058839062642823,
      "loss": 1.1354,
      "step": 1475
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5932821035385132,
      "learning_rate": 0.00019057565646899907,
      "loss": 1.2439,
      "step": 1476
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5586472153663635,
      "learning_rate": 0.00019056291412857218,
      "loss": 1.0227,
      "step": 1477
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6210311651229858,
      "learning_rate": 0.00019055016360629873,
      "loss": 1.237,
      "step": 1478
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5757390260696411,
      "learning_rate": 0.00019053740490333074,
      "loss": 1.2567,
      "step": 1479
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5853813290596008,
      "learning_rate": 0.00019052463802082082,
      "loss": 1.1551,
      "step": 1480
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5320205688476562,
      "learning_rate": 0.0001905118629599224,
      "loss": 0.8199,
      "step": 1481
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5871585607528687,
      "learning_rate": 0.00019049907972178964,
      "loss": 1.0795,
      "step": 1482
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6921295523643494,
      "learning_rate": 0.00019048628830757743,
      "loss": 1.2403,
      "step": 1483
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6031093597412109,
      "learning_rate": 0.00019047348871844145,
      "loss": 0.9655,
      "step": 1484
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6309522390365601,
      "learning_rate": 0.00019046068095553796,
      "loss": 1.0757,
      "step": 1485
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5803585648536682,
      "learning_rate": 0.00019044786502002414,
      "loss": 1.0815,
      "step": 1486
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.559934675693512,
      "learning_rate": 0.00019043504091305786,
      "loss": 1.1025,
      "step": 1487
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5299091339111328,
      "learning_rate": 0.00019042220863579767,
      "loss": 1.011,
      "step": 1488
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6164795756340027,
      "learning_rate": 0.00019040936818940287,
      "loss": 1.1809,
      "step": 1489
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6382726430892944,
      "learning_rate": 0.0001903965195750336,
      "loss": 1.0985,
      "step": 1490
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5797393918037415,
      "learning_rate": 0.00019038366279385056,
      "loss": 0.984,
      "step": 1491
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.516485869884491,
      "learning_rate": 0.00019037079784701538,
      "loss": 1.1137,
      "step": 1492
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5419387817382812,
      "learning_rate": 0.00019035792473569022,
      "loss": 1.1436,
      "step": 1493
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5802305936813354,
      "learning_rate": 0.00019034504346103823,
      "loss": 1.2381,
      "step": 1494
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5675755143165588,
      "learning_rate": 0.0001903321540242231,
      "loss": 1.029,
      "step": 1495
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6638576984405518,
      "learning_rate": 0.00019031925642640926,
      "loss": 1.0509,
      "step": 1496
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5375914573669434,
      "learning_rate": 0.00019030635066876204,
      "loss": 1.1761,
      "step": 1497
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5363042950630188,
      "learning_rate": 0.00019029343675244732,
      "loss": 0.9477,
      "step": 1498
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5889085531234741,
      "learning_rate": 0.00019028051467863185,
      "loss": 0.9483,
      "step": 1499
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6252742409706116,
      "learning_rate": 0.00019026758444848306,
      "loss": 1.1044,
      "step": 1500
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5877997279167175,
      "learning_rate": 0.00019025464606316908,
      "loss": 0.9745,
      "step": 1501
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5305163860321045,
      "learning_rate": 0.00019024169952385885,
      "loss": 0.9826,
      "step": 1502
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5609629154205322,
      "learning_rate": 0.000190228744831722,
      "loss": 0.953,
      "step": 1503
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5558766722679138,
      "learning_rate": 0.00019021578198792898,
      "loss": 1.1254,
      "step": 1504
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5775374174118042,
      "learning_rate": 0.0001902028109936508,
      "loss": 0.9384,
      "step": 1505
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5770506858825684,
      "learning_rate": 0.0001901898318500594,
      "loss": 1.051,
      "step": 1506
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5940860509872437,
      "learning_rate": 0.00019017684455832737,
      "loss": 1.1082,
      "step": 1507
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5455056428909302,
      "learning_rate": 0.000190163849119628,
      "loss": 1.1003,
      "step": 1508
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5529404878616333,
      "learning_rate": 0.00019015084553513533,
      "loss": 1.1231,
      "step": 1509
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.489065021276474,
      "learning_rate": 0.00019013783380602422,
      "loss": 0.9142,
      "step": 1510
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5688076615333557,
      "learning_rate": 0.0001901248139334702,
      "loss": 0.9035,
      "step": 1511
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5179052948951721,
      "learning_rate": 0.00019011178591864952,
      "loss": 0.978,
      "step": 1512
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.61092609167099,
      "learning_rate": 0.0001900987497627392,
      "loss": 1.1268,
      "step": 1513
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5672241449356079,
      "learning_rate": 0.00019008570546691695,
      "loss": 1.1325,
      "step": 1514
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5476197600364685,
      "learning_rate": 0.0001900726530323613,
      "loss": 0.9854,
      "step": 1515
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5623148679733276,
      "learning_rate": 0.00019005959246025142,
      "loss": 1.0171,
      "step": 1516
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5329583287239075,
      "learning_rate": 0.00019004652375176733,
      "loss": 1.0977,
      "step": 1517
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.582499086856842,
      "learning_rate": 0.0001900334469080896,
      "loss": 1.0895,
      "step": 1518
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5313141345977783,
      "learning_rate": 0.00019002036193039974,
      "loss": 1.0889,
      "step": 1519
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5586351752281189,
      "learning_rate": 0.00019000726881987986,
      "loss": 0.8848,
      "step": 1520
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6101945638656616,
      "learning_rate": 0.00018999416757771288,
      "loss": 1.1436,
      "step": 1521
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6232748627662659,
      "learning_rate": 0.00018998105820508236,
      "loss": 1.1319,
      "step": 1522
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5600299835205078,
      "learning_rate": 0.00018996794070317275,
      "loss": 1.1113,
      "step": 1523
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5319695472717285,
      "learning_rate": 0.00018995481507316905,
      "loss": 0.9104,
      "step": 1524
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5279593467712402,
      "learning_rate": 0.00018994168131625719,
      "loss": 0.9743,
      "step": 1525
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5348560214042664,
      "learning_rate": 0.00018992853943362363,
      "loss": 1.0488,
      "step": 1526
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6045165061950684,
      "learning_rate": 0.00018991538942645574,
      "loss": 1.1206,
      "step": 1527
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6501975059509277,
      "learning_rate": 0.00018990223129594148,
      "loss": 1.111,
      "step": 1528
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5018801689147949,
      "learning_rate": 0.00018988906504326965,
      "loss": 1.0066,
      "step": 1529
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5682173371315002,
      "learning_rate": 0.00018987589066962976,
      "loss": 1.1631,
      "step": 1530
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6507973670959473,
      "learning_rate": 0.00018986270817621201,
      "loss": 0.9727,
      "step": 1531
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5866438150405884,
      "learning_rate": 0.00018984951756420738,
      "loss": 1.0574,
      "step": 1532
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6363970637321472,
      "learning_rate": 0.00018983631883480757,
      "loss": 1.0447,
      "step": 1533
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5931180715560913,
      "learning_rate": 0.000189823111989205,
      "loss": 1.1534,
      "step": 1534
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.560235857963562,
      "learning_rate": 0.00018980989702859286,
      "loss": 0.9453,
      "step": 1535
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6191147565841675,
      "learning_rate": 0.00018979667395416501,
      "loss": 1.0562,
      "step": 1536
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5215852856636047,
      "learning_rate": 0.00018978344276711608,
      "loss": 1.2288,
      "step": 1537
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5872689485549927,
      "learning_rate": 0.00018977020346864145,
      "loss": 1.1983,
      "step": 1538
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5544451475143433,
      "learning_rate": 0.00018975695605993724,
      "loss": 1.1188,
      "step": 1539
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5902532935142517,
      "learning_rate": 0.00018974370054220026,
      "loss": 1.0748,
      "step": 1540
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6012886762619019,
      "learning_rate": 0.00018973043691662803,
      "loss": 1.0795,
      "step": 1541
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5270823836326599,
      "learning_rate": 0.00018971716518441888,
      "loss": 1.1419,
      "step": 1542
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6644396781921387,
      "learning_rate": 0.00018970388534677186,
      "loss": 1.0674,
      "step": 1543
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6086916923522949,
      "learning_rate": 0.0001896905974048867,
      "loss": 1.0743,
      "step": 1544
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5799842476844788,
      "learning_rate": 0.00018967730135996383,
      "loss": 1.0611,
      "step": 1545
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5766069293022156,
      "learning_rate": 0.0001896639972132046,
      "loss": 1.0784,
      "step": 1546
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6158689856529236,
      "learning_rate": 0.00018965068496581088,
      "loss": 1.2381,
      "step": 1547
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6311269402503967,
      "learning_rate": 0.00018963736461898538,
      "loss": 1.1589,
      "step": 1548
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5501083731651306,
      "learning_rate": 0.0001896240361739315,
      "loss": 1.0382,
      "step": 1549
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6331074237823486,
      "learning_rate": 0.0001896106996318534,
      "loss": 1.2187,
      "step": 1550
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.531475305557251,
      "learning_rate": 0.00018959735499395598,
      "loss": 0.9537,
      "step": 1551
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.69541996717453,
      "learning_rate": 0.00018958400226144482,
      "loss": 1.2587,
      "step": 1552
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5847852230072021,
      "learning_rate": 0.00018957064143552627,
      "loss": 1.1522,
      "step": 1553
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.549366295337677,
      "learning_rate": 0.00018955727251740743,
      "loss": 1.1589,
      "step": 1554
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6188735365867615,
      "learning_rate": 0.00018954389550829607,
      "loss": 1.0249,
      "step": 1555
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5889078974723816,
      "learning_rate": 0.00018953051040940076,
      "loss": 1.1075,
      "step": 1556
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5530256628990173,
      "learning_rate": 0.00018951711722193078,
      "loss": 1.0457,
      "step": 1557
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.586457371711731,
      "learning_rate": 0.0001895037159470961,
      "loss": 1.1271,
      "step": 1558
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5959973931312561,
      "learning_rate": 0.00018949030658610742,
      "loss": 1.0652,
      "step": 1559
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5624261498451233,
      "learning_rate": 0.00018947688914017623,
      "loss": 1.1421,
      "step": 1560
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5853245258331299,
      "learning_rate": 0.00018946346361051474,
      "loss": 1.1081,
      "step": 1561
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6063684225082397,
      "learning_rate": 0.00018945002999833584,
      "loss": 1.1567,
      "step": 1562
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5784252882003784,
      "learning_rate": 0.00018943658830485323,
      "loss": 1.2009,
      "step": 1563
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5920445322990417,
      "learning_rate": 0.00018942313853128123,
      "loss": 0.939,
      "step": 1564
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5626775622367859,
      "learning_rate": 0.00018940968067883497,
      "loss": 1.1501,
      "step": 1565
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.49538347125053406,
      "learning_rate": 0.00018939621474873031,
      "loss": 0.9118,
      "step": 1566
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6201627254486084,
      "learning_rate": 0.0001893827407421838,
      "loss": 1.2706,
      "step": 1567
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5373640656471252,
      "learning_rate": 0.00018936925866041277,
      "loss": 1.0574,
      "step": 1568
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5345717072486877,
      "learning_rate": 0.00018935576850463522,
      "loss": 1.1574,
      "step": 1569
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6629395484924316,
      "learning_rate": 0.00018934227027606987,
      "loss": 1.2365,
      "step": 1570
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6796295046806335,
      "learning_rate": 0.0001893287639759363,
      "loss": 1.1736,
      "step": 1571
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5906338691711426,
      "learning_rate": 0.00018931524960545467,
      "loss": 1.1604,
      "step": 1572
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.66068035364151,
      "learning_rate": 0.00018930172716584597,
      "loss": 1.0991,
      "step": 1573
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7229920029640198,
      "learning_rate": 0.00018928819665833178,
      "loss": 1.366,
      "step": 1574
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5846989750862122,
      "learning_rate": 0.00018927465808413465,
      "loss": 0.9767,
      "step": 1575
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5244808793067932,
      "learning_rate": 0.0001892611114444776,
      "loss": 1.1022,
      "step": 1576
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.642883837223053,
      "learning_rate": 0.0001892475567405845,
      "loss": 1.058,
      "step": 1577
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6407866477966309,
      "learning_rate": 0.00018923399397367997,
      "loss": 1.1835,
      "step": 1578
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5814583897590637,
      "learning_rate": 0.00018922042314498935,
      "loss": 1.1512,
      "step": 1579
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6063616871833801,
      "learning_rate": 0.00018920684425573865,
      "loss": 1.1334,
      "step": 1580
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6133993864059448,
      "learning_rate": 0.00018919325730715466,
      "loss": 0.945,
      "step": 1581
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5505120754241943,
      "learning_rate": 0.00018917966230046492,
      "loss": 1.1992,
      "step": 1582
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5015380382537842,
      "learning_rate": 0.00018916605923689756,
      "loss": 1.0274,
      "step": 1583
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4922814667224884,
      "learning_rate": 0.00018915244811768168,
      "loss": 0.9188,
      "step": 1584
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.562036395072937,
      "learning_rate": 0.00018913882894404683,
      "loss": 0.9782,
      "step": 1585
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.658740758895874,
      "learning_rate": 0.0001891252017172235,
      "loss": 0.9769,
      "step": 1586
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.675102949142456,
      "learning_rate": 0.00018911156643844283,
      "loss": 1.2205,
      "step": 1587
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6200870275497437,
      "learning_rate": 0.00018909792310893667,
      "loss": 1.1162,
      "step": 1588
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5535531044006348,
      "learning_rate": 0.00018908427172993766,
      "loss": 1.1918,
      "step": 1589
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6343311667442322,
      "learning_rate": 0.00018907061230267905,
      "loss": 1.0838,
      "step": 1590
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5962023735046387,
      "learning_rate": 0.00018905694482839497,
      "loss": 1.0302,
      "step": 1591
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6679428815841675,
      "learning_rate": 0.00018904326930832015,
      "loss": 1.1094,
      "step": 1592
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6458826065063477,
      "learning_rate": 0.0001890295857436901,
      "loss": 1.2363,
      "step": 1593
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5808925032615662,
      "learning_rate": 0.00018901589413574107,
      "loss": 1.1789,
      "step": 1594
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5363585352897644,
      "learning_rate": 0.00018900219448571003,
      "loss": 1.0798,
      "step": 1595
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5684288740158081,
      "learning_rate": 0.0001889884867948346,
      "loss": 1.047,
      "step": 1596
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5775579214096069,
      "learning_rate": 0.00018897477106435327,
      "loss": 1.1357,
      "step": 1597
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6103155016899109,
      "learning_rate": 0.00018896104729550515,
      "loss": 1.0776,
      "step": 1598
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6359939575195312,
      "learning_rate": 0.00018894731548953014,
      "loss": 1.1528,
      "step": 1599
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.553428590297699,
      "learning_rate": 0.00018893357564766874,
      "loss": 1.0318,
      "step": 1600
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6345550417900085,
      "learning_rate": 0.00018891982777116234,
      "loss": 1.2425,
      "step": 1601
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5610020756721497,
      "learning_rate": 0.00018890607186125292,
      "loss": 1.1793,
      "step": 1602
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6097350716590881,
      "learning_rate": 0.00018889230791918336,
      "loss": 0.9796,
      "step": 1603
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5626397132873535,
      "learning_rate": 0.00018887853594619705,
      "loss": 1.0767,
      "step": 1604
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5881405472755432,
      "learning_rate": 0.00018886475594353825,
      "loss": 1.2583,
      "step": 1605
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6646904349327087,
      "learning_rate": 0.0001888509679124519,
      "loss": 1.1146,
      "step": 1606
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5926646590232849,
      "learning_rate": 0.0001888371718541837,
      "loss": 1.2153,
      "step": 1607
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6441829204559326,
      "learning_rate": 0.00018882336776998,
      "loss": 1.1823,
      "step": 1608
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5266425013542175,
      "learning_rate": 0.0001888095556610879,
      "loss": 0.95,
      "step": 1609
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5980542302131653,
      "learning_rate": 0.00018879573552875534,
      "loss": 1.0822,
      "step": 1610
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5426793098449707,
      "learning_rate": 0.0001887819073742308,
      "loss": 1.0676,
      "step": 1611
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6400548815727234,
      "learning_rate": 0.00018876807119876362,
      "loss": 1.1233,
      "step": 1612
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5350365042686462,
      "learning_rate": 0.00018875422700360384,
      "loss": 0.9547,
      "step": 1613
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5893256068229675,
      "learning_rate": 0.00018874037479000216,
      "loss": 1.266,
      "step": 1614
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.564774751663208,
      "learning_rate": 0.00018872651455921006,
      "loss": 1.0358,
      "step": 1615
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.588233232498169,
      "learning_rate": 0.00018871264631247972,
      "loss": 1.1458,
      "step": 1616
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.567189633846283,
      "learning_rate": 0.00018869877005106407,
      "loss": 0.9776,
      "step": 1617
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5712881684303284,
      "learning_rate": 0.00018868488577621678,
      "loss": 1.1353,
      "step": 1618
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5613961219787598,
      "learning_rate": 0.00018867099348919217,
      "loss": 1.1427,
      "step": 1619
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6011113524436951,
      "learning_rate": 0.0001886570931912454,
      "loss": 1.1082,
      "step": 1620
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6284022927284241,
      "learning_rate": 0.00018864318488363217,
      "loss": 1.148,
      "step": 1621
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5748017430305481,
      "learning_rate": 0.00018862926856760912,
      "loss": 1.116,
      "step": 1622
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5744466781616211,
      "learning_rate": 0.0001886153442444335,
      "loss": 1.0798,
      "step": 1623
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5814639329910278,
      "learning_rate": 0.00018860141191536322,
      "loss": 1.1328,
      "step": 1624
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5316477417945862,
      "learning_rate": 0.00018858747158165704,
      "loss": 0.981,
      "step": 1625
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6174801588058472,
      "learning_rate": 0.0001885735232445744,
      "loss": 1.1805,
      "step": 1626
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5454065799713135,
      "learning_rate": 0.0001885595669053754,
      "loss": 1.1229,
      "step": 1627
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5038221478462219,
      "learning_rate": 0.000188545602565321,
      "loss": 1.0006,
      "step": 1628
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6330316066741943,
      "learning_rate": 0.00018853163022567273,
      "loss": 1.0873,
      "step": 1629
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5468506813049316,
      "learning_rate": 0.00018851764988769295,
      "loss": 1.034,
      "step": 1630
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5783451795578003,
      "learning_rate": 0.00018850366155264466,
      "loss": 1.0622,
      "step": 1631
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5333491563796997,
      "learning_rate": 0.00018848966522179168,
      "loss": 0.9803,
      "step": 1632
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5402332544326782,
      "learning_rate": 0.00018847566089639847,
      "loss": 1.1328,
      "step": 1633
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5634700059890747,
      "learning_rate": 0.00018846164857773024,
      "loss": 0.9754,
      "step": 1634
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5241762399673462,
      "learning_rate": 0.00018844762826705292,
      "loss": 1.1019,
      "step": 1635
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5816453099250793,
      "learning_rate": 0.0001884335999656332,
      "loss": 1.0907,
      "step": 1636
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5889614224433899,
      "learning_rate": 0.0001884195636747384,
      "loss": 1.1334,
      "step": 1637
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5321390628814697,
      "learning_rate": 0.00018840551939563666,
      "loss": 0.9717,
      "step": 1638
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5475577712059021,
      "learning_rate": 0.00018839146712959682,
      "loss": 1.0735,
      "step": 1639
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5626057386398315,
      "learning_rate": 0.00018837740687788839,
      "loss": 1.0199,
      "step": 1640
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5641455054283142,
      "learning_rate": 0.00018836333864178162,
      "loss": 1.1475,
      "step": 1641
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5993323922157288,
      "learning_rate": 0.00018834926242254754,
      "loss": 1.0956,
      "step": 1642
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5953279733657837,
      "learning_rate": 0.0001883351782214578,
      "loss": 1.021,
      "step": 1643
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5046542286872864,
      "learning_rate": 0.0001883210860397849,
      "loss": 0.8364,
      "step": 1644
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6529065370559692,
      "learning_rate": 0.00018830698587880194,
      "loss": 1.0135,
      "step": 1645
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6167691349983215,
      "learning_rate": 0.00018829287773978277,
      "loss": 1.2643,
      "step": 1646
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5951334238052368,
      "learning_rate": 0.00018827876162400201,
      "loss": 1.1444,
      "step": 1647
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.519690752029419,
      "learning_rate": 0.000188264637532735,
      "loss": 1.085,
      "step": 1648
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5396617650985718,
      "learning_rate": 0.00018825050546725773,
      "loss": 1.1176,
      "step": 1649
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5493987202644348,
      "learning_rate": 0.00018823636542884695,
      "loss": 1.0602,
      "step": 1650
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5464897155761719,
      "learning_rate": 0.00018822221741878018,
      "loss": 0.8669,
      "step": 1651
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5813970565795898,
      "learning_rate": 0.00018820806143833555,
      "loss": 1.0391,
      "step": 1652
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5767297148704529,
      "learning_rate": 0.00018819389748879202,
      "loss": 1.0945,
      "step": 1653
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6370941996574402,
      "learning_rate": 0.00018817972557142918,
      "loss": 1.1547,
      "step": 1654
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6555666923522949,
      "learning_rate": 0.0001881655456875274,
      "loss": 1.1207,
      "step": 1655
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5538688898086548,
      "learning_rate": 0.0001881513578383678,
      "loss": 0.9714,
      "step": 1656
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5678582191467285,
      "learning_rate": 0.0001881371620252321,
      "loss": 1.2347,
      "step": 1657
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5648280382156372,
      "learning_rate": 0.00018812295824940285,
      "loss": 1.0975,
      "step": 1658
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.569893479347229,
      "learning_rate": 0.0001881087465121633,
      "loss": 1.0578,
      "step": 1659
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5792575478553772,
      "learning_rate": 0.00018809452681479733,
      "loss": 1.0296,
      "step": 1660
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6080823540687561,
      "learning_rate": 0.00018808029915858967,
      "loss": 1.2352,
      "step": 1661
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6537753343582153,
      "learning_rate": 0.0001880660635448257,
      "loss": 1.1956,
      "step": 1662
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6456838250160217,
      "learning_rate": 0.00018805181997479152,
      "loss": 1.1356,
      "step": 1663
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5179287791252136,
      "learning_rate": 0.00018803756844977394,
      "loss": 1.0559,
      "step": 1664
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5787820816040039,
      "learning_rate": 0.00018802330897106051,
      "loss": 1.1527,
      "step": 1665
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6485100388526917,
      "learning_rate": 0.0001880090415399395,
      "loss": 1.1667,
      "step": 1666
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.605714738368988,
      "learning_rate": 0.00018799476615769992,
      "loss": 1.2807,
      "step": 1667
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5844491124153137,
      "learning_rate": 0.00018798048282563144,
      "loss": 1.1697,
      "step": 1668
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5353244543075562,
      "learning_rate": 0.00018796619154502447,
      "loss": 0.9961,
      "step": 1669
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5764732360839844,
      "learning_rate": 0.00018795189231717014,
      "loss": 1.0053,
      "step": 1670
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5386641025543213,
      "learning_rate": 0.00018793758514336033,
      "loss": 1.1774,
      "step": 1671
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5288216471672058,
      "learning_rate": 0.00018792327002488762,
      "loss": 1.0226,
      "step": 1672
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5615891218185425,
      "learning_rate": 0.00018790894696304525,
      "loss": 1.0576,
      "step": 1673
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5695824027061462,
      "learning_rate": 0.0001878946159591273,
      "loss": 1.1809,
      "step": 1674
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.598285436630249,
      "learning_rate": 0.00018788027701442843,
      "loss": 1.0365,
      "step": 1675
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6066529154777527,
      "learning_rate": 0.00018786593013024412,
      "loss": 1.0722,
      "step": 1676
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5309323668479919,
      "learning_rate": 0.00018785157530787052,
      "loss": 1.0409,
      "step": 1677
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5708740949630737,
      "learning_rate": 0.0001878372125486045,
      "loss": 1.1045,
      "step": 1678
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5249610543251038,
      "learning_rate": 0.00018782284185374368,
      "loss": 1.0355,
      "step": 1679
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.553139865398407,
      "learning_rate": 0.00018780846322458632,
      "loss": 0.9598,
      "step": 1680
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5853700041770935,
      "learning_rate": 0.0001877940766624315,
      "loss": 1.1801,
      "step": 1681
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.694056510925293,
      "learning_rate": 0.00018777968216857893,
      "loss": 1.1596,
      "step": 1682
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.606585681438446,
      "learning_rate": 0.00018776527974432908,
      "loss": 1.2664,
      "step": 1683
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5530321598052979,
      "learning_rate": 0.00018775086939098312,
      "loss": 1.1819,
      "step": 1684
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6291928291320801,
      "learning_rate": 0.00018773645110984298,
      "loss": 1.3132,
      "step": 1685
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5767596364021301,
      "learning_rate": 0.00018772202490221124,
      "loss": 1.0349,
      "step": 1686
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6143909096717834,
      "learning_rate": 0.0001877075907693912,
      "loss": 1.2825,
      "step": 1687
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5503374338150024,
      "learning_rate": 0.000187693148712687,
      "loss": 1.1777,
      "step": 1688
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5860001444816589,
      "learning_rate": 0.0001876786987334033,
      "loss": 0.9836,
      "step": 1689
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.645011305809021,
      "learning_rate": 0.00018766424083284557,
      "loss": 1.1498,
      "step": 1690
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5632151365280151,
      "learning_rate": 0.00018764977501232007,
      "loss": 1.0685,
      "step": 1691
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5840081572532654,
      "learning_rate": 0.00018763530127313365,
      "loss": 1.0215,
      "step": 1692
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5783519148826599,
      "learning_rate": 0.00018762081961659396,
      "loss": 1.0851,
      "step": 1693
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.554547131061554,
      "learning_rate": 0.0001876063300440093,
      "loss": 1.1261,
      "step": 1694
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5496121644973755,
      "learning_rate": 0.0001875918325566888,
      "loss": 0.8681,
      "step": 1695
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5871601104736328,
      "learning_rate": 0.00018757732715594214,
      "loss": 0.9968,
      "step": 1696
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5784563422203064,
      "learning_rate": 0.00018756281384307982,
      "loss": 1.1381,
      "step": 1697
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5940728187561035,
      "learning_rate": 0.00018754829261941308,
      "loss": 1.0509,
      "step": 1698
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5413823127746582,
      "learning_rate": 0.00018753376348625377,
      "loss": 1.091,
      "step": 1699
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5205764770507812,
      "learning_rate": 0.00018751922644491456,
      "loss": 1.077,
      "step": 1700
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5888503789901733,
      "learning_rate": 0.0001875046814967088,
      "loss": 1.3145,
      "step": 1701
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6186810731887817,
      "learning_rate": 0.00018749012864295047,
      "loss": 1.103,
      "step": 1702
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6271312832832336,
      "learning_rate": 0.00018747556788495438,
      "loss": 1.0219,
      "step": 1703
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.543907642364502,
      "learning_rate": 0.00018746099922403606,
      "loss": 0.9024,
      "step": 1704
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5604395866394043,
      "learning_rate": 0.00018744642266151163,
      "loss": 1.0485,
      "step": 1705
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5746266841888428,
      "learning_rate": 0.00018743183819869806,
      "loss": 1.0975,
      "step": 1706
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5429287552833557,
      "learning_rate": 0.00018741724583691294,
      "loss": 1.0327,
      "step": 1707
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5611682534217834,
      "learning_rate": 0.00018740264557747457,
      "loss": 0.9671,
      "step": 1708
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5795273184776306,
      "learning_rate": 0.00018738803742170213,
      "loss": 1.0083,
      "step": 1709
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.602764368057251,
      "learning_rate": 0.00018737342137091522,
      "loss": 1.2469,
      "step": 1710
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5357661843299866,
      "learning_rate": 0.00018735879742643442,
      "loss": 1.1328,
      "step": 1711
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5171231031417847,
      "learning_rate": 0.0001873441655895809,
      "loss": 1.1278,
      "step": 1712
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6350235342979431,
      "learning_rate": 0.00018732952586167655,
      "loss": 1.0691,
      "step": 1713
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.579694926738739,
      "learning_rate": 0.00018731487824404403,
      "loss": 1.1816,
      "step": 1714
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6309595108032227,
      "learning_rate": 0.0001873002227380066,
      "loss": 1.0332,
      "step": 1715
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5699979662895203,
      "learning_rate": 0.00018728555934488837,
      "loss": 1.0802,
      "step": 1716
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6824408769607544,
      "learning_rate": 0.00018727088806601405,
      "loss": 1.0947,
      "step": 1717
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5273520350456238,
      "learning_rate": 0.0001872562089027091,
      "loss": 1.0015,
      "step": 1718
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5987673997879028,
      "learning_rate": 0.00018724152185629976,
      "loss": 1.0188,
      "step": 1719
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6542358994483948,
      "learning_rate": 0.00018722682692811282,
      "loss": 1.2388,
      "step": 1720
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5636007785797119,
      "learning_rate": 0.000187212124119476,
      "loss": 1.1047,
      "step": 1721
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5961930751800537,
      "learning_rate": 0.00018719741343171755,
      "loss": 1.2205,
      "step": 1722
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5861148834228516,
      "learning_rate": 0.0001871826948661665,
      "loss": 1.1424,
      "step": 1723
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6240936517715454,
      "learning_rate": 0.00018716796842415258,
      "loss": 1.2925,
      "step": 1724
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5183912515640259,
      "learning_rate": 0.00018715323410700626,
      "loss": 0.922,
      "step": 1725
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.611001193523407,
      "learning_rate": 0.00018713849191605872,
      "loss": 1.2328,
      "step": 1726
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6203360557556152,
      "learning_rate": 0.0001871237418526418,
      "loss": 1.2026,
      "step": 1727
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6380226016044617,
      "learning_rate": 0.00018710898391808808,
      "loss": 1.193,
      "step": 1728
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.711965024471283,
      "learning_rate": 0.00018709421811373091,
      "loss": 1.1432,
      "step": 1729
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5226439833641052,
      "learning_rate": 0.0001870794444409042,
      "loss": 1.0078,
      "step": 1730
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.49361833930015564,
      "learning_rate": 0.0001870646629009428,
      "loss": 1.0736,
      "step": 1731
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5229170322418213,
      "learning_rate": 0.000187049873495182,
      "loss": 1.1184,
      "step": 1732
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.518227219581604,
      "learning_rate": 0.00018703507622495807,
      "loss": 1.1062,
      "step": 1733
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5495775938034058,
      "learning_rate": 0.00018702027109160775,
      "loss": 0.9128,
      "step": 1734
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6473329067230225,
      "learning_rate": 0.00018700545809646863,
      "loss": 0.9592,
      "step": 1735
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5847293138504028,
      "learning_rate": 0.00018699063724087904,
      "loss": 0.9249,
      "step": 1736
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6614704728126526,
      "learning_rate": 0.0001869758085261779,
      "loss": 1.0977,
      "step": 1737
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6490361094474792,
      "learning_rate": 0.0001869609719537049,
      "loss": 1.1991,
      "step": 1738
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6661361455917358,
      "learning_rate": 0.00018694612752480046,
      "loss": 1.0612,
      "step": 1739
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5234553813934326,
      "learning_rate": 0.00018693127524080572,
      "loss": 0.9233,
      "step": 1740
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.49033844470977783,
      "learning_rate": 0.0001869164151030624,
      "loss": 0.8696,
      "step": 1741
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5724113583564758,
      "learning_rate": 0.00018690154711291314,
      "loss": 1.0518,
      "step": 1742
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5647688508033752,
      "learning_rate": 0.00018688667127170113,
      "loss": 1.1055,
      "step": 1743
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5435717701911926,
      "learning_rate": 0.00018687178758077032,
      "loss": 0.9948,
      "step": 1744
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6030397415161133,
      "learning_rate": 0.00018685689604146537,
      "loss": 0.9858,
      "step": 1745
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.550806999206543,
      "learning_rate": 0.00018684199665513164,
      "loss": 1.0742,
      "step": 1746
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6276724338531494,
      "learning_rate": 0.0001868270894231152,
      "loss": 1.2271,
      "step": 1747
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7018080949783325,
      "learning_rate": 0.00018681217434676287,
      "loss": 1.1183,
      "step": 1748
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5612937211990356,
      "learning_rate": 0.00018679725142742212,
      "loss": 1.1673,
      "step": 1749
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6091915965080261,
      "learning_rate": 0.0001867823206664411,
      "loss": 1.1528,
      "step": 1750
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5641574263572693,
      "learning_rate": 0.0001867673820651688,
      "loss": 1.1589,
      "step": 1751
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.59687340259552,
      "learning_rate": 0.0001867524356249548,
      "loss": 1.1207,
      "step": 1752
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5552710294723511,
      "learning_rate": 0.00018673748134714945,
      "loss": 1.1232,
      "step": 1753
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.657662570476532,
      "learning_rate": 0.00018672251923310376,
      "loss": 1.3406,
      "step": 1754
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6064003705978394,
      "learning_rate": 0.00018670754928416946,
      "loss": 1.1582,
      "step": 1755
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5749542117118835,
      "learning_rate": 0.00018669257150169904,
      "loss": 0.9952,
      "step": 1756
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5489286780357361,
      "learning_rate": 0.0001866775858870456,
      "loss": 1.2523,
      "step": 1757
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5087161660194397,
      "learning_rate": 0.0001866625924415631,
      "loss": 0.91,
      "step": 1758
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5486624240875244,
      "learning_rate": 0.00018664759116660603,
      "loss": 0.8993,
      "step": 1759
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5084232687950134,
      "learning_rate": 0.0001866325820635297,
      "loss": 1.0808,
      "step": 1760
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6880375742912292,
      "learning_rate": 0.00018661756513369007,
      "loss": 1.1413,
      "step": 1761
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6002033948898315,
      "learning_rate": 0.00018660254037844388,
      "loss": 1.0729,
      "step": 1762
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5921014547348022,
      "learning_rate": 0.00018658750779914848,
      "loss": 1.1544,
      "step": 1763
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6466764807701111,
      "learning_rate": 0.00018657246739716205,
      "loss": 1.1871,
      "step": 1764
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6112008690834045,
      "learning_rate": 0.00018655741917384332,
      "loss": 0.9757,
      "step": 1765
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5578400492668152,
      "learning_rate": 0.00018654236313055188,
      "loss": 0.934,
      "step": 1766
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5956066250801086,
      "learning_rate": 0.00018652729926864795,
      "loss": 1.1892,
      "step": 1767
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.598323404788971,
      "learning_rate": 0.00018651222758949242,
      "loss": 0.9415,
      "step": 1768
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6082397103309631,
      "learning_rate": 0.00018649714809444696,
      "loss": 1.148,
      "step": 1769
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.566173791885376,
      "learning_rate": 0.0001864820607848739,
      "loss": 1.1214,
      "step": 1770
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5706698894500732,
      "learning_rate": 0.00018646696566213634,
      "loss": 1.2498,
      "step": 1771
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5568757653236389,
      "learning_rate": 0.00018645186272759801,
      "loss": 0.9619,
      "step": 1772
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5579240322113037,
      "learning_rate": 0.00018643675198262333,
      "loss": 0.9143,
      "step": 1773
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5389413833618164,
      "learning_rate": 0.00018642163342857756,
      "loss": 1.0979,
      "step": 1774
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6314026713371277,
      "learning_rate": 0.0001864065070668265,
      "loss": 1.2771,
      "step": 1775
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6543720960617065,
      "learning_rate": 0.00018639137289873676,
      "loss": 1.2627,
      "step": 1776
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.49429407715797424,
      "learning_rate": 0.00018637623092567564,
      "loss": 1.0445,
      "step": 1777
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5479093194007874,
      "learning_rate": 0.0001863610811490111,
      "loss": 1.0667,
      "step": 1778
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5306311845779419,
      "learning_rate": 0.00018634592357011186,
      "loss": 0.949,
      "step": 1779
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5764743089675903,
      "learning_rate": 0.00018633075819034732,
      "loss": 1.1231,
      "step": 1780
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5098384618759155,
      "learning_rate": 0.00018631558501108758,
      "loss": 1.0341,
      "step": 1781
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6208810806274414,
      "learning_rate": 0.00018630040403370344,
      "loss": 1.1288,
      "step": 1782
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6083211898803711,
      "learning_rate": 0.00018628521525956645,
      "loss": 1.1865,
      "step": 1783
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5995293259620667,
      "learning_rate": 0.0001862700186900488,
      "loss": 1.1144,
      "step": 1784
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5885659456253052,
      "learning_rate": 0.00018625481432652338,
      "loss": 1.0731,
      "step": 1785
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6302295327186584,
      "learning_rate": 0.00018623960217036388,
      "loss": 1.1223,
      "step": 1786
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5090101957321167,
      "learning_rate": 0.00018622438222294464,
      "loss": 0.9719,
      "step": 1787
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.592470645904541,
      "learning_rate": 0.00018620915448564068,
      "loss": 1.1259,
      "step": 1788
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.602279782295227,
      "learning_rate": 0.00018619391895982766,
      "loss": 1.0776,
      "step": 1789
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6333302855491638,
      "learning_rate": 0.00018617867564688214,
      "loss": 1.1915,
      "step": 1790
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6094706058502197,
      "learning_rate": 0.0001861634245481812,
      "loss": 1.101,
      "step": 1791
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5379626154899597,
      "learning_rate": 0.00018614816566510272,
      "loss": 1.0421,
      "step": 1792
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5724616050720215,
      "learning_rate": 0.00018613289899902525,
      "loss": 1.1036,
      "step": 1793
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.524731457233429,
      "learning_rate": 0.00018611762455132802,
      "loss": 0.9453,
      "step": 1794
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.631876528263092,
      "learning_rate": 0.00018610234232339102,
      "loss": 1.2017,
      "step": 1795
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5800185799598694,
      "learning_rate": 0.00018608705231659486,
      "loss": 1.0768,
      "step": 1796
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7038686275482178,
      "learning_rate": 0.00018607175453232098,
      "loss": 1.2161,
      "step": 1797
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.605658233165741,
      "learning_rate": 0.00018605644897195145,
      "loss": 1.1248,
      "step": 1798
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5098914504051208,
      "learning_rate": 0.00018604113563686897,
      "loss": 0.9575,
      "step": 1799
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.554283082485199,
      "learning_rate": 0.00018602581452845703,
      "loss": 1.0522,
      "step": 1800
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5628049373626709,
      "learning_rate": 0.00018601048564809985,
      "loss": 1.1679,
      "step": 1801
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5380632281303406,
      "learning_rate": 0.0001859951489971823,
      "loss": 1.1503,
      "step": 1802
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.638298749923706,
      "learning_rate": 0.0001859798045770899,
      "loss": 1.2309,
      "step": 1803
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5701687932014465,
      "learning_rate": 0.00018596445238920902,
      "loss": 1.2437,
      "step": 1804
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5610591173171997,
      "learning_rate": 0.00018594909243492653,
      "loss": 1.1682,
      "step": 1805
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5382736921310425,
      "learning_rate": 0.00018593372471563022,
      "loss": 0.9795,
      "step": 1806
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5971048474311829,
      "learning_rate": 0.00018591834923270843,
      "loss": 0.9871,
      "step": 1807
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5558769106864929,
      "learning_rate": 0.00018590296598755028,
      "loss": 0.8444,
      "step": 1808
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5690754055976868,
      "learning_rate": 0.0001858875749815455,
      "loss": 1.0904,
      "step": 1809
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6201112270355225,
      "learning_rate": 0.00018587217621608462,
      "loss": 1.1436,
      "step": 1810
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6081562638282776,
      "learning_rate": 0.00018585676969255885,
      "loss": 1.2893,
      "step": 1811
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6248214244842529,
      "learning_rate": 0.00018584135541236006,
      "loss": 1.1327,
      "step": 1812
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5153765082359314,
      "learning_rate": 0.0001858259333768808,
      "loss": 1.0487,
      "step": 1813
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5564947724342346,
      "learning_rate": 0.00018581050358751445,
      "loss": 1.0409,
      "step": 1814
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5596327781677246,
      "learning_rate": 0.00018579506604565494,
      "loss": 1.118,
      "step": 1815
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6628304719924927,
      "learning_rate": 0.00018577962075269698,
      "loss": 1.182,
      "step": 1816
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5848866701126099,
      "learning_rate": 0.00018576416771003595,
      "loss": 1.2644,
      "step": 1817
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.600448489189148,
      "learning_rate": 0.00018574870691906797,
      "loss": 1.1461,
      "step": 1818
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5690786242485046,
      "learning_rate": 0.00018573323838118986,
      "loss": 1.1681,
      "step": 1819
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5750014781951904,
      "learning_rate": 0.00018571776209779905,
      "loss": 0.9931,
      "step": 1820
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5702171921730042,
      "learning_rate": 0.00018570227807029378,
      "loss": 1.0003,
      "step": 1821
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5780484676361084,
      "learning_rate": 0.00018568678630007294,
      "loss": 1.1079,
      "step": 1822
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5208010077476501,
      "learning_rate": 0.00018567128678853608,
      "loss": 1.1168,
      "step": 1823
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5132521986961365,
      "learning_rate": 0.00018565577953708358,
      "loss": 0.9701,
      "step": 1824
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5545271635055542,
      "learning_rate": 0.00018564026454711633,
      "loss": 0.9203,
      "step": 1825
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5755243897438049,
      "learning_rate": 0.0001856247418200361,
      "loss": 1.049,
      "step": 1826
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.601019024848938,
      "learning_rate": 0.00018560921135724528,
      "loss": 1.1816,
      "step": 1827
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6014050841331482,
      "learning_rate": 0.00018559367316014688,
      "loss": 1.1182,
      "step": 1828
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5794886946678162,
      "learning_rate": 0.00018557812723014476,
      "loss": 0.9821,
      "step": 1829
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6025217175483704,
      "learning_rate": 0.00018556257356864338,
      "loss": 1.0827,
      "step": 1830
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.581674337387085,
      "learning_rate": 0.00018554701217704796,
      "loss": 0.8873,
      "step": 1831
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.669243574142456,
      "learning_rate": 0.00018553144305676433,
      "loss": 0.9918,
      "step": 1832
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6118124723434448,
      "learning_rate": 0.0001855158662091991,
      "loss": 1.0832,
      "step": 1833
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5944288969039917,
      "learning_rate": 0.00018550028163575954,
      "loss": 1.2669,
      "step": 1834
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6186957359313965,
      "learning_rate": 0.00018548468933785365,
      "loss": 0.9488,
      "step": 1835
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5669781565666199,
      "learning_rate": 0.00018546908931689007,
      "loss": 1.1178,
      "step": 1836
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6318771839141846,
      "learning_rate": 0.00018545348157427823,
      "loss": 1.061,
      "step": 1837
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6139348745346069,
      "learning_rate": 0.00018543786611142815,
      "loss": 1.0234,
      "step": 1838
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4509977102279663,
      "learning_rate": 0.00018542224292975062,
      "loss": 0.7523,
      "step": 1839
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5004023909568787,
      "learning_rate": 0.0001854066120306571,
      "loss": 1.0223,
      "step": 1840
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6322842836380005,
      "learning_rate": 0.00018539097341555973,
      "loss": 1.1863,
      "step": 1841
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5344803333282471,
      "learning_rate": 0.00018537532708587143,
      "loss": 1.0257,
      "step": 1842
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5191022753715515,
      "learning_rate": 0.0001853596730430057,
      "loss": 0.9801,
      "step": 1843
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5770233869552612,
      "learning_rate": 0.00018534401128837683,
      "loss": 1.0975,
      "step": 1844
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6385951638221741,
      "learning_rate": 0.0001853283418233997,
      "loss": 0.9929,
      "step": 1845
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5498313903808594,
      "learning_rate": 0.0001853126646494901,
      "loss": 0.9341,
      "step": 1846
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6085447669029236,
      "learning_rate": 0.00018529697976806423,
      "loss": 1.057,
      "step": 1847
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5516953468322754,
      "learning_rate": 0.0001852812871805392,
      "loss": 1.0169,
      "step": 1848
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5250498652458191,
      "learning_rate": 0.00018526558688833274,
      "loss": 0.9131,
      "step": 1849
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7380767464637756,
      "learning_rate": 0.0001852498788928633,
      "loss": 1.2816,
      "step": 1850
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5584884285926819,
      "learning_rate": 0.00018523416319554995,
      "loss": 0.9567,
      "step": 1851
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5699655413627625,
      "learning_rate": 0.00018521843979781258,
      "loss": 1.1744,
      "step": 1852
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5850598216056824,
      "learning_rate": 0.00018520270870107166,
      "loss": 1.0347,
      "step": 1853
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5637402534484863,
      "learning_rate": 0.00018518696990674842,
      "loss": 1.2212,
      "step": 1854
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6142070889472961,
      "learning_rate": 0.00018517122341626477,
      "loss": 1.07,
      "step": 1855
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5141515135765076,
      "learning_rate": 0.00018515546923104337,
      "loss": 0.9331,
      "step": 1856
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.473111093044281,
      "learning_rate": 0.00018513970735250742,
      "loss": 0.7911,
      "step": 1857
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6113271713256836,
      "learning_rate": 0.000185123937782081,
      "loss": 1.2151,
      "step": 1858
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6086576581001282,
      "learning_rate": 0.00018510816052118875,
      "loss": 1.1222,
      "step": 1859
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5792139768600464,
      "learning_rate": 0.0001850923755712561,
      "loss": 0.9931,
      "step": 1860
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5663713216781616,
      "learning_rate": 0.00018507658293370912,
      "loss": 1.1144,
      "step": 1861
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6032007336616516,
      "learning_rate": 0.0001850607826099745,
      "loss": 1.1773,
      "step": 1862
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5945179462432861,
      "learning_rate": 0.00018504497460147988,
      "loss": 0.9072,
      "step": 1863
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6139575839042664,
      "learning_rate": 0.00018502915890965328,
      "loss": 1.1492,
      "step": 1864
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5732980370521545,
      "learning_rate": 0.0001850133355359236,
      "loss": 0.9399,
      "step": 1865
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6116182804107666,
      "learning_rate": 0.00018499750448172045,
      "loss": 1.1455,
      "step": 1866
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5685417652130127,
      "learning_rate": 0.00018498166574847398,
      "loss": 1.1286,
      "step": 1867
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6494985222816467,
      "learning_rate": 0.00018496581933761516,
      "loss": 1.1148,
      "step": 1868
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5793482065200806,
      "learning_rate": 0.00018494996525057566,
      "loss": 1.1552,
      "step": 1869
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5607535243034363,
      "learning_rate": 0.00018493410348878779,
      "loss": 1.0392,
      "step": 1870
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5485084652900696,
      "learning_rate": 0.00018491823405368456,
      "loss": 0.8355,
      "step": 1871
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5895860195159912,
      "learning_rate": 0.0001849023569466997,
      "loss": 1.3229,
      "step": 1872
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6231148838996887,
      "learning_rate": 0.00018488647216926757,
      "loss": 1.0216,
      "step": 1873
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5446336269378662,
      "learning_rate": 0.00018487057972282336,
      "loss": 1.0869,
      "step": 1874
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6365904211997986,
      "learning_rate": 0.00018485467960880275,
      "loss": 1.0267,
      "step": 1875
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5633032917976379,
      "learning_rate": 0.0001848387718286423,
      "loss": 1.0315,
      "step": 1876
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5365302562713623,
      "learning_rate": 0.0001848228563837792,
      "loss": 1.2495,
      "step": 1877
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6207290291786194,
      "learning_rate": 0.0001848069332756513,
      "loss": 1.1013,
      "step": 1878
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6174927949905396,
      "learning_rate": 0.0001847910025056971,
      "loss": 1.2725,
      "step": 1879
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.565679132938385,
      "learning_rate": 0.00018477506407535595,
      "loss": 0.9717,
      "step": 1880
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6149706244468689,
      "learning_rate": 0.00018475911798606772,
      "loss": 0.9632,
      "step": 1881
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5572057962417603,
      "learning_rate": 0.00018474316423927313,
      "loss": 0.9965,
      "step": 1882
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5754985213279724,
      "learning_rate": 0.00018472720283641345,
      "loss": 0.8882,
      "step": 1883
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6410358548164368,
      "learning_rate": 0.0001847112337789307,
      "loss": 0.9525,
      "step": 1884
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6138419508934021,
      "learning_rate": 0.00018469525706826763,
      "loss": 1.1237,
      "step": 1885
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5606120824813843,
      "learning_rate": 0.0001846792727058676,
      "loss": 0.9746,
      "step": 1886
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5778054594993591,
      "learning_rate": 0.0001846632806931748,
      "loss": 1.0737,
      "step": 1887
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5834203958511353,
      "learning_rate": 0.00018464728103163386,
      "loss": 1.2133,
      "step": 1888
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6452996134757996,
      "learning_rate": 0.0001846312737226904,
      "loss": 1.1829,
      "step": 1889
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5307098627090454,
      "learning_rate": 0.00018461525876779057,
      "loss": 0.9321,
      "step": 1890
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5604325532913208,
      "learning_rate": 0.00018459923616838113,
      "loss": 1.0976,
      "step": 1891
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5541103482246399,
      "learning_rate": 0.00018458320592590975,
      "loss": 1.0378,
      "step": 1892
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5926581621170044,
      "learning_rate": 0.0001845671680418246,
      "loss": 0.9836,
      "step": 1893
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6027371883392334,
      "learning_rate": 0.00018455112251757465,
      "loss": 1.2251,
      "step": 1894
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6586189270019531,
      "learning_rate": 0.0001845350693546095,
      "loss": 1.2198,
      "step": 1895
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5091075897216797,
      "learning_rate": 0.0001845190085543795,
      "loss": 1.041,
      "step": 1896
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5649153590202332,
      "learning_rate": 0.00018450294011833557,
      "loss": 1.1006,
      "step": 1897
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6161060929298401,
      "learning_rate": 0.0001844868640479295,
      "loss": 1.0289,
      "step": 1898
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5821948647499084,
      "learning_rate": 0.0001844707803446136,
      "loss": 0.9917,
      "step": 1899
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6699268221855164,
      "learning_rate": 0.00018445468900984098,
      "loss": 1.2236,
      "step": 1900
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5464971661567688,
      "learning_rate": 0.0001844385900450654,
      "loss": 1.0474,
      "step": 1901
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6147408485412598,
      "learning_rate": 0.00018442248345174126,
      "loss": 1.1151,
      "step": 1902
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6387994289398193,
      "learning_rate": 0.00018440636923132376,
      "loss": 1.2701,
      "step": 1903
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6076620221138,
      "learning_rate": 0.00018439024738526873,
      "loss": 0.957,
      "step": 1904
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6402626037597656,
      "learning_rate": 0.00018437411791503262,
      "loss": 1.032,
      "step": 1905
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6451716423034668,
      "learning_rate": 0.00018435798082207274,
      "loss": 1.0105,
      "step": 1906
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7566270232200623,
      "learning_rate": 0.0001843418361078469,
      "loss": 0.8896,
      "step": 1907
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.589716911315918,
      "learning_rate": 0.0001843256837738137,
      "loss": 1.0328,
      "step": 1908
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7324854731559753,
      "learning_rate": 0.00018430952382143242,
      "loss": 1.0414,
      "step": 1909
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6748380064964294,
      "learning_rate": 0.00018429335625216303,
      "loss": 1.1903,
      "step": 1910
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5136976838111877,
      "learning_rate": 0.00018427718106746617,
      "loss": 0.9545,
      "step": 1911
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6081967949867249,
      "learning_rate": 0.00018426099826880318,
      "loss": 1.0921,
      "step": 1912
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5389492511749268,
      "learning_rate": 0.00018424480785763606,
      "loss": 1.0173,
      "step": 1913
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6011937260627747,
      "learning_rate": 0.00018422860983542758,
      "loss": 0.9444,
      "step": 1914
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6137906908988953,
      "learning_rate": 0.00018421240420364108,
      "loss": 1.0585,
      "step": 1915
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6015632748603821,
      "learning_rate": 0.00018419619096374067,
      "loss": 1.1256,
      "step": 1916
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6502439975738525,
      "learning_rate": 0.00018417997011719111,
      "loss": 1.158,
      "step": 1917
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5486273765563965,
      "learning_rate": 0.00018416374166545788,
      "loss": 0.9079,
      "step": 1918
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5800416469573975,
      "learning_rate": 0.0001841475056100072,
      "loss": 1.1742,
      "step": 1919
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5277450084686279,
      "learning_rate": 0.00018413126195230572,
      "loss": 1.0423,
      "step": 1920
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5868902802467346,
      "learning_rate": 0.00018411501069382112,
      "loss": 1.1206,
      "step": 1921
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5652511119842529,
      "learning_rate": 0.00018409875183602158,
      "loss": 1.0377,
      "step": 1922
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5721745491027832,
      "learning_rate": 0.00018408248538037594,
      "loss": 0.9665,
      "step": 1923
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6899604201316833,
      "learning_rate": 0.00018406621132835383,
      "loss": 1.2337,
      "step": 1924
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.604383647441864,
      "learning_rate": 0.00018404992968142552,
      "loss": 1.0569,
      "step": 1925
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5258768200874329,
      "learning_rate": 0.00018403364044106195,
      "loss": 1.067,
      "step": 1926
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5347247123718262,
      "learning_rate": 0.00018401734360873478,
      "loss": 1.0679,
      "step": 1927
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5636218786239624,
      "learning_rate": 0.0001840010391859163,
      "loss": 1.0746,
      "step": 1928
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6444405913352966,
      "learning_rate": 0.00018398472717407954,
      "loss": 1.1184,
      "step": 1929
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6043636202812195,
      "learning_rate": 0.00018396840757469818,
      "loss": 1.0266,
      "step": 1930
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5988621115684509,
      "learning_rate": 0.00018395208038924667,
      "loss": 0.9645,
      "step": 1931
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5031013488769531,
      "learning_rate": 0.0001839357456192,
      "loss": 0.9798,
      "step": 1932
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6296197175979614,
      "learning_rate": 0.00018391940326603395,
      "loss": 0.9574,
      "step": 1933
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7154589891433716,
      "learning_rate": 0.000183903053331225,
      "loss": 1.233,
      "step": 1934
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5253052711486816,
      "learning_rate": 0.00018388669581625023,
      "loss": 1.051,
      "step": 1935
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6451107263565063,
      "learning_rate": 0.00018387033072258743,
      "loss": 1.1662,
      "step": 1936
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6134210228919983,
      "learning_rate": 0.00018385395805171515,
      "loss": 1.2124,
      "step": 1937
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.521445631980896,
      "learning_rate": 0.00018383757780511252,
      "loss": 1.066,
      "step": 1938
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6145197749137878,
      "learning_rate": 0.00018382118998425944,
      "loss": 1.1536,
      "step": 1939
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5896132588386536,
      "learning_rate": 0.00018380479459063643,
      "loss": 1.1371,
      "step": 1940
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5753506422042847,
      "learning_rate": 0.0001837883916257247,
      "loss": 1.2127,
      "step": 1941
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6061387062072754,
      "learning_rate": 0.00018377198109100626,
      "loss": 0.9953,
      "step": 1942
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6697282195091248,
      "learning_rate": 0.0001837555629879636,
      "loss": 1.2423,
      "step": 1943
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5806127786636353,
      "learning_rate": 0.00018373913731808007,
      "loss": 1.1304,
      "step": 1944
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.538391649723053,
      "learning_rate": 0.0001837227040828396,
      "loss": 1.0623,
      "step": 1945
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6214932799339294,
      "learning_rate": 0.00018370626328372683,
      "loss": 1.147,
      "step": 1946
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5707278251647949,
      "learning_rate": 0.00018368981492222717,
      "loss": 1.0347,
      "step": 1947
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5317419767379761,
      "learning_rate": 0.00018367335899982653,
      "loss": 1.1575,
      "step": 1948
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6756814122200012,
      "learning_rate": 0.00018365689551801172,
      "loss": 1.1641,
      "step": 1949
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5761415958404541,
      "learning_rate": 0.00018364042447827,
      "loss": 1.2391,
      "step": 1950
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6367659568786621,
      "learning_rate": 0.00018362394588208953,
      "loss": 1.0521,
      "step": 1951
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.612143874168396,
      "learning_rate": 0.00018360745973095902,
      "loss": 1.1476,
      "step": 1952
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6905182600021362,
      "learning_rate": 0.00018359096602636794,
      "loss": 1.1497,
      "step": 1953
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6150467395782471,
      "learning_rate": 0.00018357446476980634,
      "loss": 1.0446,
      "step": 1954
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6133366227149963,
      "learning_rate": 0.00018355795596276502,
      "loss": 1.0702,
      "step": 1955
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8109869360923767,
      "learning_rate": 0.00018354143960673552,
      "loss": 1.1479,
      "step": 1956
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5375014543533325,
      "learning_rate": 0.00018352491570320997,
      "loss": 1.0464,
      "step": 1957
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5784035325050354,
      "learning_rate": 0.00018350838425368115,
      "loss": 1.0479,
      "step": 1958
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6554840207099915,
      "learning_rate": 0.0001834918452596427,
      "loss": 1.0229,
      "step": 1959
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5233038067817688,
      "learning_rate": 0.00018347529872258874,
      "loss": 1.1293,
      "step": 1960
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5993428826332092,
      "learning_rate": 0.00018345874464401412,
      "loss": 1.0875,
      "step": 1961
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5326312184333801,
      "learning_rate": 0.00018344218302541454,
      "loss": 1.0039,
      "step": 1962
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6286762356758118,
      "learning_rate": 0.00018342561386828615,
      "loss": 1.1151,
      "step": 1963
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6470648646354675,
      "learning_rate": 0.00018340903717412588,
      "loss": 1.2054,
      "step": 1964
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5421403050422668,
      "learning_rate": 0.0001833924529444314,
      "loss": 1.1086,
      "step": 1965
    },
    {
      "epoch": 0.75,
      "eval_loss": 0.9726712107658386,
      "eval_runtime": 17.8102,
      "eval_samples_per_second": 153.227,
      "eval_steps_per_second": 38.349,
      "step": 1965
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.59051114320755,
      "learning_rate": 0.00018337586118070091,
      "loss": 1.1655,
      "step": 1966
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5358890891075134,
      "learning_rate": 0.00018335926188443348,
      "loss": 0.8833,
      "step": 1967
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.590603232383728,
      "learning_rate": 0.0001833426550571287,
      "loss": 1.2004,
      "step": 1968
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5690146684646606,
      "learning_rate": 0.00018332604070028693,
      "loss": 1.0583,
      "step": 1969
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5618229508399963,
      "learning_rate": 0.00018330941881540915,
      "loss": 1.0755,
      "step": 1970
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6254324316978455,
      "learning_rate": 0.00018329278940399713,
      "loss": 1.214,
      "step": 1971
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6175144910812378,
      "learning_rate": 0.0001832761524675531,
      "loss": 1.192,
      "step": 1972
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5359880924224854,
      "learning_rate": 0.0001832595080075803,
      "loss": 0.9377,
      "step": 1973
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5933471322059631,
      "learning_rate": 0.0001832428560255823,
      "loss": 1.1249,
      "step": 1974
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7249235510826111,
      "learning_rate": 0.00018322619652306362,
      "loss": 1.0241,
      "step": 1975
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6383464932441711,
      "learning_rate": 0.00018320952950152928,
      "loss": 1.2613,
      "step": 1976
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5812982320785522,
      "learning_rate": 0.00018319285496248508,
      "loss": 1.0963,
      "step": 1977
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6639915704727173,
      "learning_rate": 0.00018317617290743748,
      "loss": 1.0557,
      "step": 1978
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6422559022903442,
      "learning_rate": 0.00018315948333789362,
      "loss": 1.0014,
      "step": 1979
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5967164039611816,
      "learning_rate": 0.00018314278625536126,
      "loss": 1.0322,
      "step": 1980
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6338431239128113,
      "learning_rate": 0.00018312608166134892,
      "loss": 1.1217,
      "step": 1981
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5600549578666687,
      "learning_rate": 0.00018310936955736579,
      "loss": 1.2031,
      "step": 1982
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.640558660030365,
      "learning_rate": 0.00018309264994492163,
      "loss": 1.1203,
      "step": 1983
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6528109312057495,
      "learning_rate": 0.00018307592282552706,
      "loss": 1.1409,
      "step": 1984
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5706397891044617,
      "learning_rate": 0.0001830591882006932,
      "loss": 1.0011,
      "step": 1985
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6276172399520874,
      "learning_rate": 0.00018304244607193198,
      "loss": 1.2437,
      "step": 1986
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6464489102363586,
      "learning_rate": 0.00018302569644075597,
      "loss": 1.049,
      "step": 1987
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5966342091560364,
      "learning_rate": 0.00018300893930867838,
      "loss": 1.2014,
      "step": 1988
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5682297945022583,
      "learning_rate": 0.0001829921746772131,
      "loss": 0.9777,
      "step": 1989
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6807078123092651,
      "learning_rate": 0.00018297540254787472,
      "loss": 1.0618,
      "step": 1990
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5794547200202942,
      "learning_rate": 0.00018295862292217857,
      "loss": 0.9338,
      "step": 1991
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5602817535400391,
      "learning_rate": 0.0001829418358016405,
      "loss": 1.1029,
      "step": 1992
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6040206551551819,
      "learning_rate": 0.0001829250411877772,
      "loss": 1.0249,
      "step": 1993
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6918087601661682,
      "learning_rate": 0.00018290823908210594,
      "loss": 1.2849,
      "step": 1994
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6619887351989746,
      "learning_rate": 0.0001828914294861447,
      "loss": 1.1101,
      "step": 1995
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6102719902992249,
      "learning_rate": 0.00018287461240141216,
      "loss": 1.0639,
      "step": 1996
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6364079117774963,
      "learning_rate": 0.00018285778782942762,
      "loss": 1.2199,
      "step": 1997
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6381446123123169,
      "learning_rate": 0.0001828409557717111,
      "loss": 1.1562,
      "step": 1998
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6087682843208313,
      "learning_rate": 0.00018282411622978323,
      "loss": 1.0874,
      "step": 1999
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6255144476890564,
      "learning_rate": 0.00018280726920516546,
      "loss": 1.3216,
      "step": 2000
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5824018716812134,
      "learning_rate": 0.00018279041469937974,
      "loss": 1.2186,
      "step": 2001
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6130787134170532,
      "learning_rate": 0.0001827735527139488,
      "loss": 1.164,
      "step": 2002
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6037015318870544,
      "learning_rate": 0.00018275668325039605,
      "loss": 1.2456,
      "step": 2003
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5351065397262573,
      "learning_rate": 0.00018273980631024555,
      "loss": 0.9892,
      "step": 2004
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6300187706947327,
      "learning_rate": 0.00018272292189502198,
      "loss": 1.0067,
      "step": 2005
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5483942031860352,
      "learning_rate": 0.00018270603000625083,
      "loss": 1.0425,
      "step": 2006
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.607746422290802,
      "learning_rate": 0.00018268913064545814,
      "loss": 0.9913,
      "step": 2007
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6014259457588196,
      "learning_rate": 0.00018267222381417066,
      "loss": 1.0893,
      "step": 2008
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6995277404785156,
      "learning_rate": 0.0001826553095139159,
      "loss": 1.2404,
      "step": 2009
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6149386763572693,
      "learning_rate": 0.00018263838774622187,
      "loss": 1.2246,
      "step": 2010
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5554689764976501,
      "learning_rate": 0.0001826214585126174,
      "loss": 1.1945,
      "step": 2011
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6338213086128235,
      "learning_rate": 0.00018260452181463198,
      "loss": 1.1132,
      "step": 2012
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5740888118743896,
      "learning_rate": 0.0001825875776537957,
      "loss": 1.1365,
      "step": 2013
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6145991683006287,
      "learning_rate": 0.0001825706260316394,
      "loss": 1.1372,
      "step": 2014
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5843204855918884,
      "learning_rate": 0.00018255366694969457,
      "loss": 1.0638,
      "step": 2015
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7211287021636963,
      "learning_rate": 0.00018253670040949333,
      "loss": 1.1015,
      "step": 2016
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6173563599586487,
      "learning_rate": 0.00018251972641256854,
      "loss": 1.1816,
      "step": 2017
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6879176497459412,
      "learning_rate": 0.00018250274496045367,
      "loss": 1.2768,
      "step": 2018
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5964239239692688,
      "learning_rate": 0.00018248575605468294,
      "loss": 1.1307,
      "step": 2019
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5030563473701477,
      "learning_rate": 0.00018246875969679118,
      "loss": 0.709,
      "step": 2020
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6097801327705383,
      "learning_rate": 0.00018245175588831392,
      "loss": 1.0092,
      "step": 2021
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6590239405632019,
      "learning_rate": 0.00018243474463078735,
      "loss": 1.1461,
      "step": 2022
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5847734808921814,
      "learning_rate": 0.00018241772592574834,
      "loss": 1.1958,
      "step": 2023
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5965940356254578,
      "learning_rate": 0.00018240069977473446,
      "loss": 1.2313,
      "step": 2024
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6437633633613586,
      "learning_rate": 0.0001823836661792839,
      "loss": 1.0947,
      "step": 2025
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5720009803771973,
      "learning_rate": 0.00018236662514093557,
      "loss": 1.0711,
      "step": 2026
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6008542776107788,
      "learning_rate": 0.000182349576661229,
      "loss": 1.1588,
      "step": 2027
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.612463116645813,
      "learning_rate": 0.00018233252074170442,
      "loss": 0.9719,
      "step": 2028
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5129092931747437,
      "learning_rate": 0.0001823154573839028,
      "loss": 0.932,
      "step": 2029
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7150174379348755,
      "learning_rate": 0.00018229838658936564,
      "loss": 1.2633,
      "step": 2030
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6581935286521912,
      "learning_rate": 0.00018228130835963526,
      "loss": 1.0565,
      "step": 2031
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5770524144172668,
      "learning_rate": 0.0001822642226962545,
      "loss": 1.0407,
      "step": 2032
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5931928157806396,
      "learning_rate": 0.000182247129600767,
      "loss": 1.1073,
      "step": 2033
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5641647577285767,
      "learning_rate": 0.00018223002907471708,
      "loss": 1.0942,
      "step": 2034
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5922268629074097,
      "learning_rate": 0.00018221292111964959,
      "loss": 1.1774,
      "step": 2035
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.637740969657898,
      "learning_rate": 0.00018219580573711016,
      "loss": 1.0944,
      "step": 2036
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5891668200492859,
      "learning_rate": 0.00018217868292864505,
      "loss": 1.2074,
      "step": 2037
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5810216665267944,
      "learning_rate": 0.00018216155269580123,
      "loss": 1.1021,
      "step": 2038
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5588609576225281,
      "learning_rate": 0.00018214441504012637,
      "loss": 1.0642,
      "step": 2039
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5198193192481995,
      "learning_rate": 0.00018212726996316868,
      "loss": 0.8633,
      "step": 2040
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6176603436470032,
      "learning_rate": 0.00018211011746647717,
      "loss": 1.0139,
      "step": 2041
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5617534518241882,
      "learning_rate": 0.00018209295755160142,
      "loss": 1.0715,
      "step": 2042
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6227233409881592,
      "learning_rate": 0.0001820757902200918,
      "loss": 1.0053,
      "step": 2043
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5912638306617737,
      "learning_rate": 0.00018205861547349923,
      "loss": 1.147,
      "step": 2044
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5654580593109131,
      "learning_rate": 0.00018204143331337536,
      "loss": 1.0996,
      "step": 2045
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7012020945549011,
      "learning_rate": 0.0001820242437412725,
      "loss": 1.0514,
      "step": 2046
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6119007468223572,
      "learning_rate": 0.00018200704675874367,
      "loss": 1.0164,
      "step": 2047
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6899349689483643,
      "learning_rate": 0.00018198984236734246,
      "loss": 1.0248,
      "step": 2048
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6254860758781433,
      "learning_rate": 0.00018197263056862325,
      "loss": 1.1599,
      "step": 2049
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5398896932601929,
      "learning_rate": 0.00018195541136414097,
      "loss": 1.0355,
      "step": 2050
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5548747777938843,
      "learning_rate": 0.00018193818475545129,
      "loss": 1.0282,
      "step": 2051
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.591780424118042,
      "learning_rate": 0.00018192095074411058,
      "loss": 1.2166,
      "step": 2052
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6100587844848633,
      "learning_rate": 0.00018190370933167578,
      "loss": 1.1181,
      "step": 2053
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.699583113193512,
      "learning_rate": 0.0001818864605197046,
      "loss": 1.1278,
      "step": 2054
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6051424741744995,
      "learning_rate": 0.00018186920430975534,
      "loss": 1.0937,
      "step": 2055
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.550672173500061,
      "learning_rate": 0.00018185194070338703,
      "loss": 1.0416,
      "step": 2056
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5645923018455505,
      "learning_rate": 0.0001818346697021593,
      "loss": 1.0335,
      "step": 2057
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5920846462249756,
      "learning_rate": 0.0001818173913076325,
      "loss": 1.173,
      "step": 2058
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7178803086280823,
      "learning_rate": 0.00018180010552136769,
      "loss": 0.9348,
      "step": 2059
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5697028040885925,
      "learning_rate": 0.00018178281234492648,
      "loss": 1.1084,
      "step": 2060
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6761431097984314,
      "learning_rate": 0.0001817655117798712,
      "loss": 1.1577,
      "step": 2061
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.573343813419342,
      "learning_rate": 0.00018174820382776488,
      "loss": 1.1266,
      "step": 2062
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5408897399902344,
      "learning_rate": 0.00018173088849017123,
      "loss": 1.054,
      "step": 2063
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6230862140655518,
      "learning_rate": 0.00018171356576865457,
      "loss": 1.0538,
      "step": 2064
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5958811640739441,
      "learning_rate": 0.0001816962356647799,
      "loss": 1.226,
      "step": 2065
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5987549424171448,
      "learning_rate": 0.0001816788981801129,
      "loss": 1.1471,
      "step": 2066
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6187098622322083,
      "learning_rate": 0.0001816615533162199,
      "loss": 1.242,
      "step": 2067
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5478816032409668,
      "learning_rate": 0.0001816442010746679,
      "loss": 1.0554,
      "step": 2068
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6243496537208557,
      "learning_rate": 0.00018162684145702466,
      "loss": 0.9563,
      "step": 2069
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6071329712867737,
      "learning_rate": 0.00018160947446485844,
      "loss": 1.1325,
      "step": 2070
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6612725853919983,
      "learning_rate": 0.00018159210009973826,
      "loss": 1.1448,
      "step": 2071
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6862730979919434,
      "learning_rate": 0.00018157471836323383,
      "loss": 1.2824,
      "step": 2072
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.604934811592102,
      "learning_rate": 0.00018155732925691543,
      "loss": 1.0633,
      "step": 2073
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6084781885147095,
      "learning_rate": 0.00018153993278235416,
      "loss": 1.1974,
      "step": 2074
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6043122410774231,
      "learning_rate": 0.00018152252894112163,
      "loss": 1.1033,
      "step": 2075
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6114141941070557,
      "learning_rate": 0.00018150511773479017,
      "loss": 1.1386,
      "step": 2076
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6719943284988403,
      "learning_rate": 0.0001814876991649328,
      "loss": 1.0866,
      "step": 2077
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5567785501480103,
      "learning_rate": 0.00018147027323312323,
      "loss": 0.9184,
      "step": 2078
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5580743551254272,
      "learning_rate": 0.0001814528399409357,
      "loss": 1.0469,
      "step": 2079
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5721143484115601,
      "learning_rate": 0.00018143539928994532,
      "loss": 1.0805,
      "step": 2080
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5623195171356201,
      "learning_rate": 0.00018141795128172767,
      "loss": 1.1016,
      "step": 2081
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6376780271530151,
      "learning_rate": 0.0001814004959178591,
      "loss": 1.002,
      "step": 2082
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6629352569580078,
      "learning_rate": 0.00018138303319991663,
      "loss": 1.1561,
      "step": 2083
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5993563532829285,
      "learning_rate": 0.0001813655631294779,
      "loss": 1.063,
      "step": 2084
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5828850865364075,
      "learning_rate": 0.00018134808570812125,
      "loss": 1.1868,
      "step": 2085
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5925980806350708,
      "learning_rate": 0.0001813306009374256,
      "loss": 1.0633,
      "step": 2086
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6339349746704102,
      "learning_rate": 0.0001813131088189707,
      "loss": 1.0335,
      "step": 2087
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5569525361061096,
      "learning_rate": 0.00018129560935433678,
      "loss": 0.9869,
      "step": 2088
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5449732542037964,
      "learning_rate": 0.00018127810254510486,
      "loss": 1.2209,
      "step": 2089
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6194440722465515,
      "learning_rate": 0.00018126058839285656,
      "loss": 1.0826,
      "step": 2090
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5619876384735107,
      "learning_rate": 0.0001812430668991742,
      "loss": 1.1765,
      "step": 2091
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5702633261680603,
      "learning_rate": 0.0001812255380656408,
      "loss": 1.1087,
      "step": 2092
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6091815233230591,
      "learning_rate": 0.00018120800189383985,
      "loss": 1.1604,
      "step": 2093
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6504197120666504,
      "learning_rate": 0.00018119045838535573,
      "loss": 0.9817,
      "step": 2094
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5884442329406738,
      "learning_rate": 0.00018117290754177345,
      "loss": 0.9892,
      "step": 2095
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5776565074920654,
      "learning_rate": 0.00018115534936467853,
      "loss": 1.0363,
      "step": 2096
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6103533506393433,
      "learning_rate": 0.00018113778385565733,
      "loss": 1.244,
      "step": 2097
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6736485958099365,
      "learning_rate": 0.0001811202110162967,
      "loss": 1.0085,
      "step": 2098
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7142513990402222,
      "learning_rate": 0.00018110263084818436,
      "loss": 1.1899,
      "step": 2099
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6481114029884338,
      "learning_rate": 0.0001810850433529085,
      "loss": 1.1406,
      "step": 2100
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5257217288017273,
      "learning_rate": 0.0001810674485320581,
      "loss": 1.1025,
      "step": 2101
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.538659393787384,
      "learning_rate": 0.0001810498463872227,
      "loss": 1.0308,
      "step": 2102
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5943474769592285,
      "learning_rate": 0.0001810322369199926,
      "loss": 1.0064,
      "step": 2103
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6814498901367188,
      "learning_rate": 0.00018101462013195868,
      "loss": 1.0761,
      "step": 2104
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6166239976882935,
      "learning_rate": 0.00018099699602471255,
      "loss": 0.9444,
      "step": 2105
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6369709372520447,
      "learning_rate": 0.00018097936459984647,
      "loss": 1.1837,
      "step": 2106
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5224598050117493,
      "learning_rate": 0.00018096172585895323,
      "loss": 1.2496,
      "step": 2107
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6199701428413391,
      "learning_rate": 0.0001809440798036265,
      "loss": 1.0938,
      "step": 2108
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6109820008277893,
      "learning_rate": 0.00018092642643546047,
      "loss": 1.0366,
      "step": 2109
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6200912594795227,
      "learning_rate": 0.00018090876575604998,
      "loss": 1.094,
      "step": 2110
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5744567513465881,
      "learning_rate": 0.00018089109776699066,
      "loss": 1.0253,
      "step": 2111
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6033176779747009,
      "learning_rate": 0.00018087342246987862,
      "loss": 0.9497,
      "step": 2112
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6371310949325562,
      "learning_rate": 0.0001808557398663108,
      "loss": 1.0781,
      "step": 2113
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6146370768547058,
      "learning_rate": 0.00018083804995788465,
      "loss": 1.0779,
      "step": 2114
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5866037011146545,
      "learning_rate": 0.00018082035274619843,
      "loss": 1.1211,
      "step": 2115
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5422040820121765,
      "learning_rate": 0.0001808026482328509,
      "loss": 0.8366,
      "step": 2116
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5723130106925964,
      "learning_rate": 0.0001807849364194416,
      "loss": 1.0537,
      "step": 2117
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5562235713005066,
      "learning_rate": 0.00018076721730757073,
      "loss": 1.0842,
      "step": 2118
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.608705997467041,
      "learning_rate": 0.0001807494908988391,
      "loss": 1.1393,
      "step": 2119
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6037261486053467,
      "learning_rate": 0.0001807317571948481,
      "loss": 1.3242,
      "step": 2120
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6175596117973328,
      "learning_rate": 0.0001807140161972,
      "loss": 1.1733,
      "step": 2121
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.615269660949707,
      "learning_rate": 0.00018069626790749753,
      "loss": 1.0321,
      "step": 2122
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.603236734867096,
      "learning_rate": 0.00018067851232734413,
      "loss": 1.2018,
      "step": 2123
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5883955955505371,
      "learning_rate": 0.00018066074945834396,
      "loss": 1.1032,
      "step": 2124
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5292845368385315,
      "learning_rate": 0.00018064297930210174,
      "loss": 0.9332,
      "step": 2125
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6402313709259033,
      "learning_rate": 0.000180625201860223,
      "loss": 1.0735,
      "step": 2126
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5985416769981384,
      "learning_rate": 0.00018060741713431372,
      "loss": 1.0645,
      "step": 2127
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5230291485786438,
      "learning_rate": 0.00018058962512598075,
      "loss": 0.9831,
      "step": 2128
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6098759770393372,
      "learning_rate": 0.0001805718258368314,
      "loss": 1.1876,
      "step": 2129
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7077792882919312,
      "learning_rate": 0.00018055401926847379,
      "loss": 0.9392,
      "step": 2130
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6847090721130371,
      "learning_rate": 0.00018053620542251666,
      "loss": 1.0481,
      "step": 2131
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5998674631118774,
      "learning_rate": 0.00018051838430056935,
      "loss": 0.916,
      "step": 2132
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6179681420326233,
      "learning_rate": 0.0001805005559042419,
      "loss": 1.0466,
      "step": 2133
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6567022204399109,
      "learning_rate": 0.00018048272023514503,
      "loss": 1.2206,
      "step": 2134
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6526135206222534,
      "learning_rate": 0.00018046487729489008,
      "loss": 1.2091,
      "step": 2135
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.605446457862854,
      "learning_rate": 0.00018044702708508908,
      "loss": 1.2436,
      "step": 2136
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6243354678153992,
      "learning_rate": 0.00018042916960735463,
      "loss": 1.3055,
      "step": 2137
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5263623595237732,
      "learning_rate": 0.0001804113048633001,
      "loss": 0.8699,
      "step": 2138
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6840293407440186,
      "learning_rate": 0.0001803934328545395,
      "loss": 1.0613,
      "step": 2139
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6112801432609558,
      "learning_rate": 0.00018037555358268737,
      "loss": 1.0718,
      "step": 2140
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5487591028213501,
      "learning_rate": 0.0001803576670493591,
      "loss": 0.9538,
      "step": 2141
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5530500411987305,
      "learning_rate": 0.00018033977325617058,
      "loss": 0.9213,
      "step": 2142
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5962365865707397,
      "learning_rate": 0.00018032187220473847,
      "loss": 1.0206,
      "step": 2143
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6077033877372742,
      "learning_rate": 0.00018030396389667996,
      "loss": 1.1739,
      "step": 2144
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6256138682365417,
      "learning_rate": 0.00018028604833361298,
      "loss": 1.2033,
      "step": 2145
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5926710367202759,
      "learning_rate": 0.0001802681255171561,
      "loss": 0.9579,
      "step": 2146
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.603348433971405,
      "learning_rate": 0.00018025019544892857,
      "loss": 1.2326,
      "step": 2147
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.576933741569519,
      "learning_rate": 0.00018023225813055028,
      "loss": 1.1138,
      "step": 2148
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6577118635177612,
      "learning_rate": 0.00018021431356364173,
      "loss": 1.1375,
      "step": 2149
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6184905767440796,
      "learning_rate": 0.00018019636174982414,
      "loss": 1.2606,
      "step": 2150
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5773909687995911,
      "learning_rate": 0.00018017840269071928,
      "loss": 1.1278,
      "step": 2151
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.593681812286377,
      "learning_rate": 0.00018016043638794974,
      "loss": 1.0683,
      "step": 2152
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5299533009529114,
      "learning_rate": 0.00018014246284313865,
      "loss": 1.0488,
      "step": 2153
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5954784750938416,
      "learning_rate": 0.00018012448205790977,
      "loss": 0.9962,
      "step": 2154
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5385948419570923,
      "learning_rate": 0.00018010649403388764,
      "loss": 0.9248,
      "step": 2155
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5898330807685852,
      "learning_rate": 0.0001800884987726973,
      "loss": 1.1551,
      "step": 2156
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6078624129295349,
      "learning_rate": 0.0001800704962759646,
      "loss": 1.0074,
      "step": 2157
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5408656597137451,
      "learning_rate": 0.00018005248654531588,
      "loss": 0.9395,
      "step": 2158
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5705100893974304,
      "learning_rate": 0.00018003446958237828,
      "loss": 1.0721,
      "step": 2159
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5572664737701416,
      "learning_rate": 0.0001800164453887795,
      "loss": 0.8564,
      "step": 2160
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6223136186599731,
      "learning_rate": 0.0001799984139661479,
      "loss": 1.0628,
      "step": 2161
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5901376008987427,
      "learning_rate": 0.00017998037531611263,
      "loss": 1.1279,
      "step": 2162
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.581200122833252,
      "learning_rate": 0.00017996232944030323,
      "loss": 1.2537,
      "step": 2163
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6000811457633972,
      "learning_rate": 0.00017994427634035015,
      "loss": 1.0852,
      "step": 2164
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.608558177947998,
      "learning_rate": 0.00017992621601788428,
      "loss": 1.1352,
      "step": 2165
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6038708090782166,
      "learning_rate": 0.00017990814847453738,
      "loss": 1.3391,
      "step": 2166
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.588092029094696,
      "learning_rate": 0.00017989007371194173,
      "loss": 1.2182,
      "step": 2167
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5584761500358582,
      "learning_rate": 0.00017987199173173023,
      "loss": 1.1986,
      "step": 2168
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6125190258026123,
      "learning_rate": 0.0001798539025355365,
      "loss": 1.1643,
      "step": 2169
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5272698402404785,
      "learning_rate": 0.00017983580612499482,
      "loss": 0.9866,
      "step": 2170
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6117221117019653,
      "learning_rate": 0.00017981770250174005,
      "loss": 0.988,
      "step": 2171
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5380080342292786,
      "learning_rate": 0.00017979959166740782,
      "loss": 1.0272,
      "step": 2172
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6345922350883484,
      "learning_rate": 0.00017978147362363432,
      "loss": 0.9944,
      "step": 2173
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.685167670249939,
      "learning_rate": 0.0001797633483720564,
      "loss": 1.0927,
      "step": 2174
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.682863712310791,
      "learning_rate": 0.00017974521591431156,
      "loss": 1.1548,
      "step": 2175
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5539425015449524,
      "learning_rate": 0.00017972707625203796,
      "loss": 1.1734,
      "step": 2176
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6049844026565552,
      "learning_rate": 0.00017970892938687445,
      "loss": 1.0368,
      "step": 2177
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5493163466453552,
      "learning_rate": 0.00017969077532046048,
      "loss": 1.0534,
      "step": 2178
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5946351885795593,
      "learning_rate": 0.00017967261405443617,
      "loss": 1.0142,
      "step": 2179
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5894339680671692,
      "learning_rate": 0.00017965444559044227,
      "loss": 1.118,
      "step": 2180
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5812546014785767,
      "learning_rate": 0.00017963626993012022,
      "loss": 1.1465,
      "step": 2181
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7265496850013733,
      "learning_rate": 0.0001796180870751121,
      "loss": 0.9756,
      "step": 2182
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5403939485549927,
      "learning_rate": 0.00017959989702706058,
      "loss": 0.9606,
      "step": 2183
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6526712775230408,
      "learning_rate": 0.0001795816997876091,
      "loss": 1.0271,
      "step": 2184
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5255093574523926,
      "learning_rate": 0.0001795634953584016,
      "loss": 0.8756,
      "step": 2185
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6554507613182068,
      "learning_rate": 0.0001795452837410828,
      "loss": 1.2341,
      "step": 2186
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6162861585617065,
      "learning_rate": 0.000179527064937298,
      "loss": 1.2453,
      "step": 2187
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5804990530014038,
      "learning_rate": 0.0001795088389486931,
      "loss": 1.172,
      "step": 2188
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6742739677429199,
      "learning_rate": 0.00017949060577691488,
      "loss": 1.0933,
      "step": 2189
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5167546272277832,
      "learning_rate": 0.00017947236542361044,
      "loss": 1.0269,
      "step": 2190
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6206005215644836,
      "learning_rate": 0.00017945411789042776,
      "loss": 0.9725,
      "step": 2191
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.58721524477005,
      "learning_rate": 0.0001794358631790154,
      "loss": 1.0485,
      "step": 2192
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5348027348518372,
      "learning_rate": 0.00017941760129102255,
      "loss": 0.9812,
      "step": 2193
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5457707047462463,
      "learning_rate": 0.00017939933222809915,
      "loss": 1.0419,
      "step": 2194
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6082179546356201,
      "learning_rate": 0.00017938105599189558,
      "loss": 1.1514,
      "step": 2195
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5660126209259033,
      "learning_rate": 0.00017936277258406307,
      "loss": 1.0089,
      "step": 2196
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5888327956199646,
      "learning_rate": 0.00017934448200625342,
      "loss": 0.9221,
      "step": 2197
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.638171374797821,
      "learning_rate": 0.00017932618426011907,
      "loss": 1.0189,
      "step": 2198
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7061122059822083,
      "learning_rate": 0.00017930787934731311,
      "loss": 1.1216,
      "step": 2199
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5726503729820251,
      "learning_rate": 0.00017928956726948926,
      "loss": 1.075,
      "step": 2200
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5943982601165771,
      "learning_rate": 0.000179271248028302,
      "loss": 0.8889,
      "step": 2201
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6812533736228943,
      "learning_rate": 0.00017925292162540628,
      "loss": 1.2345,
      "step": 2202
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6768470406532288,
      "learning_rate": 0.00017923458806245782,
      "loss": 1.2374,
      "step": 2203
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5991680026054382,
      "learning_rate": 0.00017921624734111292,
      "loss": 1.08,
      "step": 2204
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6172266602516174,
      "learning_rate": 0.00017919789946302866,
      "loss": 1.0463,
      "step": 2205
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8502815961837769,
      "learning_rate": 0.00017917954442986256,
      "loss": 1.089,
      "step": 2206
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5796491503715515,
      "learning_rate": 0.00017916118224327296,
      "loss": 1.1364,
      "step": 2207
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6637051105499268,
      "learning_rate": 0.00017914281290491873,
      "loss": 1.205,
      "step": 2208
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6383735537528992,
      "learning_rate": 0.0001791244364164595,
      "loss": 0.9944,
      "step": 2209
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5945635437965393,
      "learning_rate": 0.0001791060527795554,
      "loss": 1.0487,
      "step": 2210
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6733584403991699,
      "learning_rate": 0.00017908766199586733,
      "loss": 1.1305,
      "step": 2211
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5791917443275452,
      "learning_rate": 0.00017906926406705678,
      "loss": 1.17,
      "step": 2212
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5425911545753479,
      "learning_rate": 0.00017905085899478593,
      "loss": 1.1022,
      "step": 2213
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6071683168411255,
      "learning_rate": 0.00017903244678071757,
      "loss": 1.1017,
      "step": 2214
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6542885899543762,
      "learning_rate": 0.00017901402742651512,
      "loss": 1.1805,
      "step": 2215
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6654801964759827,
      "learning_rate": 0.00017899560093384263,
      "loss": 1.2689,
      "step": 2216
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5979436039924622,
      "learning_rate": 0.00017897716730436492,
      "loss": 0.9942,
      "step": 2217
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6463541388511658,
      "learning_rate": 0.00017895872653974729,
      "loss": 1.0664,
      "step": 2218
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5794387459754944,
      "learning_rate": 0.00017894027864165577,
      "loss": 0.9069,
      "step": 2219
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6222195029258728,
      "learning_rate": 0.00017892182361175702,
      "loss": 1.1433,
      "step": 2220
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5939921140670776,
      "learning_rate": 0.0001789033614517184,
      "loss": 1.1035,
      "step": 2221
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6801263689994812,
      "learning_rate": 0.00017888489216320782,
      "loss": 1.3398,
      "step": 2222
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6073567867279053,
      "learning_rate": 0.0001788664157478938,
      "loss": 1.1463,
      "step": 2223
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5933486223220825,
      "learning_rate": 0.00017884793220744575,
      "loss": 1.0512,
      "step": 2224
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6283493638038635,
      "learning_rate": 0.00017882944154353343,
      "loss": 1.1488,
      "step": 2225
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5770262479782104,
      "learning_rate": 0.0001788109437578274,
      "loss": 1.1998,
      "step": 2226
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5506201982498169,
      "learning_rate": 0.00017879243885199884,
      "loss": 1.1062,
      "step": 2227
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5642320513725281,
      "learning_rate": 0.00017877392682771954,
      "loss": 1.0807,
      "step": 2228
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5464077591896057,
      "learning_rate": 0.00017875540768666194,
      "loss": 0.9634,
      "step": 2229
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5671146512031555,
      "learning_rate": 0.0001787368814304992,
      "loss": 1.189,
      "step": 2230
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5678768754005432,
      "learning_rate": 0.00017871834806090501,
      "loss": 1.0745,
      "step": 2231
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5408884286880493,
      "learning_rate": 0.0001786998075795538,
      "loss": 1.152,
      "step": 2232
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5448105931282043,
      "learning_rate": 0.00017868125998812056,
      "loss": 1.0465,
      "step": 2233
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6558840870857239,
      "learning_rate": 0.000178662705288281,
      "loss": 1.053,
      "step": 2234
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5681583881378174,
      "learning_rate": 0.00017864414348171134,
      "loss": 0.8673,
      "step": 2235
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.577218234539032,
      "learning_rate": 0.00017862557457008866,
      "loss": 1.0605,
      "step": 2236
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6256998181343079,
      "learning_rate": 0.00017860699855509044,
      "loss": 1.087,
      "step": 2237
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7601686120033264,
      "learning_rate": 0.000178588415438395,
      "loss": 1.138,
      "step": 2238
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6093456745147705,
      "learning_rate": 0.0001785698252216812,
      "loss": 1.0941,
      "step": 2239
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5740756392478943,
      "learning_rate": 0.00017855122790662854,
      "loss": 1.2238,
      "step": 2240
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5732688307762146,
      "learning_rate": 0.00017853262349491723,
      "loss": 1.0036,
      "step": 2241
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5456250905990601,
      "learning_rate": 0.000178514011988228,
      "loss": 0.8441,
      "step": 2242
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5977802276611328,
      "learning_rate": 0.00017849539338824231,
      "loss": 1.0905,
      "step": 2243
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5805135369300842,
      "learning_rate": 0.00017847676769664232,
      "loss": 1.0668,
      "step": 2244
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.544512152671814,
      "learning_rate": 0.00017845813491511068,
      "loss": 1.0293,
      "step": 2245
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6788541078567505,
      "learning_rate": 0.00017843949504533078,
      "loss": 1.0246,
      "step": 2246
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5601325631141663,
      "learning_rate": 0.00017842084808898661,
      "loss": 1.0222,
      "step": 2247
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5927838683128357,
      "learning_rate": 0.00017840219404776283,
      "loss": 0.9626,
      "step": 2248
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.565467894077301,
      "learning_rate": 0.00017838353292334478,
      "loss": 0.9815,
      "step": 2249
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5185115933418274,
      "learning_rate": 0.0001783648647174183,
      "loss": 0.9391,
      "step": 2250
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6661456823348999,
      "learning_rate": 0.00017834618943167,
      "loss": 1.1952,
      "step": 2251
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5766065120697021,
      "learning_rate": 0.00017832750706778708,
      "loss": 1.0535,
      "step": 2252
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5823925733566284,
      "learning_rate": 0.0001783088176274574,
      "loss": 1.0761,
      "step": 2253
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5491415858268738,
      "learning_rate": 0.0001782901211123694,
      "loss": 1.0468,
      "step": 2254
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5908526182174683,
      "learning_rate": 0.00017827141752421223,
      "loss": 1.0112,
      "step": 2255
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5098611116409302,
      "learning_rate": 0.00017825270686467568,
      "loss": 0.8527,
      "step": 2256
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.536832869052887,
      "learning_rate": 0.00017823398913545014,
      "loss": 0.9449,
      "step": 2257
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5211824178695679,
      "learning_rate": 0.00017821526433822661,
      "loss": 1.105,
      "step": 2258
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6235193610191345,
      "learning_rate": 0.00017819653247469683,
      "loss": 1.048,
      "step": 2259
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5279130935668945,
      "learning_rate": 0.00017817779354655308,
      "loss": 1.1058,
      "step": 2260
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5793086886405945,
      "learning_rate": 0.0001781590475554883,
      "loss": 1.0004,
      "step": 2261
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5962411165237427,
      "learning_rate": 0.00017814029450319616,
      "loss": 1.0824,
      "step": 2262
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5811233520507812,
      "learning_rate": 0.00017812153439137078,
      "loss": 0.9964,
      "step": 2263
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5765413641929626,
      "learning_rate": 0.00017810276722170714,
      "loss": 1.0507,
      "step": 2264
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6243969202041626,
      "learning_rate": 0.00017808399299590067,
      "loss": 0.9657,
      "step": 2265
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6792131066322327,
      "learning_rate": 0.00017806521171564757,
      "loss": 1.0217,
      "step": 2266
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6137617230415344,
      "learning_rate": 0.00017804642338264457,
      "loss": 1.0808,
      "step": 2267
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6335213780403137,
      "learning_rate": 0.00017802762799858912,
      "loss": 1.0909,
      "step": 2268
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5913608074188232,
      "learning_rate": 0.0001780088255651793,
      "loss": 1.1797,
      "step": 2269
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5856280326843262,
      "learning_rate": 0.00017799001608411375,
      "loss": 1.122,
      "step": 2270
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5854422450065613,
      "learning_rate": 0.0001779711995570918,
      "loss": 1.0876,
      "step": 2271
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5743233561515808,
      "learning_rate": 0.0001779523759858135,
      "loss": 1.0314,
      "step": 2272
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5637683868408203,
      "learning_rate": 0.00017793354537197934,
      "loss": 1.0113,
      "step": 2273
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5439072847366333,
      "learning_rate": 0.00017791470771729063,
      "loss": 1.1764,
      "step": 2274
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6216204166412354,
      "learning_rate": 0.00017789586302344924,
      "loss": 0.9853,
      "step": 2275
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5557587742805481,
      "learning_rate": 0.00017787701129215767,
      "loss": 0.9668,
      "step": 2276
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6265153884887695,
      "learning_rate": 0.00017785815252511908,
      "loss": 1.0337,
      "step": 2277
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6366235613822937,
      "learning_rate": 0.00017783928672403723,
      "loss": 1.3435,
      "step": 2278
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6350502967834473,
      "learning_rate": 0.00017782041389061655,
      "loss": 1.0933,
      "step": 2279
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6751552820205688,
      "learning_rate": 0.00017780153402656207,
      "loss": 1.0766,
      "step": 2280
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5644265413284302,
      "learning_rate": 0.00017778264713357952,
      "loss": 0.9216,
      "step": 2281
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5799874663352966,
      "learning_rate": 0.00017776375321337521,
      "loss": 0.9968,
      "step": 2282
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6158801913261414,
      "learning_rate": 0.0001777448522676561,
      "loss": 1.2637,
      "step": 2283
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.571167528629303,
      "learning_rate": 0.00017772594429812976,
      "loss": 1.0365,
      "step": 2284
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6398135423660278,
      "learning_rate": 0.00017770702930650445,
      "loss": 1.1111,
      "step": 2285
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5383583307266235,
      "learning_rate": 0.000177688107294489,
      "loss": 1.0693,
      "step": 2286
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.592910647392273,
      "learning_rate": 0.00017766917826379295,
      "loss": 1.0026,
      "step": 2287
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5524596571922302,
      "learning_rate": 0.0001776502422161264,
      "loss": 0.9801,
      "step": 2288
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5932091474533081,
      "learning_rate": 0.0001776312991532001,
      "loss": 1.1859,
      "step": 2289
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6059234738349915,
      "learning_rate": 0.00017761234907672549,
      "loss": 1.1686,
      "step": 2290
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5730354189872742,
      "learning_rate": 0.00017759339198841455,
      "loss": 1.1097,
      "step": 2291
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.545337975025177,
      "learning_rate": 0.00017757442788998002,
      "loss": 0.8885,
      "step": 2292
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7098504304885864,
      "learning_rate": 0.00017755545678313514,
      "loss": 1.2253,
      "step": 2293
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5590181946754456,
      "learning_rate": 0.00017753647866959385,
      "loss": 0.8259,
      "step": 2294
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.804749071598053,
      "learning_rate": 0.00017751749355107073,
      "loss": 1.0091,
      "step": 2295
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.540513277053833,
      "learning_rate": 0.00017749850142928096,
      "loss": 0.9203,
      "step": 2296
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5944012403488159,
      "learning_rate": 0.00017747950230594039,
      "loss": 1.0141,
      "step": 2297
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5729084014892578,
      "learning_rate": 0.00017746049618276545,
      "loss": 1.2178,
      "step": 2298
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6585800051689148,
      "learning_rate": 0.0001774414830614733,
      "loss": 0.9808,
      "step": 2299
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6925854086875916,
      "learning_rate": 0.00017742246294378159,
      "loss": 1.0978,
      "step": 2300
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6280820965766907,
      "learning_rate": 0.00017740343583140877,
      "loss": 1.0309,
      "step": 2301
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6000059247016907,
      "learning_rate": 0.00017738440172607373,
      "loss": 1.1107,
      "step": 2302
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5967001914978027,
      "learning_rate": 0.00017736536062949615,
      "loss": 1.0392,
      "step": 2303
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4993484318256378,
      "learning_rate": 0.0001773463125433963,
      "loss": 0.8711,
      "step": 2304
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5703268647193909,
      "learning_rate": 0.00017732725746949504,
      "loss": 1.0827,
      "step": 2305
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5676262378692627,
      "learning_rate": 0.0001773081954095139,
      "loss": 0.9512,
      "step": 2306
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5920840501785278,
      "learning_rate": 0.00017728912636517499,
      "loss": 1.0837,
      "step": 2307
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5826390385627747,
      "learning_rate": 0.00017727005033820115,
      "loss": 0.9103,
      "step": 2308
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.718740701675415,
      "learning_rate": 0.00017725096733031576,
      "loss": 1.1066,
      "step": 2309
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.593582034111023,
      "learning_rate": 0.00017723187734324287,
      "loss": 0.9903,
      "step": 2310
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5643889904022217,
      "learning_rate": 0.00017721278037870715,
      "loss": 1.0689,
      "step": 2311
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5696119666099548,
      "learning_rate": 0.00017719367643843388,
      "loss": 1.01,
      "step": 2312
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5642480254173279,
      "learning_rate": 0.00017717456552414902,
      "loss": 1.1625,
      "step": 2313
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.506824791431427,
      "learning_rate": 0.00017715544763757915,
      "loss": 0.9964,
      "step": 2314
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7369400262832642,
      "learning_rate": 0.0001771363227804514,
      "loss": 1.2391,
      "step": 2315
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5189867615699768,
      "learning_rate": 0.00017711719095449366,
      "loss": 1.0312,
      "step": 2316
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5520070791244507,
      "learning_rate": 0.00017709805216143432,
      "loss": 1.1712,
      "step": 2317
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5789135098457336,
      "learning_rate": 0.00017707890640300254,
      "loss": 1.1366,
      "step": 2318
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6521427631378174,
      "learning_rate": 0.00017705975368092794,
      "loss": 1.0508,
      "step": 2319
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5626633167266846,
      "learning_rate": 0.00017704059399694096,
      "loss": 1.1635,
      "step": 2320
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.685055136680603,
      "learning_rate": 0.00017702142735277247,
      "loss": 1.1603,
      "step": 2321
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5628002285957336,
      "learning_rate": 0.0001770022537501541,
      "loss": 0.9771,
      "step": 2322
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.631969153881073,
      "learning_rate": 0.00017698307319081815,
      "loss": 1.1721,
      "step": 2323
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.691218912601471,
      "learning_rate": 0.00017696388567649737,
      "loss": 1.2345,
      "step": 2324
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7056717276573181,
      "learning_rate": 0.0001769446912089253,
      "loss": 1.0213,
      "step": 2325
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5915741324424744,
      "learning_rate": 0.00017692548978983605,
      "loss": 1.2047,
      "step": 2326
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5657656192779541,
      "learning_rate": 0.00017690628142096433,
      "loss": 1.0482,
      "step": 2327
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6127615571022034,
      "learning_rate": 0.00017688706610404555,
      "loss": 1.1017,
      "step": 2328
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5989276170730591,
      "learning_rate": 0.00017686784384081565,
      "loss": 1.2391,
      "step": 2329
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.49377620220184326,
      "learning_rate": 0.0001768486146330113,
      "loss": 0.9627,
      "step": 2330
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6099086999893188,
      "learning_rate": 0.00017682937848236972,
      "loss": 1.0978,
      "step": 2331
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6590273976325989,
      "learning_rate": 0.00017681013539062884,
      "loss": 1.0869,
      "step": 2332
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5015077590942383,
      "learning_rate": 0.00017679088535952709,
      "loss": 1.082,
      "step": 2333
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5334069132804871,
      "learning_rate": 0.00017677162839080363,
      "loss": 1.0413,
      "step": 2334
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6499801278114319,
      "learning_rate": 0.00017675236448619824,
      "loss": 1.1829,
      "step": 2335
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.744327962398529,
      "learning_rate": 0.0001767330936474513,
      "loss": 1.2113,
      "step": 2336
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6160250902175903,
      "learning_rate": 0.0001767138158763038,
      "loss": 1.1968,
      "step": 2337
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6244109869003296,
      "learning_rate": 0.0001766945311744974,
      "loss": 1.1436,
      "step": 2338
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5933175683021545,
      "learning_rate": 0.00017667523954377433,
      "loss": 0.966,
      "step": 2339
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.621781051158905,
      "learning_rate": 0.0001766559409858775,
      "loss": 1.0309,
      "step": 2340
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5982803702354431,
      "learning_rate": 0.00017663663550255044,
      "loss": 1.2157,
      "step": 2341
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6001816391944885,
      "learning_rate": 0.00017661732309553728,
      "loss": 1.0603,
      "step": 2342
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5300409197807312,
      "learning_rate": 0.00017659800376658278,
      "loss": 1.0337,
      "step": 2343
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5697085857391357,
      "learning_rate": 0.00017657867751743234,
      "loss": 0.9433,
      "step": 2344
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5476548671722412,
      "learning_rate": 0.00017655934434983198,
      "loss": 1.0204,
      "step": 2345
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5725254416465759,
      "learning_rate": 0.0001765400042655283,
      "loss": 1.1516,
      "step": 2346
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.666840672492981,
      "learning_rate": 0.00017652065726626862,
      "loss": 1.0513,
      "step": 2347
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6322093605995178,
      "learning_rate": 0.00017650130335380084,
      "loss": 1.1666,
      "step": 2348
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6339440941810608,
      "learning_rate": 0.00017648194252987344,
      "loss": 1.02,
      "step": 2349
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5936554074287415,
      "learning_rate": 0.00017646257479623557,
      "loss": 1.0905,
      "step": 2350
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6231417655944824,
      "learning_rate": 0.000176443200154637,
      "loss": 1.1442,
      "step": 2351
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5736232399940491,
      "learning_rate": 0.00017642381860682812,
      "loss": 1.0681,
      "step": 2352
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6252473592758179,
      "learning_rate": 0.0001764044301545599,
      "loss": 1.1322,
      "step": 2353
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.56855708360672,
      "learning_rate": 0.00017638503479958404,
      "loss": 1.1346,
      "step": 2354
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6396982073783875,
      "learning_rate": 0.0001763656325436528,
      "loss": 1.1003,
      "step": 2355
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7210072875022888,
      "learning_rate": 0.000176346223388519,
      "loss": 1.1606,
      "step": 2356
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6793826222419739,
      "learning_rate": 0.00017632680733593623,
      "loss": 1.0003,
      "step": 2357
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6272571682929993,
      "learning_rate": 0.00017630738438765853,
      "loss": 1.0742,
      "step": 2358
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5288051962852478,
      "learning_rate": 0.00017628795454544075,
      "loss": 1.093,
      "step": 2359
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6225223541259766,
      "learning_rate": 0.0001762685178110382,
      "loss": 1.0729,
      "step": 2360
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5822621583938599,
      "learning_rate": 0.00017624907418620691,
      "loss": 0.9997,
      "step": 2361
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5485069751739502,
      "learning_rate": 0.00017622962367270346,
      "loss": 1.0984,
      "step": 2362
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6365689039230347,
      "learning_rate": 0.00017621016627228516,
      "loss": 1.087,
      "step": 2363
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6249110698699951,
      "learning_rate": 0.00017619070198670981,
      "loss": 1.0656,
      "step": 2364
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5620887279510498,
      "learning_rate": 0.00017617123081773591,
      "loss": 1.1772,
      "step": 2365
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5600476264953613,
      "learning_rate": 0.00017615175276712263,
      "loss": 1.0348,
      "step": 2366
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7794972062110901,
      "learning_rate": 0.00017613226783662965,
      "loss": 1.0991,
      "step": 2367
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5799472332000732,
      "learning_rate": 0.0001761127760280173,
      "loss": 0.9012,
      "step": 2368
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6447935104370117,
      "learning_rate": 0.00017609327734304658,
      "loss": 1.0947,
      "step": 2369
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6327322721481323,
      "learning_rate": 0.0001760737717834791,
      "loss": 1.0207,
      "step": 2370
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6323292851448059,
      "learning_rate": 0.00017605425935107708,
      "loss": 1.1466,
      "step": 2371
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6471980214118958,
      "learning_rate": 0.00017603474004760337,
      "loss": 1.0487,
      "step": 2372
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6297856569290161,
      "learning_rate": 0.00017601521387482135,
      "loss": 1.2354,
      "step": 2373
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6297726631164551,
      "learning_rate": 0.00017599568083449522,
      "loss": 0.926,
      "step": 2374
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.716359555721283,
      "learning_rate": 0.00017597614092838956,
      "loss": 1.0752,
      "step": 2375
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6178591251373291,
      "learning_rate": 0.00017595659415826977,
      "loss": 1.1487,
      "step": 2376
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5869417786598206,
      "learning_rate": 0.00017593704052590173,
      "loss": 1.0316,
      "step": 2377
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5557582974433899,
      "learning_rate": 0.0001759174800330521,
      "loss": 1.0543,
      "step": 2378
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5786042213439941,
      "learning_rate": 0.00017589791268148792,
      "loss": 1.0099,
      "step": 2379
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6963510513305664,
      "learning_rate": 0.0001758783384729771,
      "loss": 1.1509,
      "step": 2380
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5620383024215698,
      "learning_rate": 0.000175858757409288,
      "loss": 0.8293,
      "step": 2381
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5947554111480713,
      "learning_rate": 0.0001758391694921897,
      "loss": 1.1298,
      "step": 2382
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5493107438087463,
      "learning_rate": 0.00017581957472345182,
      "loss": 1.153,
      "step": 2383
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.664533793926239,
      "learning_rate": 0.00017579997310484467,
      "loss": 1.0192,
      "step": 2384
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.591524064540863,
      "learning_rate": 0.0001757803646381391,
      "loss": 1.1513,
      "step": 2385
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5664580464363098,
      "learning_rate": 0.00017576074932510667,
      "loss": 1.0468,
      "step": 2386
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.631363034248352,
      "learning_rate": 0.0001757411271675195,
      "loss": 1.1021,
      "step": 2387
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5752427577972412,
      "learning_rate": 0.00017572149816715032,
      "loss": 1.0026,
      "step": 2388
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.561696469783783,
      "learning_rate": 0.00017570186232577256,
      "loss": 1.0329,
      "step": 2389
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.588409423828125,
      "learning_rate": 0.0001756822196451601,
      "loss": 0.9594,
      "step": 2390
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5990773439407349,
      "learning_rate": 0.00017566257012708762,
      "loss": 1.0183,
      "step": 2391
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5430911183357239,
      "learning_rate": 0.00017564291377333035,
      "loss": 1.0256,
      "step": 2392
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6061146259307861,
      "learning_rate": 0.00017562325058566408,
      "loss": 1.0068,
      "step": 2393
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.718323290348053,
      "learning_rate": 0.00017560358056586535,
      "loss": 1.1601,
      "step": 2394
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5919448137283325,
      "learning_rate": 0.00017558390371571117,
      "loss": 1.1336,
      "step": 2395
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6243352293968201,
      "learning_rate": 0.00017556422003697919,
      "loss": 1.084,
      "step": 2396
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5345236659049988,
      "learning_rate": 0.00017554452953144782,
      "loss": 1.029,
      "step": 2397
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6053178906440735,
      "learning_rate": 0.00017552483220089593,
      "loss": 1.0874,
      "step": 2398
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6026574373245239,
      "learning_rate": 0.0001755051280471031,
      "loss": 1.1356,
      "step": 2399
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5435675382614136,
      "learning_rate": 0.0001754854170718494,
      "loss": 1.022,
      "step": 2400
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5904632806777954,
      "learning_rate": 0.0001754656992769157,
      "loss": 1.0769,
      "step": 2401
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6012750864028931,
      "learning_rate": 0.00017544597466408334,
      "loss": 1.0033,
      "step": 2402
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5695236325263977,
      "learning_rate": 0.00017542624323513436,
      "loss": 0.9283,
      "step": 2403
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7243800759315491,
      "learning_rate": 0.00017540650499185136,
      "loss": 1.2303,
      "step": 2404
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6951823830604553,
      "learning_rate": 0.00017538675993601757,
      "loss": 1.144,
      "step": 2405
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6282033920288086,
      "learning_rate": 0.00017536700806941688,
      "loss": 0.8576,
      "step": 2406
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6453385949134827,
      "learning_rate": 0.00017534724939383372,
      "loss": 1.1128,
      "step": 2407
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6555474400520325,
      "learning_rate": 0.0001753274839110532,
      "loss": 1.0861,
      "step": 2408
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5819057822227478,
      "learning_rate": 0.00017530771162286101,
      "loss": 0.9768,
      "step": 2409
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5652881264686584,
      "learning_rate": 0.00017528793253104348,
      "loss": 1.1722,
      "step": 2410
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5035662055015564,
      "learning_rate": 0.0001752681466373875,
      "loss": 0.8257,
      "step": 2411
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6228415369987488,
      "learning_rate": 0.00017524835394368065,
      "loss": 0.9524,
      "step": 2412
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5981736779212952,
      "learning_rate": 0.0001752285544517111,
      "loss": 1.1211,
      "step": 2413
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.623490035533905,
      "learning_rate": 0.00017520874816326755,
      "loss": 1.1335,
      "step": 2414
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6124420762062073,
      "learning_rate": 0.0001751889350801395,
      "loss": 1.1728,
      "step": 2415
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6511784791946411,
      "learning_rate": 0.00017516911520411684,
      "loss": 1.0441,
      "step": 2416
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6015797853469849,
      "learning_rate": 0.0001751492885369902,
      "loss": 0.908,
      "step": 2417
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6543349027633667,
      "learning_rate": 0.0001751294550805509,
      "loss": 1.1191,
      "step": 2418
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6257116198539734,
      "learning_rate": 0.00017510961483659066,
      "loss": 1.0538,
      "step": 2419
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5716103315353394,
      "learning_rate": 0.00017508976780690203,
      "loss": 1.0378,
      "step": 2420
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5667004585266113,
      "learning_rate": 0.000175069913993278,
      "loss": 1.0132,
      "step": 2421
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5886109471321106,
      "learning_rate": 0.0001750500533975123,
      "loss": 0.9482,
      "step": 2422
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6722474694252014,
      "learning_rate": 0.00017503018602139925,
      "loss": 1.3397,
      "step": 2423
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5232262015342712,
      "learning_rate": 0.00017501031186673366,
      "loss": 1.035,
      "step": 2424
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5664156079292297,
      "learning_rate": 0.00017499043093531112,
      "loss": 1.0928,
      "step": 2425
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5467848181724548,
      "learning_rate": 0.00017497054322892773,
      "loss": 1.1207,
      "step": 2426
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.561660647392273,
      "learning_rate": 0.00017495064874938025,
      "loss": 0.9222,
      "step": 2427
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.659675657749176,
      "learning_rate": 0.00017493074749846602,
      "loss": 0.7637,
      "step": 2428
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5654435753822327,
      "learning_rate": 0.000174910839477983,
      "loss": 1.1098,
      "step": 2429
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6611302495002747,
      "learning_rate": 0.0001748909246897298,
      "loss": 1.1681,
      "step": 2430
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5984782576560974,
      "learning_rate": 0.00017487100313550555,
      "loss": 1.1746,
      "step": 2431
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7086513638496399,
      "learning_rate": 0.00017485107481711012,
      "loss": 1.1975,
      "step": 2432
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.639403760433197,
      "learning_rate": 0.00017483113973634388,
      "loss": 1.2715,
      "step": 2433
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6293968558311462,
      "learning_rate": 0.00017481119789500787,
      "loss": 1.0939,
      "step": 2434
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6304031610488892,
      "learning_rate": 0.00017479124929490366,
      "loss": 1.0429,
      "step": 2435
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6312044858932495,
      "learning_rate": 0.0001747712939378336,
      "loss": 1.2176,
      "step": 2436
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7063558101654053,
      "learning_rate": 0.00017475133182560045,
      "loss": 1.1881,
      "step": 2437
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6040260791778564,
      "learning_rate": 0.00017473136296000772,
      "loss": 1.1133,
      "step": 2438
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6029963493347168,
      "learning_rate": 0.00017471138734285948,
      "loss": 1.071,
      "step": 2439
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5627874135971069,
      "learning_rate": 0.00017469140497596038,
      "loss": 0.9752,
      "step": 2440
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5305935740470886,
      "learning_rate": 0.0001746714158611158,
      "loss": 1.0148,
      "step": 2441
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5892313122749329,
      "learning_rate": 0.00017465142000013155,
      "loss": 1.0881,
      "step": 2442
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6180660724639893,
      "learning_rate": 0.00017463141739481418,
      "loss": 1.071,
      "step": 2443
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.559648334980011,
      "learning_rate": 0.0001746114080469708,
      "loss": 0.9962,
      "step": 2444
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6373436450958252,
      "learning_rate": 0.00017459139195840918,
      "loss": 1.1598,
      "step": 2445
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6689919829368591,
      "learning_rate": 0.00017457136913093761,
      "loss": 1.1186,
      "step": 2446
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.49747997522354126,
      "learning_rate": 0.00017455133956636507,
      "loss": 0.9106,
      "step": 2447
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6205312609672546,
      "learning_rate": 0.0001745313032665011,
      "loss": 0.9376,
      "step": 2448
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5468583703041077,
      "learning_rate": 0.00017451126023315588,
      "loss": 0.8509,
      "step": 2449
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.720237672328949,
      "learning_rate": 0.00017449121046814016,
      "loss": 1.0091,
      "step": 2450
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6043667793273926,
      "learning_rate": 0.0001744711539732654,
      "loss": 1.0248,
      "step": 2451
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5988819599151611,
      "learning_rate": 0.00017445109075034347,
      "loss": 1.1108,
      "step": 2452
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.62172931432724,
      "learning_rate": 0.00017443102080118704,
      "loss": 1.127,
      "step": 2453
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5728936791419983,
      "learning_rate": 0.00017441094412760935,
      "loss": 1.0426,
      "step": 2454
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5967807173728943,
      "learning_rate": 0.00017439086073142414,
      "loss": 1.2082,
      "step": 2455
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5857275724411011,
      "learning_rate": 0.00017437077061444587,
      "loss": 1.1496,
      "step": 2456
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.569225549697876,
      "learning_rate": 0.00017435067377848955,
      "loss": 0.8923,
      "step": 2457
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5777881741523743,
      "learning_rate": 0.00017433057022537085,
      "loss": 1.0297,
      "step": 2458
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6602001786231995,
      "learning_rate": 0.00017431045995690595,
      "loss": 0.9484,
      "step": 2459
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6171995997428894,
      "learning_rate": 0.00017429034297491178,
      "loss": 1.053,
      "step": 2460
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5582581758499146,
      "learning_rate": 0.00017427021928120574,
      "loss": 0.9963,
      "step": 2461
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6040249466896057,
      "learning_rate": 0.0001742500888776059,
      "loss": 1.0635,
      "step": 2462
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6399948596954346,
      "learning_rate": 0.00017422995176593093,
      "loss": 1.1903,
      "step": 2463
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6034595966339111,
      "learning_rate": 0.00017420980794800013,
      "loss": 1.1586,
      "step": 2464
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6262980103492737,
      "learning_rate": 0.00017418965742563332,
      "loss": 1.0906,
      "step": 2465
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6708945035934448,
      "learning_rate": 0.00017416950020065103,
      "loss": 0.9319,
      "step": 2466
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7182316780090332,
      "learning_rate": 0.00017414933627487436,
      "loss": 1.1271,
      "step": 2467
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5667241811752319,
      "learning_rate": 0.000174129165650125,
      "loss": 1.1542,
      "step": 2468
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6271225214004517,
      "learning_rate": 0.0001741089883282252,
      "loss": 1.0171,
      "step": 2469
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7024561166763306,
      "learning_rate": 0.00017408880431099794,
      "loss": 0.9372,
      "step": 2470
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5504549145698547,
      "learning_rate": 0.0001740686136002667,
      "loss": 1.0443,
      "step": 2471
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5320385098457336,
      "learning_rate": 0.00017404841619785558,
      "loss": 1.0304,
      "step": 2472
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6917148232460022,
      "learning_rate": 0.00017402821210558934,
      "loss": 0.9552,
      "step": 2473
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5703467726707458,
      "learning_rate": 0.00017400800132529327,
      "loss": 1.1731,
      "step": 2474
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5451155304908752,
      "learning_rate": 0.0001739877838587933,
      "loss": 1.0043,
      "step": 2475
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.653205156326294,
      "learning_rate": 0.000173967559707916,
      "loss": 1.0316,
      "step": 2476
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6300532817840576,
      "learning_rate": 0.00017394732887448847,
      "loss": 1.0411,
      "step": 2477
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5900035500526428,
      "learning_rate": 0.00017392709136033847,
      "loss": 1.117,
      "step": 2478
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6111758351325989,
      "learning_rate": 0.00017390684716729434,
      "loss": 1.1896,
      "step": 2479
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5686002969741821,
      "learning_rate": 0.000173886596297185,
      "loss": 1.1497,
      "step": 2480
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6367338299751282,
      "learning_rate": 0.00017386633875184006,
      "loss": 1.2072,
      "step": 2481
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6269874572753906,
      "learning_rate": 0.00017384607453308965,
      "loss": 1.0348,
      "step": 2482
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6493733525276184,
      "learning_rate": 0.00017382580364276451,
      "loss": 1.0231,
      "step": 2483
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7141128778457642,
      "learning_rate": 0.00017380552608269604,
      "loss": 1.1997,
      "step": 2484
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5094428658485413,
      "learning_rate": 0.0001737852418547161,
      "loss": 0.8767,
      "step": 2485
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.635170578956604,
      "learning_rate": 0.00017376495096065739,
      "loss": 1.0596,
      "step": 2486
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5259461998939514,
      "learning_rate": 0.00017374465340235298,
      "loss": 0.9905,
      "step": 2487
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.47776302695274353,
      "learning_rate": 0.00017372434918163672,
      "loss": 0.864,
      "step": 2488
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5800095200538635,
      "learning_rate": 0.0001737040383003429,
      "loss": 0.9926,
      "step": 2489
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6474664211273193,
      "learning_rate": 0.00017368372076030655,
      "loss": 1.183,
      "step": 2490
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6319321393966675,
      "learning_rate": 0.0001736633965633632,
      "loss": 1.1478,
      "step": 2491
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5537928342819214,
      "learning_rate": 0.00017364306571134909,
      "loss": 1.0313,
      "step": 2492
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6153177618980408,
      "learning_rate": 0.00017362272820610095,
      "loss": 0.9352,
      "step": 2493
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6004303097724915,
      "learning_rate": 0.00017360238404945616,
      "loss": 0.8545,
      "step": 2494
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6483585834503174,
      "learning_rate": 0.0001735820332432527,
      "loss": 1.0338,
      "step": 2495
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5554136633872986,
      "learning_rate": 0.00017356167578932915,
      "loss": 1.1218,
      "step": 2496
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.582006573677063,
      "learning_rate": 0.00017354131168952473,
      "loss": 1.0416,
      "step": 2497
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.582472562789917,
      "learning_rate": 0.00017352094094567915,
      "loss": 0.9425,
      "step": 2498
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5998402237892151,
      "learning_rate": 0.00017350056355963287,
      "loss": 0.9139,
      "step": 2499
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5740723609924316,
      "learning_rate": 0.00017348017953322683,
      "loss": 1.2384,
      "step": 2500
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5838708281517029,
      "learning_rate": 0.00017345978886830257,
      "loss": 1.1484,
      "step": 2501
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6191444396972656,
      "learning_rate": 0.00017343939156670235,
      "loss": 0.9293,
      "step": 2502
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.569363534450531,
      "learning_rate": 0.00017341898763026888,
      "loss": 0.8821,
      "step": 2503
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6086107492446899,
      "learning_rate": 0.0001733985770608456,
      "loss": 1.1533,
      "step": 2504
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6043040156364441,
      "learning_rate": 0.00017337815986027648,
      "loss": 1.116,
      "step": 2505
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5391929149627686,
      "learning_rate": 0.00017335773603040604,
      "loss": 1.0267,
      "step": 2506
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5986438989639282,
      "learning_rate": 0.00017333730557307953,
      "loss": 1.194,
      "step": 2507
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5731180310249329,
      "learning_rate": 0.00017331686849014267,
      "loss": 1.1851,
      "step": 2508
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6088848114013672,
      "learning_rate": 0.0001732964247834419,
      "loss": 1.0551,
      "step": 2509
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5705273747444153,
      "learning_rate": 0.00017327597445482415,
      "loss": 0.9899,
      "step": 2510
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7256391644477844,
      "learning_rate": 0.00017325551750613696,
      "loss": 1.2971,
      "step": 2511
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.659708559513092,
      "learning_rate": 0.00017323505393922856,
      "loss": 1.1237,
      "step": 2512
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6605255603790283,
      "learning_rate": 0.0001732145837559477,
      "loss": 1.0982,
      "step": 2513
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5549720525741577,
      "learning_rate": 0.0001731941069581437,
      "loss": 1.0215,
      "step": 2514
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5785793662071228,
      "learning_rate": 0.0001731736235476666,
      "loss": 0.9916,
      "step": 2515
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5541704297065735,
      "learning_rate": 0.0001731531335263669,
      "loss": 1.0952,
      "step": 2516
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5742295980453491,
      "learning_rate": 0.0001731326368960958,
      "loss": 1.049,
      "step": 2517
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5734418630599976,
      "learning_rate": 0.000173112133658705,
      "loss": 1.2567,
      "step": 2518
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5206436514854431,
      "learning_rate": 0.00017309162381604691,
      "loss": 0.9286,
      "step": 2519
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.624740481376648,
      "learning_rate": 0.00017307110736997444,
      "loss": 1.0974,
      "step": 2520
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8554314970970154,
      "learning_rate": 0.00017305058432234117,
      "loss": 1.1071,
      "step": 2521
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5889794230461121,
      "learning_rate": 0.00017303005467500117,
      "loss": 1.1397,
      "step": 2522
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5917510986328125,
      "learning_rate": 0.00017300951842980926,
      "loss": 1.0686,
      "step": 2523
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6076456308364868,
      "learning_rate": 0.00017298897558862072,
      "loss": 1.1758,
      "step": 2524
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5394027829170227,
      "learning_rate": 0.00017296842615329152,
      "loss": 1.0842,
      "step": 2525
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5625893473625183,
      "learning_rate": 0.00017294787012567814,
      "loss": 0.9352,
      "step": 2526
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6789762377738953,
      "learning_rate": 0.00017292730750763773,
      "loss": 1.2374,
      "step": 2527
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6913893818855286,
      "learning_rate": 0.00017290673830102802,
      "loss": 1.1266,
      "step": 2528
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5793453454971313,
      "learning_rate": 0.0001728861625077073,
      "loss": 0.9519,
      "step": 2529
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5948161482810974,
      "learning_rate": 0.00017286558012953444,
      "loss": 1.0246,
      "step": 2530
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5573294758796692,
      "learning_rate": 0.00017284499116836902,
      "loss": 1.0322,
      "step": 2531
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6499856114387512,
      "learning_rate": 0.00017282439562607107,
      "loss": 1.0345,
      "step": 2532
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5683536529541016,
      "learning_rate": 0.0001728037935045013,
      "loss": 0.9092,
      "step": 2533
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6661322712898254,
      "learning_rate": 0.000172783184805521,
      "loss": 1.0459,
      "step": 2534
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5732278823852539,
      "learning_rate": 0.0001727625695309921,
      "loss": 0.981,
      "step": 2535
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6493289470672607,
      "learning_rate": 0.00017274194768277697,
      "loss": 1.1291,
      "step": 2536
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6122599840164185,
      "learning_rate": 0.00017272131926273875,
      "loss": 1.0597,
      "step": 2537
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6348492503166199,
      "learning_rate": 0.00017270068427274109,
      "loss": 1.1,
      "step": 2538
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5080891847610474,
      "learning_rate": 0.00017268004271464823,
      "loss": 0.9227,
      "step": 2539
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5522621273994446,
      "learning_rate": 0.00017265939459032504,
      "loss": 0.9829,
      "step": 2540
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6143836975097656,
      "learning_rate": 0.00017263873990163692,
      "loss": 0.9779,
      "step": 2541
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5581562519073486,
      "learning_rate": 0.00017261807865044993,
      "loss": 0.9664,
      "step": 2542
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6005426049232483,
      "learning_rate": 0.0001725974108386307,
      "loss": 1.0958,
      "step": 2543
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6451091766357422,
      "learning_rate": 0.00017257673646804643,
      "loss": 0.9855,
      "step": 2544
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6124886274337769,
      "learning_rate": 0.00017255605554056496,
      "loss": 1.02,
      "step": 2545
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6112679243087769,
      "learning_rate": 0.00017253536805805474,
      "loss": 0.9499,
      "step": 2546
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5976582765579224,
      "learning_rate": 0.00017251467402238464,
      "loss": 0.8698,
      "step": 2547
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6313769817352295,
      "learning_rate": 0.00017249397343542433,
      "loss": 1.163,
      "step": 2548
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6301968097686768,
      "learning_rate": 0.000172473266299044,
      "loss": 1.0018,
      "step": 2549
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5670313835144043,
      "learning_rate": 0.0001724525526151144,
      "loss": 1.1514,
      "step": 2550
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5379245281219482,
      "learning_rate": 0.0001724318323855069,
      "loss": 0.9412,
      "step": 2551
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5700402855873108,
      "learning_rate": 0.00017241110561209342,
      "loss": 1.2688,
      "step": 2552
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6314685940742493,
      "learning_rate": 0.00017239037229674654,
      "loss": 1.1394,
      "step": 2553
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6362370252609253,
      "learning_rate": 0.00017236963244133944,
      "loss": 1.147,
      "step": 2554
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6027795672416687,
      "learning_rate": 0.00017234888604774574,
      "loss": 1.0329,
      "step": 2555
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6116631031036377,
      "learning_rate": 0.00017232813311783988,
      "loss": 0.9994,
      "step": 2556
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6491808891296387,
      "learning_rate": 0.00017230737365349672,
      "loss": 0.9094,
      "step": 2557
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5615575909614563,
      "learning_rate": 0.0001722866076565917,
      "loss": 1.1796,
      "step": 2558
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5690613389015198,
      "learning_rate": 0.000172265835129001,
      "loss": 1.0572,
      "step": 2559
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5822938680648804,
      "learning_rate": 0.00017224505607260128,
      "loss": 1.1249,
      "step": 2560
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5287598967552185,
      "learning_rate": 0.00017222427048926977,
      "loss": 0.9504,
      "step": 2561
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5965141654014587,
      "learning_rate": 0.00017220347838088436,
      "loss": 1.0806,
      "step": 2562
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6563447713851929,
      "learning_rate": 0.00017218267974932348,
      "loss": 1.0252,
      "step": 2563
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5706745386123657,
      "learning_rate": 0.0001721618745964662,
      "loss": 1.1904,
      "step": 2564
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6056011915206909,
      "learning_rate": 0.00017214106292419212,
      "loss": 1.2825,
      "step": 2565
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6403053402900696,
      "learning_rate": 0.00017212024473438147,
      "loss": 0.9715,
      "step": 2566
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6166976690292358,
      "learning_rate": 0.00017209942002891504,
      "loss": 1.0931,
      "step": 2567
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6298193335533142,
      "learning_rate": 0.00017207858880967426,
      "loss": 1.1611,
      "step": 2568
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5657613277435303,
      "learning_rate": 0.0001720577510785411,
      "loss": 1.0368,
      "step": 2569
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8520029187202454,
      "learning_rate": 0.00017203690683739807,
      "loss": 1.2765,
      "step": 2570
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5342356562614441,
      "learning_rate": 0.0001720160560881284,
      "loss": 1.1253,
      "step": 2571
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6659888029098511,
      "learning_rate": 0.00017199519883261583,
      "loss": 1.0596,
      "step": 2572
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6437025666236877,
      "learning_rate": 0.00017197433507274465,
      "loss": 1.2124,
      "step": 2573
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6187822818756104,
      "learning_rate": 0.00017195346481039986,
      "loss": 1.0649,
      "step": 2574
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5546347498893738,
      "learning_rate": 0.0001719325880474669,
      "loss": 1.1158,
      "step": 2575
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6503075361251831,
      "learning_rate": 0.00017191170478583187,
      "loss": 1.1343,
      "step": 2576
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6349958777427673,
      "learning_rate": 0.0001718908150273815,
      "loss": 0.935,
      "step": 2577
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6106913089752197,
      "learning_rate": 0.00017186991877400304,
      "loss": 1.1899,
      "step": 2578
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6525137424468994,
      "learning_rate": 0.00017184901602758433,
      "loss": 1.0135,
      "step": 2579
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5244576334953308,
      "learning_rate": 0.00017182810679001385,
      "loss": 1.0043,
      "step": 2580
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5864934921264648,
      "learning_rate": 0.0001718071910631806,
      "loss": 1.0766,
      "step": 2581
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6121466159820557,
      "learning_rate": 0.0001717862688489742,
      "loss": 0.9471,
      "step": 2582
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5585313439369202,
      "learning_rate": 0.00017176534014928487,
      "loss": 1.0705,
      "step": 2583
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5822098851203918,
      "learning_rate": 0.0001717444049660034,
      "loss": 1.0575,
      "step": 2584
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5953527092933655,
      "learning_rate": 0.00017172346330102115,
      "loss": 0.9999,
      "step": 2585
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5756900906562805,
      "learning_rate": 0.00017170251515623007,
      "loss": 1.0358,
      "step": 2586
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.714945912361145,
      "learning_rate": 0.00017168156053352273,
      "loss": 1.2321,
      "step": 2587
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6626083850860596,
      "learning_rate": 0.0001716605994347923,
      "loss": 1.1833,
      "step": 2588
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5595118999481201,
      "learning_rate": 0.00017163963186193242,
      "loss": 0.9832,
      "step": 2589
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5993494391441345,
      "learning_rate": 0.00017161865781683741,
      "loss": 1.1693,
      "step": 2590
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5663042068481445,
      "learning_rate": 0.0001715976773014022,
      "loss": 1.1872,
      "step": 2591
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5583269000053406,
      "learning_rate": 0.00017157669031752222,
      "loss": 0.9559,
      "step": 2592
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6719366312026978,
      "learning_rate": 0.0001715556968670935,
      "loss": 0.9072,
      "step": 2593
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5847353935241699,
      "learning_rate": 0.00017153469695201277,
      "loss": 0.9896,
      "step": 2594
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5907499194145203,
      "learning_rate": 0.00017151369057417717,
      "loss": 1.1507,
      "step": 2595
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5951755046844482,
      "learning_rate": 0.00017149267773548455,
      "loss": 1.1303,
      "step": 2596
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5762290954589844,
      "learning_rate": 0.00017147165843783325,
      "loss": 1.0785,
      "step": 2597
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5002837181091309,
      "learning_rate": 0.0001714506326831223,
      "loss": 0.8424,
      "step": 2598
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5358802676200867,
      "learning_rate": 0.00017142960047325127,
      "loss": 0.9938,
      "step": 2599
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6255794167518616,
      "learning_rate": 0.00017140856181012027,
      "loss": 1.0858,
      "step": 2600
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7021377086639404,
      "learning_rate": 0.00017138751669562997,
      "loss": 1.1827,
      "step": 2601
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6269333958625793,
      "learning_rate": 0.00017136646513168179,
      "loss": 1.1467,
      "step": 2602
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5892650485038757,
      "learning_rate": 0.00017134540712017752,
      "loss": 0.9954,
      "step": 2603
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5547927021980286,
      "learning_rate": 0.0001713243426630197,
      "loss": 1.1428,
      "step": 2604
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5758942365646362,
      "learning_rate": 0.00017130327176211133,
      "loss": 1.0432,
      "step": 2605
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6165540218353271,
      "learning_rate": 0.0001712821944193561,
      "loss": 1.1172,
      "step": 2606
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6256709694862366,
      "learning_rate": 0.0001712611106366582,
      "loss": 0.9987,
      "step": 2607
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6734545230865479,
      "learning_rate": 0.00017124002041592245,
      "loss": 0.9803,
      "step": 2608
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5715333819389343,
      "learning_rate": 0.0001712189237590542,
      "loss": 1.1031,
      "step": 2609
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6795665621757507,
      "learning_rate": 0.0001711978206679594,
      "loss": 1.2371,
      "step": 2610
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6391113996505737,
      "learning_rate": 0.00017117671114454463,
      "loss": 1.2347,
      "step": 2611
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5957304835319519,
      "learning_rate": 0.00017115559519071704,
      "loss": 1.138,
      "step": 2612
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6582696437835693,
      "learning_rate": 0.0001711344728083843,
      "loss": 1.2427,
      "step": 2613
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.610765814781189,
      "learning_rate": 0.00017111334399945468,
      "loss": 1.0626,
      "step": 2614
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6205570101737976,
      "learning_rate": 0.0001710922087658371,
      "loss": 0.9739,
      "step": 2615
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5730193853378296,
      "learning_rate": 0.00017107106710944096,
      "loss": 1.0047,
      "step": 2616
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5567137598991394,
      "learning_rate": 0.00017104991903217634,
      "loss": 1.2085,
      "step": 2617
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5671441555023193,
      "learning_rate": 0.0001710287645359538,
      "loss": 1.1132,
      "step": 2618
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5712023973464966,
      "learning_rate": 0.00017100760362268454,
      "loss": 1.052,
      "step": 2619
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6445497274398804,
      "learning_rate": 0.00017098643629428036,
      "loss": 0.9957,
      "step": 2620
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.959863007068634,
      "eval_runtime": 17.8087,
      "eval_samples_per_second": 153.24,
      "eval_steps_per_second": 38.352,
      "step": 2620
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5775024890899658,
      "learning_rate": 0.00017096526255265356,
      "loss": 1.0435,
      "step": 2621
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.603359580039978,
      "learning_rate": 0.0001709440823997171,
      "loss": 1.1612,
      "step": 2622
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5576832890510559,
      "learning_rate": 0.00017092289583738447,
      "loss": 0.8875,
      "step": 2623
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5404002070426941,
      "learning_rate": 0.00017090170286756978,
      "loss": 1.0764,
      "step": 2624
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6048089861869812,
      "learning_rate": 0.00017088050349218768,
      "loss": 0.9467,
      "step": 2625
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6076110005378723,
      "learning_rate": 0.00017085929771315338,
      "loss": 1.0436,
      "step": 2626
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6579375863075256,
      "learning_rate": 0.00017083808553238278,
      "loss": 1.0591,
      "step": 2627
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.515089213848114,
      "learning_rate": 0.00017081686695179221,
      "loss": 0.9998,
      "step": 2628
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6199119091033936,
      "learning_rate": 0.0001707956419732987,
      "loss": 0.9776,
      "step": 2629
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6219343543052673,
      "learning_rate": 0.00017077441059881978,
      "loss": 1.0895,
      "step": 2630
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7458560466766357,
      "learning_rate": 0.00017075317283027355,
      "loss": 1.0071,
      "step": 2631
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6140450239181519,
      "learning_rate": 0.00017073192866957882,
      "loss": 1.0347,
      "step": 2632
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6091210246086121,
      "learning_rate": 0.00017071067811865476,
      "loss": 0.9943,
      "step": 2633
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6610286235809326,
      "learning_rate": 0.0001706894211794213,
      "loss": 1.2903,
      "step": 2634
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5978135466575623,
      "learning_rate": 0.00017066815785379892,
      "loss": 1.0517,
      "step": 2635
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6416430473327637,
      "learning_rate": 0.00017064688814370858,
      "loss": 1.2506,
      "step": 2636
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6144441962242126,
      "learning_rate": 0.0001706256120510719,
      "loss": 1.0151,
      "step": 2637
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6584609150886536,
      "learning_rate": 0.00017060432957781104,
      "loss": 1.0537,
      "step": 2638
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6167966723442078,
      "learning_rate": 0.00017058304072584878,
      "loss": 1.0422,
      "step": 2639
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5490273237228394,
      "learning_rate": 0.00017056174549710842,
      "loss": 1.1807,
      "step": 2640
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5920153856277466,
      "learning_rate": 0.00017054044389351383,
      "loss": 1.0671,
      "step": 2641
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6045279502868652,
      "learning_rate": 0.00017051913591698958,
      "loss": 1.0548,
      "step": 2642
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6131229400634766,
      "learning_rate": 0.0001704978215694607,
      "loss": 1.0219,
      "step": 2643
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.578783392906189,
      "learning_rate": 0.00017047650085285276,
      "loss": 1.1181,
      "step": 2644
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5949702858924866,
      "learning_rate": 0.00017045517376909205,
      "loss": 1.0815,
      "step": 2645
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6813076734542847,
      "learning_rate": 0.00017043384032010522,
      "loss": 1.2671,
      "step": 2646
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6607581377029419,
      "learning_rate": 0.00017041250050781978,
      "loss": 0.9366,
      "step": 2647
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6341320276260376,
      "learning_rate": 0.0001703911543341636,
      "loss": 1.2367,
      "step": 2648
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5848129987716675,
      "learning_rate": 0.00017036980180106515,
      "loss": 1.0295,
      "step": 2649
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5940001010894775,
      "learning_rate": 0.00017034844291045356,
      "loss": 1.0005,
      "step": 2650
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5980600118637085,
      "learning_rate": 0.00017032707766425844,
      "loss": 1.0023,
      "step": 2651
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6173316240310669,
      "learning_rate": 0.00017030570606441005,
      "loss": 1.1167,
      "step": 2652
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5713648796081543,
      "learning_rate": 0.00017028432811283918,
      "loss": 0.9816,
      "step": 2653
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.690399169921875,
      "learning_rate": 0.00017026294381147726,
      "loss": 1.2074,
      "step": 2654
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6921668648719788,
      "learning_rate": 0.00017024155316225614,
      "loss": 1.1248,
      "step": 2655
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5611989498138428,
      "learning_rate": 0.0001702201561671084,
      "loss": 1.0523,
      "step": 2656
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6552536487579346,
      "learning_rate": 0.0001701987528279672,
      "loss": 1.0417,
      "step": 2657
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.5799632668495178,
      "learning_rate": 0.0001701773431467661,
      "loss": 1.0133,
      "step": 2658
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.5619133710861206,
      "learning_rate": 0.0001701559271254394,
      "loss": 1.0702,
      "step": 2659
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6272364854812622,
      "learning_rate": 0.00017013450476592189,
      "loss": 1.029,
      "step": 2660
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6038535237312317,
      "learning_rate": 0.000170113076070149,
      "loss": 1.0668,
      "step": 2661
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.5458052158355713,
      "learning_rate": 0.00017009164104005663,
      "loss": 1.0212,
      "step": 2662
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.5674341320991516,
      "learning_rate": 0.00017007019967758138,
      "loss": 1.0305,
      "step": 2663
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6523269414901733,
      "learning_rate": 0.00017004875198466031,
      "loss": 0.983,
      "step": 2664
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6157127022743225,
      "learning_rate": 0.00017002729796323113,
      "loss": 0.8491,
      "step": 2665
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.581650972366333,
      "learning_rate": 0.00017000583761523204,
      "loss": 1.0964,
      "step": 2666
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.5540935397148132,
      "learning_rate": 0.00016998437094260194,
      "loss": 1.0273,
      "step": 2667
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.565251350402832,
      "learning_rate": 0.00016996289794728013,
      "loss": 1.0797,
      "step": 2668
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5763837099075317,
      "learning_rate": 0.00016994141863120666,
      "loss": 1.0146,
      "step": 2669
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5568943023681641,
      "learning_rate": 0.000169919932996322,
      "loss": 0.8508,
      "step": 2670
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.639254629611969,
      "learning_rate": 0.00016989844104456727,
      "loss": 1.1043,
      "step": 2671
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5605832934379578,
      "learning_rate": 0.00016987694277788417,
      "loss": 0.9942,
      "step": 2672
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6005716323852539,
      "learning_rate": 0.00016985543819821492,
      "loss": 0.9229,
      "step": 2673
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5722012519836426,
      "learning_rate": 0.0001698339273075024,
      "loss": 1.0464,
      "step": 2674
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5585935711860657,
      "learning_rate": 0.0001698124101076899,
      "loss": 0.754,
      "step": 2675
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6257965564727783,
      "learning_rate": 0.0001697908866007214,
      "loss": 1.1917,
      "step": 2676
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5758666396141052,
      "learning_rate": 0.00016976935678854148,
      "loss": 1.0265,
      "step": 2677
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6103315949440002,
      "learning_rate": 0.00016974782067309522,
      "loss": 1.1078,
      "step": 2678
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6202539205551147,
      "learning_rate": 0.00016972627825632828,
      "loss": 1.119,
      "step": 2679
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6085442900657654,
      "learning_rate": 0.00016970472954018687,
      "loss": 0.987,
      "step": 2680
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6156657338142395,
      "learning_rate": 0.0001696831745266178,
      "loss": 1.0947,
      "step": 2681
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5341770648956299,
      "learning_rate": 0.00016966161321756848,
      "loss": 0.7939,
      "step": 2682
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5123429894447327,
      "learning_rate": 0.0001696400456149868,
      "loss": 0.9622,
      "step": 2683
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6400977969169617,
      "learning_rate": 0.00016961847172082135,
      "loss": 0.6955,
      "step": 2684
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9171468615531921,
      "learning_rate": 0.0001695968915370211,
      "loss": 0.8986,
      "step": 2685
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6452581286430359,
      "learning_rate": 0.00016957530506553578,
      "loss": 0.8457,
      "step": 2686
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6027610301971436,
      "learning_rate": 0.0001695537123083156,
      "loss": 0.9135,
      "step": 2687
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5201352834701538,
      "learning_rate": 0.00016953211326731132,
      "loss": 0.8204,
      "step": 2688
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.611875593662262,
      "learning_rate": 0.00016951050794447426,
      "loss": 0.9314,
      "step": 2689
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7601920962333679,
      "learning_rate": 0.0001694888963417564,
      "loss": 0.8574,
      "step": 2690
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6485564112663269,
      "learning_rate": 0.0001694672784611102,
      "loss": 1.0105,
      "step": 2691
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7118054032325745,
      "learning_rate": 0.00016944565430448874,
      "loss": 1.0206,
      "step": 2692
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.624322772026062,
      "learning_rate": 0.00016942402387384558,
      "loss": 0.8937,
      "step": 2693
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6083493232727051,
      "learning_rate": 0.00016940238717113494,
      "loss": 0.9125,
      "step": 2694
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6026355624198914,
      "learning_rate": 0.0001693807441983116,
      "loss": 0.905,
      "step": 2695
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7009247541427612,
      "learning_rate": 0.00016935909495733082,
      "loss": 1.2486,
      "step": 2696
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5891932845115662,
      "learning_rate": 0.00016933743945014856,
      "loss": 0.8315,
      "step": 2697
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.656340479850769,
      "learning_rate": 0.0001693157776787212,
      "loss": 1.048,
      "step": 2698
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6057400107383728,
      "learning_rate": 0.0001692941096450058,
      "loss": 0.973,
      "step": 2699
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.680121123790741,
      "learning_rate": 0.00016927243535095997,
      "loss": 1.0588,
      "step": 2700
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6800230741500854,
      "learning_rate": 0.00016925075479854175,
      "loss": 0.9017,
      "step": 2701
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.589103102684021,
      "learning_rate": 0.00016922906798971,
      "loss": 0.8216,
      "step": 2702
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6343634724617004,
      "learning_rate": 0.0001692073749264239,
      "loss": 0.883,
      "step": 2703
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6688684821128845,
      "learning_rate": 0.00016918567561064336,
      "loss": 0.8563,
      "step": 2704
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6180203557014465,
      "learning_rate": 0.00016916397004432872,
      "loss": 1.0055,
      "step": 2705
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6281496286392212,
      "learning_rate": 0.00016914225822944098,
      "loss": 0.9944,
      "step": 2706
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7170965671539307,
      "learning_rate": 0.00016912054016794174,
      "loss": 1.145,
      "step": 2707
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6044749021530151,
      "learning_rate": 0.00016909881586179304,
      "loss": 0.9927,
      "step": 2708
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6250945329666138,
      "learning_rate": 0.00016907708531295755,
      "loss": 0.8074,
      "step": 2709
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7008484601974487,
      "learning_rate": 0.0001690553485233985,
      "loss": 0.9173,
      "step": 2710
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7483898997306824,
      "learning_rate": 0.0001690336054950797,
      "loss": 0.8804,
      "step": 2711
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6421265006065369,
      "learning_rate": 0.00016901185622996554,
      "loss": 1.0949,
      "step": 2712
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6270898580551147,
      "learning_rate": 0.00016899010073002092,
      "loss": 0.8987,
      "step": 2713
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6082978844642639,
      "learning_rate": 0.0001689683389972113,
      "loss": 0.7974,
      "step": 2714
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.5921205878257751,
      "learning_rate": 0.00016894657103350274,
      "loss": 1.0373,
      "step": 2715
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6575934886932373,
      "learning_rate": 0.00016892479684086186,
      "loss": 0.944,
      "step": 2716
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6679452657699585,
      "learning_rate": 0.00016890301642125588,
      "loss": 0.9262,
      "step": 2717
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7515625357627869,
      "learning_rate": 0.0001688812297766525,
      "loss": 0.944,
      "step": 2718
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6797657608985901,
      "learning_rate": 0.00016885943690901994,
      "loss": 0.8558,
      "step": 2719
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6033437848091125,
      "learning_rate": 0.00016883763782032718,
      "loss": 0.9379,
      "step": 2720
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6078616976737976,
      "learning_rate": 0.0001688158325125436,
      "loss": 0.7629,
      "step": 2721
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6192131638526917,
      "learning_rate": 0.00016879402098763917,
      "loss": 1.0707,
      "step": 2722
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6294496059417725,
      "learning_rate": 0.00016877220324758448,
      "loss": 0.8759,
      "step": 2723
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6831909418106079,
      "learning_rate": 0.0001687503792943506,
      "loss": 0.9751,
      "step": 2724
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6219105124473572,
      "learning_rate": 0.00016872854912990923,
      "loss": 0.8283,
      "step": 2725
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6364472508430481,
      "learning_rate": 0.00016870671275623256,
      "loss": 1.0008,
      "step": 2726
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6604315042495728,
      "learning_rate": 0.00016868487017529342,
      "loss": 1.0695,
      "step": 2727
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6434652805328369,
      "learning_rate": 0.00016866302138906516,
      "loss": 0.8471,
      "step": 2728
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6357761025428772,
      "learning_rate": 0.00016864116639952168,
      "loss": 0.9723,
      "step": 2729
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7642642259597778,
      "learning_rate": 0.00016861930520863748,
      "loss": 1.154,
      "step": 2730
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6624659895896912,
      "learning_rate": 0.0001685974378183875,
      "loss": 0.8315,
      "step": 2731
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6619630455970764,
      "learning_rate": 0.00016857556423074748,
      "loss": 0.8667,
      "step": 2732
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6592692136764526,
      "learning_rate": 0.0001685536844476935,
      "loss": 1.0389,
      "step": 2733
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6404480338096619,
      "learning_rate": 0.0001685317984712023,
      "loss": 1.0442,
      "step": 2734
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.5911348462104797,
      "learning_rate": 0.0001685099063032511,
      "loss": 0.7707,
      "step": 2735
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6581184267997742,
      "learning_rate": 0.00016848800794581775,
      "loss": 0.9005,
      "step": 2736
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6644151210784912,
      "learning_rate": 0.0001684661034008807,
      "loss": 0.9491,
      "step": 2737
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6669042706489563,
      "learning_rate": 0.00016844419267041886,
      "loss": 0.9936,
      "step": 2738
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6686713099479675,
      "learning_rate": 0.00016842227575641173,
      "loss": 1.0133,
      "step": 2739
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.639211118221283,
      "learning_rate": 0.0001684003526608394,
      "loss": 0.8211,
      "step": 2740
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6182137727737427,
      "learning_rate": 0.00016837842338568245,
      "loss": 0.9588,
      "step": 2741
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7168389558792114,
      "learning_rate": 0.00016835648793292216,
      "loss": 0.7844,
      "step": 2742
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.5751428008079529,
      "learning_rate": 0.00016833454630454017,
      "loss": 0.8838,
      "step": 2743
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6621695756912231,
      "learning_rate": 0.00016831259850251886,
      "loss": 1.0437,
      "step": 2744
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.646742045879364,
      "learning_rate": 0.00016829064452884108,
      "loss": 0.9284,
      "step": 2745
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6294827461242676,
      "learning_rate": 0.00016826868438549018,
      "loss": 0.9306,
      "step": 2746
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6392683386802673,
      "learning_rate": 0.0001682467180744502,
      "loss": 0.9498,
      "step": 2747
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6557241082191467,
      "learning_rate": 0.00016822474559770564,
      "loss": 1.1063,
      "step": 2748
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6353508830070496,
      "learning_rate": 0.00016820276695724164,
      "loss": 0.8402,
      "step": 2749
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6505456566810608,
      "learning_rate": 0.0001681807821550438,
      "loss": 0.9349,
      "step": 2750
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7468292713165283,
      "learning_rate": 0.0001681587911930983,
      "loss": 1.1376,
      "step": 2751
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.608337938785553,
      "learning_rate": 0.00016813679407339197,
      "loss": 1.0272,
      "step": 2752
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6507999897003174,
      "learning_rate": 0.00016811479079791204,
      "loss": 1.1247,
      "step": 2753
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6512243151664734,
      "learning_rate": 0.00016809278136864646,
      "loss": 0.901,
      "step": 2754
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.5723342299461365,
      "learning_rate": 0.0001680707657875836,
      "loss": 0.8635,
      "step": 2755
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6601933240890503,
      "learning_rate": 0.0001680487440567125,
      "loss": 0.876,
      "step": 2756
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6232926845550537,
      "learning_rate": 0.00016802671617802264,
      "loss": 0.9157,
      "step": 2757
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6270859241485596,
      "learning_rate": 0.00016800468215350412,
      "loss": 0.9454,
      "step": 2758
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6160213351249695,
      "learning_rate": 0.00016798264198514767,
      "loss": 0.8243,
      "step": 2759
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6409580111503601,
      "learning_rate": 0.00016796059567494439,
      "loss": 1.014,
      "step": 2760
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7464109063148499,
      "learning_rate": 0.00016793854322488607,
      "loss": 1.0413,
      "step": 2761
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.58450847864151,
      "learning_rate": 0.00016791648463696505,
      "loss": 0.7682,
      "step": 2762
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6073700189590454,
      "learning_rate": 0.0001678944199131742,
      "loss": 0.8193,
      "step": 2763
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6749382615089417,
      "learning_rate": 0.00016787234905550692,
      "loss": 0.9554,
      "step": 2764
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6255000829696655,
      "learning_rate": 0.0001678502720659572,
      "loss": 0.8979,
      "step": 2765
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.625436544418335,
      "learning_rate": 0.00016782818894651956,
      "loss": 0.9665,
      "step": 2766
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6645047664642334,
      "learning_rate": 0.0001678060996991891,
      "loss": 0.993,
      "step": 2767
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7271724939346313,
      "learning_rate": 0.00016778400432596144,
      "loss": 0.9922,
      "step": 2768
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7113057374954224,
      "learning_rate": 0.00016776190282883277,
      "loss": 1.029,
      "step": 2769
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.650078535079956,
      "learning_rate": 0.00016773979520979987,
      "loss": 0.9139,
      "step": 2770
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6887482404708862,
      "learning_rate": 0.00016771768147086,
      "loss": 1.0467,
      "step": 2771
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.675455629825592,
      "learning_rate": 0.00016769556161401103,
      "loss": 0.9832,
      "step": 2772
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.690849244594574,
      "learning_rate": 0.00016767343564125136,
      "loss": 0.9428,
      "step": 2773
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7624968886375427,
      "learning_rate": 0.00016765130355457996,
      "loss": 1.0352,
      "step": 2774
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6946026682853699,
      "learning_rate": 0.00016762916535599631,
      "loss": 0.9284,
      "step": 2775
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6219006776809692,
      "learning_rate": 0.00016760702104750045,
      "loss": 0.8471,
      "step": 2776
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6577418446540833,
      "learning_rate": 0.0001675848706310931,
      "loss": 1.0124,
      "step": 2777
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6933895945549011,
      "learning_rate": 0.00016756271410877532,
      "loss": 0.8792,
      "step": 2778
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6292385458946228,
      "learning_rate": 0.00016754055148254885,
      "loss": 0.936,
      "step": 2779
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6794418692588806,
      "learning_rate": 0.00016751838275441596,
      "loss": 0.9986,
      "step": 2780
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6378180980682373,
      "learning_rate": 0.00016749620792637948,
      "loss": 0.942,
      "step": 2781
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6403058767318726,
      "learning_rate": 0.0001674740270004428,
      "loss": 0.844,
      "step": 2782
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7571424841880798,
      "learning_rate": 0.0001674518399786098,
      "loss": 0.9293,
      "step": 2783
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6841951012611389,
      "learning_rate": 0.00016742964686288495,
      "loss": 0.7867,
      "step": 2784
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7075207233428955,
      "learning_rate": 0.00016740744765527334,
      "loss": 0.9051,
      "step": 2785
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6528226733207703,
      "learning_rate": 0.00016738524235778045,
      "loss": 0.8037,
      "step": 2786
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6237984895706177,
      "learning_rate": 0.00016736303097241245,
      "loss": 0.9536,
      "step": 2787
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6472983360290527,
      "learning_rate": 0.00016734081350117604,
      "loss": 0.8897,
      "step": 2788
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6402902007102966,
      "learning_rate": 0.00016731858994607838,
      "loss": 0.8731,
      "step": 2789
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7781495451927185,
      "learning_rate": 0.0001672963603091273,
      "loss": 1.0459,
      "step": 2790
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7584108710289001,
      "learning_rate": 0.0001672741245923311,
      "loss": 1.1999,
      "step": 2791
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6754958629608154,
      "learning_rate": 0.00016725188279769865,
      "loss": 0.8714,
      "step": 2792
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6159543395042419,
      "learning_rate": 0.00016722963492723936,
      "loss": 0.9094,
      "step": 2793
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7796074748039246,
      "learning_rate": 0.0001672073809829632,
      "loss": 1.037,
      "step": 2794
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6288584470748901,
      "learning_rate": 0.0001671851209668807,
      "loss": 0.8631,
      "step": 2795
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6400723457336426,
      "learning_rate": 0.00016716285488100292,
      "loss": 0.8373,
      "step": 2796
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6898773312568665,
      "learning_rate": 0.00016714058272734148,
      "loss": 0.9886,
      "step": 2797
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6610246300697327,
      "learning_rate": 0.0001671183045079085,
      "loss": 0.9466,
      "step": 2798
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7253816723823547,
      "learning_rate": 0.0001670960202247168,
      "loss": 1.0565,
      "step": 2799
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6904633641242981,
      "learning_rate": 0.00016707372987977953,
      "loss": 0.9772,
      "step": 2800
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7199797630310059,
      "learning_rate": 0.00016705143347511055,
      "loss": 1.0893,
      "step": 2801
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.650538980960846,
      "learning_rate": 0.0001670291310127242,
      "loss": 0.9298,
      "step": 2802
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6637884378433228,
      "learning_rate": 0.00016700682249463534,
      "loss": 1.1245,
      "step": 2803
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7162839770317078,
      "learning_rate": 0.00016698450792285952,
      "loss": 0.9432,
      "step": 2804
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6651907563209534,
      "learning_rate": 0.00016696218729941265,
      "loss": 0.7817,
      "step": 2805
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6402869820594788,
      "learning_rate": 0.00016693986062631127,
      "loss": 0.9302,
      "step": 2806
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6910731196403503,
      "learning_rate": 0.00016691752790557254,
      "loss": 0.9083,
      "step": 2807
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6888274550437927,
      "learning_rate": 0.000166895189139214,
      "loss": 1.0216,
      "step": 2808
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7038168907165527,
      "learning_rate": 0.00016687284432925393,
      "loss": 0.9906,
      "step": 2809
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7101284265518188,
      "learning_rate": 0.000166850493477711,
      "loss": 0.8463,
      "step": 2810
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6386854648590088,
      "learning_rate": 0.00016682813658660446,
      "loss": 1.073,
      "step": 2811
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6103797554969788,
      "learning_rate": 0.00016680577365795415,
      "loss": 0.9996,
      "step": 2812
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6694577932357788,
      "learning_rate": 0.00016678340469378049,
      "loss": 0.9692,
      "step": 2813
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6260998845100403,
      "learning_rate": 0.00016676102969610432,
      "loss": 0.9769,
      "step": 2814
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6177822351455688,
      "learning_rate": 0.00016673864866694716,
      "loss": 0.8814,
      "step": 2815
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7221593260765076,
      "learning_rate": 0.00016671626160833089,
      "loss": 1.1002,
      "step": 2816
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7593296766281128,
      "learning_rate": 0.00016669386852227817,
      "loss": 0.7499,
      "step": 2817
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.676494836807251,
      "learning_rate": 0.00016667146941081205,
      "loss": 0.942,
      "step": 2818
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.693713903427124,
      "learning_rate": 0.00016664906427595616,
      "loss": 0.992,
      "step": 2819
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7619845867156982,
      "learning_rate": 0.00016662665311973466,
      "loss": 1.044,
      "step": 2820
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6537740230560303,
      "learning_rate": 0.00016660423594417231,
      "loss": 0.988,
      "step": 2821
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7443823218345642,
      "learning_rate": 0.00016658181275129436,
      "loss": 1.0774,
      "step": 2822
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7338739633560181,
      "learning_rate": 0.00016655938354312658,
      "loss": 1.006,
      "step": 2823
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7927771806716919,
      "learning_rate": 0.0001665369483216954,
      "loss": 1.2174,
      "step": 2824
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6222347021102905,
      "learning_rate": 0.00016651450708902765,
      "loss": 0.8883,
      "step": 2825
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6589560508728027,
      "learning_rate": 0.00016649205984715077,
      "loss": 0.9508,
      "step": 2826
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6752550601959229,
      "learning_rate": 0.00016646960659809276,
      "loss": 0.9706,
      "step": 2827
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6681761741638184,
      "learning_rate": 0.00016644714734388217,
      "loss": 0.9949,
      "step": 2828
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6704165935516357,
      "learning_rate": 0.00016642468208654805,
      "loss": 1.1179,
      "step": 2829
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7063694596290588,
      "learning_rate": 0.00016640221082811997,
      "loss": 0.9225,
      "step": 2830
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7499638199806213,
      "learning_rate": 0.00016637973357062812,
      "loss": 0.8692,
      "step": 2831
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6111122965812683,
      "learning_rate": 0.0001663572503161032,
      "loss": 0.9102,
      "step": 2832
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6734527945518494,
      "learning_rate": 0.0001663347610665764,
      "loss": 0.9193,
      "step": 2833
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7178148031234741,
      "learning_rate": 0.00016631226582407952,
      "loss": 1.1397,
      "step": 2834
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7453985810279846,
      "learning_rate": 0.0001662897645906449,
      "loss": 0.7565,
      "step": 2835
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7964417338371277,
      "learning_rate": 0.00016626725736830538,
      "loss": 0.9191,
      "step": 2836
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6423993706703186,
      "learning_rate": 0.00016624474415909436,
      "loss": 0.8625,
      "step": 2837
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6401289105415344,
      "learning_rate": 0.00016622222496504577,
      "loss": 0.7904,
      "step": 2838
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6781225204467773,
      "learning_rate": 0.00016619969978819415,
      "loss": 0.8491,
      "step": 2839
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.672544538974762,
      "learning_rate": 0.00016617716863057442,
      "loss": 1.0306,
      "step": 2840
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6570728421211243,
      "learning_rate": 0.0001661546314942222,
      "loss": 0.8669,
      "step": 2841
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7074602842330933,
      "learning_rate": 0.00016613208838117362,
      "loss": 1.0322,
      "step": 2842
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7880011200904846,
      "learning_rate": 0.00016610953929346528,
      "loss": 1.106,
      "step": 2843
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7000258564949036,
      "learning_rate": 0.0001660869842331344,
      "loss": 0.9041,
      "step": 2844
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7121238708496094,
      "learning_rate": 0.00016606442320221863,
      "loss": 1.0979,
      "step": 2845
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6337499022483826,
      "learning_rate": 0.0001660418562027563,
      "loss": 1.0048,
      "step": 2846
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6515573263168335,
      "learning_rate": 0.0001660192832367862,
      "loss": 0.9387,
      "step": 2847
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.5596105456352234,
      "learning_rate": 0.0001659967043063477,
      "loss": 0.8096,
      "step": 2848
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.5970057249069214,
      "learning_rate": 0.00016597411941348056,
      "loss": 0.7847,
      "step": 2849
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6861267685890198,
      "learning_rate": 0.00016595152856022533,
      "loss": 0.9259,
      "step": 2850
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6038246750831604,
      "learning_rate": 0.00016592893174862286,
      "loss": 0.7822,
      "step": 2851
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7510673403739929,
      "learning_rate": 0.00016590632898071475,
      "loss": 1.031,
      "step": 2852
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7487673759460449,
      "learning_rate": 0.00016588372025854296,
      "loss": 0.9588,
      "step": 2853
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6498749852180481,
      "learning_rate": 0.00016586110558415004,
      "loss": 0.8258,
      "step": 2854
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6465758085250854,
      "learning_rate": 0.0001658384849595792,
      "loss": 1.0278,
      "step": 2855
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6948072910308838,
      "learning_rate": 0.00016581585838687396,
      "loss": 1.0128,
      "step": 2856
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7159262895584106,
      "learning_rate": 0.0001657932258680786,
      "loss": 0.8581,
      "step": 2857
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.672397255897522,
      "learning_rate": 0.0001657705874052378,
      "loss": 0.8933,
      "step": 2858
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6281071305274963,
      "learning_rate": 0.0001657479430003968,
      "loss": 0.8891,
      "step": 2859
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7124696969985962,
      "learning_rate": 0.00016572529265560144,
      "loss": 0.7881,
      "step": 2860
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8034883737564087,
      "learning_rate": 0.000165702636372898,
      "loss": 0.9732,
      "step": 2861
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7420786619186401,
      "learning_rate": 0.00016567997415433338,
      "loss": 0.9587,
      "step": 2862
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6224200129508972,
      "learning_rate": 0.00016565730600195497,
      "loss": 0.9015,
      "step": 2863
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7891188859939575,
      "learning_rate": 0.00016563463191781073,
      "loss": 1.0276,
      "step": 2864
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.659827709197998,
      "learning_rate": 0.00016561195190394908,
      "loss": 0.9243,
      "step": 2865
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6081632375717163,
      "learning_rate": 0.0001655892659624191,
      "loss": 0.8685,
      "step": 2866
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7055192589759827,
      "learning_rate": 0.0001655665740952703,
      "loss": 0.92,
      "step": 2867
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6848809719085693,
      "learning_rate": 0.0001655438763045528,
      "loss": 0.8801,
      "step": 2868
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7062857747077942,
      "learning_rate": 0.00016552117259231714,
      "loss": 0.8699,
      "step": 2869
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7173714637756348,
      "learning_rate": 0.00016549846296061454,
      "loss": 1.0067,
      "step": 2870
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6474378108978271,
      "learning_rate": 0.00016547574741149665,
      "loss": 0.9772,
      "step": 2871
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7030521631240845,
      "learning_rate": 0.0001654530259470157,
      "loss": 0.906,
      "step": 2872
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7073749303817749,
      "learning_rate": 0.00016543029856922449,
      "loss": 1.0654,
      "step": 2873
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6579055786132812,
      "learning_rate": 0.00016540756528017622,
      "loss": 1.1778,
      "step": 2874
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7000943422317505,
      "learning_rate": 0.00016538482608192476,
      "loss": 1.0288,
      "step": 2875
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6581107378005981,
      "learning_rate": 0.00016536208097652448,
      "loss": 0.7962,
      "step": 2876
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7475805282592773,
      "learning_rate": 0.00016533932996603027,
      "loss": 1.0036,
      "step": 2877
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.5845509171485901,
      "learning_rate": 0.00016531657305249756,
      "loss": 0.927,
      "step": 2878
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7028359174728394,
      "learning_rate": 0.00016529381023798232,
      "loss": 0.9387,
      "step": 2879
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.763886570930481,
      "learning_rate": 0.00016527104152454095,
      "loss": 1.0064,
      "step": 2880
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6850041747093201,
      "learning_rate": 0.0001652482669142306,
      "loss": 0.8708,
      "step": 2881
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8486610054969788,
      "learning_rate": 0.00016522548640910873,
      "loss": 1.0059,
      "step": 2882
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6982050538063049,
      "learning_rate": 0.0001652027000112335,
      "loss": 0.8896,
      "step": 2883
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6296117305755615,
      "learning_rate": 0.0001651799077226635,
      "loss": 0.8382,
      "step": 2884
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6951981782913208,
      "learning_rate": 0.00016515710954545786,
      "loss": 0.9466,
      "step": 2885
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6647318005561829,
      "learning_rate": 0.0001651343054816763,
      "loss": 0.9548,
      "step": 2886
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6394795775413513,
      "learning_rate": 0.00016511149553337904,
      "loss": 0.8903,
      "step": 2887
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6727846264839172,
      "learning_rate": 0.0001650886797026268,
      "loss": 0.9943,
      "step": 2888
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6580429673194885,
      "learning_rate": 0.00016506585799148092,
      "loss": 0.9331,
      "step": 2889
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6348177194595337,
      "learning_rate": 0.00016504303040200312,
      "loss": 1.1005,
      "step": 2890
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.709359884262085,
      "learning_rate": 0.00016502019693625584,
      "loss": 0.9764,
      "step": 2891
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7029663920402527,
      "learning_rate": 0.0001649973575963019,
      "loss": 1.0282,
      "step": 2892
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7602015137672424,
      "learning_rate": 0.00016497451238420467,
      "loss": 1.0261,
      "step": 2893
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6693970561027527,
      "learning_rate": 0.00016495166130202818,
      "loss": 1.0764,
      "step": 2894
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6728341579437256,
      "learning_rate": 0.00016492880435183685,
      "loss": 0.8672,
      "step": 2895
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6348689794540405,
      "learning_rate": 0.00016490594153569562,
      "loss": 0.9328,
      "step": 2896
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7363536357879639,
      "learning_rate": 0.00016488307285567012,
      "loss": 0.8982,
      "step": 2897
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6520256400108337,
      "learning_rate": 0.00016486019831382634,
      "loss": 1.1188,
      "step": 2898
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7071019411087036,
      "learning_rate": 0.00016483731791223084,
      "loss": 0.9359,
      "step": 2899
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9263062477111816,
      "learning_rate": 0.00016481443165295085,
      "loss": 1.1566,
      "step": 2900
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7108592987060547,
      "learning_rate": 0.0001647915395380539,
      "loss": 0.8543,
      "step": 2901
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7122064232826233,
      "learning_rate": 0.00016476864156960818,
      "loss": 0.9805,
      "step": 2902
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6613624691963196,
      "learning_rate": 0.00016474573774968244,
      "loss": 0.9594,
      "step": 2903
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6157373785972595,
      "learning_rate": 0.0001647228280803459,
      "loss": 0.9154,
      "step": 2904
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6679324507713318,
      "learning_rate": 0.00016469991256366824,
      "loss": 1.0046,
      "step": 2905
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6375901699066162,
      "learning_rate": 0.00016467699120171987,
      "loss": 0.9018,
      "step": 2906
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6826538443565369,
      "learning_rate": 0.00016465406399657155,
      "loss": 0.9845,
      "step": 2907
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7681721448898315,
      "learning_rate": 0.0001646311309502946,
      "loss": 0.8873,
      "step": 2908
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7108680605888367,
      "learning_rate": 0.00016460819206496087,
      "loss": 1.0239,
      "step": 2909
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6449684500694275,
      "learning_rate": 0.00016458524734264284,
      "loss": 1.0171,
      "step": 2910
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6971015334129333,
      "learning_rate": 0.00016456229678541337,
      "loss": 0.8124,
      "step": 2911
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6763261556625366,
      "learning_rate": 0.00016453934039534598,
      "loss": 0.9602,
      "step": 2912
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7010252475738525,
      "learning_rate": 0.00016451637817451457,
      "loss": 0.8596,
      "step": 2913
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6786449551582336,
      "learning_rate": 0.00016449341012499368,
      "loss": 1.0699,
      "step": 2914
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.718379557132721,
      "learning_rate": 0.00016447043624885836,
      "loss": 0.852,
      "step": 2915
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6729423999786377,
      "learning_rate": 0.00016444745654818415,
      "loss": 1.0049,
      "step": 2916
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6310647130012512,
      "learning_rate": 0.00016442447102504716,
      "loss": 0.852,
      "step": 2917
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.5880166888237,
      "learning_rate": 0.00016440147968152394,
      "loss": 0.9871,
      "step": 2918
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6796461343765259,
      "learning_rate": 0.00016437848251969173,
      "loss": 1.1046,
      "step": 2919
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.750541627407074,
      "learning_rate": 0.0001643554795416281,
      "loss": 0.9814,
      "step": 2920
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7341058850288391,
      "learning_rate": 0.00016433247074941124,
      "loss": 1.0621,
      "step": 2921
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6043815612792969,
      "learning_rate": 0.00016430945614511996,
      "loss": 0.9073,
      "step": 2922
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6640098690986633,
      "learning_rate": 0.0001642864357308334,
      "loss": 1.0701,
      "step": 2923
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6690117716789246,
      "learning_rate": 0.00016426340950863137,
      "loss": 0.9208,
      "step": 2924
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6856228709220886,
      "learning_rate": 0.00016424037748059416,
      "loss": 0.998,
      "step": 2925
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7497897148132324,
      "learning_rate": 0.00016421733964880252,
      "loss": 0.8903,
      "step": 2926
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8069843053817749,
      "learning_rate": 0.0001641942960153379,
      "loss": 0.9125,
      "step": 2927
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7221735119819641,
      "learning_rate": 0.00016417124658228208,
      "loss": 1.0176,
      "step": 2928
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6690991520881653,
      "learning_rate": 0.00016414819135171749,
      "loss": 0.8777,
      "step": 2929
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7555813789367676,
      "learning_rate": 0.000164125130325727,
      "loss": 1.0156,
      "step": 2930
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.615086555480957,
      "learning_rate": 0.00016410206350639403,
      "loss": 0.8144,
      "step": 2931
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6863670349121094,
      "learning_rate": 0.00016407899089580262,
      "loss": 1.0584,
      "step": 2932
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7557387351989746,
      "learning_rate": 0.00016405591249603715,
      "loss": 1.0769,
      "step": 2933
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6460722088813782,
      "learning_rate": 0.0001640328283091827,
      "loss": 0.8902,
      "step": 2934
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6990112662315369,
      "learning_rate": 0.00016400973833732477,
      "loss": 0.9806,
      "step": 2935
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7056894302368164,
      "learning_rate": 0.0001639866425825494,
      "loss": 0.8175,
      "step": 2936
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6659440994262695,
      "learning_rate": 0.00016396354104694316,
      "loss": 0.9374,
      "step": 2937
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7583017945289612,
      "learning_rate": 0.00016394043373259311,
      "loss": 1.0661,
      "step": 2938
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7467166185379028,
      "learning_rate": 0.00016391732064158695,
      "loss": 0.9484,
      "step": 2939
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6958436369895935,
      "learning_rate": 0.00016389420177601277,
      "loss": 0.978,
      "step": 2940
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7176093459129333,
      "learning_rate": 0.0001638710771379592,
      "loss": 0.9416,
      "step": 2941
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6962888240814209,
      "learning_rate": 0.00016384794672951545,
      "loss": 0.9828,
      "step": 2942
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6958292126655579,
      "learning_rate": 0.00016382481055277125,
      "loss": 0.9887,
      "step": 2943
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7200213074684143,
      "learning_rate": 0.00016380166860981675,
      "loss": 1.0077,
      "step": 2944
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.668902575969696,
      "learning_rate": 0.00016377852090274276,
      "loss": 0.9923,
      "step": 2945
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6741634011268616,
      "learning_rate": 0.0001637553674336405,
      "loss": 0.871,
      "step": 2946
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6794645190238953,
      "learning_rate": 0.0001637322082046018,
      "loss": 1.0129,
      "step": 2947
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.68716961145401,
      "learning_rate": 0.00016370904321771893,
      "loss": 0.8878,
      "step": 2948
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7383739948272705,
      "learning_rate": 0.00016368587247508472,
      "loss": 1.0159,
      "step": 2949
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7284265756607056,
      "learning_rate": 0.00016366269597879255,
      "loss": 1.0451,
      "step": 2950
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6858283281326294,
      "learning_rate": 0.0001636395137309362,
      "loss": 0.9618,
      "step": 2951
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7832133769989014,
      "learning_rate": 0.00016361632573361014,
      "loss": 1.0356,
      "step": 2952
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7388131618499756,
      "learning_rate": 0.00016359313198890924,
      "loss": 0.991,
      "step": 2953
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6978746056556702,
      "learning_rate": 0.00016356993249892895,
      "loss": 0.8741,
      "step": 2954
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7001532316207886,
      "learning_rate": 0.0001635467272657652,
      "loss": 0.817,
      "step": 2955
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6634668111801147,
      "learning_rate": 0.0001635235162915144,
      "loss": 0.8336,
      "step": 2956
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.669437050819397,
      "learning_rate": 0.00016350029957827355,
      "loss": 0.9604,
      "step": 2957
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7847330570220947,
      "learning_rate": 0.00016347707712814023,
      "loss": 1.0175,
      "step": 2958
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6637473702430725,
      "learning_rate": 0.0001634538489432124,
      "loss": 0.9076,
      "step": 2959
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.766559362411499,
      "learning_rate": 0.00016343061502558856,
      "loss": 1.0943,
      "step": 2960
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7539775371551514,
      "learning_rate": 0.00016340737537736777,
      "loss": 1.1895,
      "step": 2961
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6763971447944641,
      "learning_rate": 0.00016338413000064968,
      "loss": 1.0199,
      "step": 2962
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6886430978775024,
      "learning_rate": 0.00016336087889753432,
      "loss": 0.9311,
      "step": 2963
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6830310821533203,
      "learning_rate": 0.0001633376220701223,
      "loss": 0.8723,
      "step": 2964
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6681257486343384,
      "learning_rate": 0.0001633143595205147,
      "loss": 0.871,
      "step": 2965
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7215926051139832,
      "learning_rate": 0.0001632910912508132,
      "loss": 0.9047,
      "step": 2966
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6446928381919861,
      "learning_rate": 0.00016326781726312,
      "loss": 0.8068,
      "step": 2967
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9273528456687927,
      "learning_rate": 0.00016324453755953773,
      "loss": 0.9627,
      "step": 2968
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.70436692237854,
      "learning_rate": 0.00016322125214216957,
      "loss": 1.0719,
      "step": 2969
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6768920421600342,
      "learning_rate": 0.00016319796101311925,
      "loss": 0.8396,
      "step": 2970
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6657084822654724,
      "learning_rate": 0.00016317466417449094,
      "loss": 0.7522,
      "step": 2971
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7143867015838623,
      "learning_rate": 0.00016315136162838948,
      "loss": 1.0002,
      "step": 2972
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.808310866355896,
      "learning_rate": 0.00016312805337691998,
      "loss": 1.0182,
      "step": 2973
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6517670750617981,
      "learning_rate": 0.00016310473942218837,
      "loss": 0.9692,
      "step": 2974
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6355635523796082,
      "learning_rate": 0.0001630814197663008,
      "loss": 0.9723,
      "step": 2975
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6290289163589478,
      "learning_rate": 0.00016305809441136414,
      "loss": 1.1122,
      "step": 2976
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7016021609306335,
      "learning_rate": 0.00016303476335948567,
      "loss": 0.7976,
      "step": 2977
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7270112037658691,
      "learning_rate": 0.00016301142661277323,
      "loss": 0.9878,
      "step": 2978
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7442315220832825,
      "learning_rate": 0.00016298808417333516,
      "loss": 0.9822,
      "step": 2979
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6447054147720337,
      "learning_rate": 0.00016296473604328034,
      "loss": 0.6995,
      "step": 2980
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.76390141248703,
      "learning_rate": 0.0001629413822247181,
      "loss": 1.0782,
      "step": 2981
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7055835723876953,
      "learning_rate": 0.00016291802271975838,
      "loss": 1.006,
      "step": 2982
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.648766040802002,
      "learning_rate": 0.00016289465753051153,
      "loss": 0.8927,
      "step": 2983
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6227762699127197,
      "learning_rate": 0.0001628712866590885,
      "loss": 0.9941,
      "step": 2984
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7030214667320251,
      "learning_rate": 0.00016284791010760067,
      "loss": 0.6903,
      "step": 2985
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6903812289237976,
      "learning_rate": 0.00016282452787816002,
      "loss": 0.8768,
      "step": 2986
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6601381301879883,
      "learning_rate": 0.000162801139972879,
      "loss": 0.95,
      "step": 2987
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6409837603569031,
      "learning_rate": 0.00016277774639387053,
      "loss": 0.8799,
      "step": 2988
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7373121380805969,
      "learning_rate": 0.00016275434714324816,
      "loss": 0.8155,
      "step": 2989
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6737827062606812,
      "learning_rate": 0.0001627309422231258,
      "loss": 0.9852,
      "step": 2990
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.680059552192688,
      "learning_rate": 0.00016270753163561801,
      "loss": 0.9731,
      "step": 2991
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.722334623336792,
      "learning_rate": 0.00016268411538283977,
      "loss": 1.1091,
      "step": 2992
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6662548184394836,
      "learning_rate": 0.00016266069346690666,
      "loss": 0.7737,
      "step": 2993
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.6991024017333984,
      "learning_rate": 0.00016263726588993467,
      "loss": 0.8982,
      "step": 2994
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7102436423301697,
      "learning_rate": 0.00016261383265404034,
      "loss": 0.9056,
      "step": 2995
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6874223947525024,
      "learning_rate": 0.00016259039376134075,
      "loss": 0.9987,
      "step": 2996
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7075768709182739,
      "learning_rate": 0.00016256694921395348,
      "loss": 1.0127,
      "step": 2997
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6447400450706482,
      "learning_rate": 0.0001625434990139966,
      "loss": 0.8433,
      "step": 2998
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7359217405319214,
      "learning_rate": 0.0001625200431635887,
      "loss": 1.0163,
      "step": 2999
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7554381489753723,
      "learning_rate": 0.0001624965816648489,
      "loss": 0.9526,
      "step": 3000
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6736909747123718,
      "learning_rate": 0.00016247311451989677,
      "loss": 0.9332,
      "step": 3001
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6281025409698486,
      "learning_rate": 0.0001624496417308525,
      "loss": 0.8652,
      "step": 3002
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8367699384689331,
      "learning_rate": 0.00016242616329983665,
      "loss": 1.0465,
      "step": 3003
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6902986168861389,
      "learning_rate": 0.00016240267922897043,
      "loss": 0.7807,
      "step": 3004
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6087586879730225,
      "learning_rate": 0.00016237918952037545,
      "loss": 0.8017,
      "step": 3005
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6701231002807617,
      "learning_rate": 0.0001623556941761739,
      "loss": 0.9221,
      "step": 3006
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6698204278945923,
      "learning_rate": 0.00016233219319848843,
      "loss": 0.9335,
      "step": 3007
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7633811831474304,
      "learning_rate": 0.00016230868658944223,
      "loss": 0.9768,
      "step": 3008
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7034717202186584,
      "learning_rate": 0.000162285174351159,
      "loss": 0.9893,
      "step": 3009
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7095836997032166,
      "learning_rate": 0.0001622616564857629,
      "loss": 0.8335,
      "step": 3010
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7027610540390015,
      "learning_rate": 0.0001622381329953787,
      "loss": 0.9,
      "step": 3011
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6865144371986389,
      "learning_rate": 0.00016221460388213154,
      "loss": 0.7792,
      "step": 3012
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7391175627708435,
      "learning_rate": 0.0001621910691481472,
      "loss": 0.9275,
      "step": 3013
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8024181723594666,
      "learning_rate": 0.00016216752879555187,
      "loss": 0.956,
      "step": 3014
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6928214430809021,
      "learning_rate": 0.00016214398282647232,
      "loss": 0.9826,
      "step": 3015
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.669105052947998,
      "learning_rate": 0.00016212043124303572,
      "loss": 0.9564,
      "step": 3016
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7236560583114624,
      "learning_rate": 0.00016209687404736996,
      "loss": 0.9277,
      "step": 3017
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8411710262298584,
      "learning_rate": 0.0001620733112416032,
      "loss": 1.0461,
      "step": 3018
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.6422951221466064,
      "learning_rate": 0.00016204974282786424,
      "loss": 0.8462,
      "step": 3019
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7195187211036682,
      "learning_rate": 0.0001620261688082823,
      "loss": 0.94,
      "step": 3020
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7143303155899048,
      "learning_rate": 0.00016200258918498724,
      "loss": 0.9178,
      "step": 3021
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7510295510292053,
      "learning_rate": 0.0001619790039601093,
      "loss": 1.0951,
      "step": 3022
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7327381372451782,
      "learning_rate": 0.00016195541313577923,
      "loss": 1.0614,
      "step": 3023
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7491642236709595,
      "learning_rate": 0.0001619318167141284,
      "loss": 1.097,
      "step": 3024
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7633213996887207,
      "learning_rate": 0.00016190821469728862,
      "loss": 0.8888,
      "step": 3025
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7773520946502686,
      "learning_rate": 0.0001618846070873921,
      "loss": 1.1425,
      "step": 3026
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7367396950721741,
      "learning_rate": 0.00016186099388657176,
      "loss": 0.8996,
      "step": 3027
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7460110783576965,
      "learning_rate": 0.00016183737509696089,
      "loss": 1.006,
      "step": 3028
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7337568402290344,
      "learning_rate": 0.00016181375072069328,
      "loss": 0.9945,
      "step": 3029
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.6886031031608582,
      "learning_rate": 0.0001617901207599033,
      "loss": 0.9766,
      "step": 3030
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7199036478996277,
      "learning_rate": 0.00016176648521672575,
      "loss": 0.9924,
      "step": 3031
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7087488770484924,
      "learning_rate": 0.00016174284409329598,
      "loss": 0.9453,
      "step": 3032
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7542042136192322,
      "learning_rate": 0.00016171919739174984,
      "loss": 0.8999,
      "step": 3033
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7130687832832336,
      "learning_rate": 0.0001616955451142237,
      "loss": 1.0728,
      "step": 3034
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7343533039093018,
      "learning_rate": 0.00016167188726285434,
      "loss": 0.9288,
      "step": 3035
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.6714083552360535,
      "learning_rate": 0.00016164822383977914,
      "loss": 1.029,
      "step": 3036
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7564050555229187,
      "learning_rate": 0.00016162455484713598,
      "loss": 0.873,
      "step": 3037
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.6941550374031067,
      "learning_rate": 0.0001616008802870632,
      "loss": 0.862,
      "step": 3038
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7139540314674377,
      "learning_rate": 0.0001615772001616997,
      "loss": 1.0259,
      "step": 3039
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7256305813789368,
      "learning_rate": 0.00016155351447318477,
      "loss": 0.9499,
      "step": 3040
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7770005464553833,
      "learning_rate": 0.00016152982322365834,
      "loss": 0.9313,
      "step": 3041
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7816600203514099,
      "learning_rate": 0.00016150612641526078,
      "loss": 1.0,
      "step": 3042
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9455228447914124,
      "learning_rate": 0.00016148242405013294,
      "loss": 0.9221,
      "step": 3043
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7099979519844055,
      "learning_rate": 0.00016145871613041618,
      "loss": 0.964,
      "step": 3044
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8116803765296936,
      "learning_rate": 0.00016143500265825242,
      "loss": 0.9761,
      "step": 3045
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7324463129043579,
      "learning_rate": 0.000161411283635784,
      "loss": 1.0304,
      "step": 3046
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7201613783836365,
      "learning_rate": 0.00016138755906515382,
      "loss": 1.1438,
      "step": 3047
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6950251460075378,
      "learning_rate": 0.00016136382894850525,
      "loss": 0.9848,
      "step": 3048
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6754145622253418,
      "learning_rate": 0.00016134009328798218,
      "loss": 1.0516,
      "step": 3049
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7823755145072937,
      "learning_rate": 0.000161316352085729,
      "loss": 1.0282,
      "step": 3050
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7333949208259583,
      "learning_rate": 0.00016129260534389057,
      "loss": 0.8644,
      "step": 3051
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7001619338989258,
      "learning_rate": 0.00016126885306461225,
      "loss": 0.9158,
      "step": 3052
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7424895763397217,
      "learning_rate": 0.00016124509525004002,
      "loss": 1.1199,
      "step": 3053
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7295093536376953,
      "learning_rate": 0.00016122133190232018,
      "loss": 0.931,
      "step": 3054
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7016628384590149,
      "learning_rate": 0.00016119756302359963,
      "loss": 1.1039,
      "step": 3055
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7510272264480591,
      "learning_rate": 0.00016117378861602575,
      "loss": 0.9644,
      "step": 3056
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6671194434165955,
      "learning_rate": 0.00016115000868174644,
      "loss": 0.7718,
      "step": 3057
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7244356870651245,
      "learning_rate": 0.0001611262232229101,
      "loss": 0.9918,
      "step": 3058
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6976906657218933,
      "learning_rate": 0.00016110243224166557,
      "loss": 0.8962,
      "step": 3059
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6452173590660095,
      "learning_rate": 0.00016107863574016222,
      "loss": 0.8798,
      "step": 3060
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6883850693702698,
      "learning_rate": 0.00016105483372055,
      "loss": 1.0826,
      "step": 3061
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6843141913414001,
      "learning_rate": 0.00016103102618497922,
      "loss": 1.0096,
      "step": 3062
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7020462155342102,
      "learning_rate": 0.0001610072131356008,
      "loss": 0.8949,
      "step": 3063
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7438017129898071,
      "learning_rate": 0.00016098339457456607,
      "loss": 0.9674,
      "step": 3064
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6534295678138733,
      "learning_rate": 0.00016095957050402696,
      "loss": 0.8764,
      "step": 3065
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6544593572616577,
      "learning_rate": 0.00016093574092613578,
      "loss": 0.8218,
      "step": 3066
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6656880974769592,
      "learning_rate": 0.0001609119058430454,
      "loss": 0.999,
      "step": 3067
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7666613459587097,
      "learning_rate": 0.00016088806525690926,
      "loss": 1.0574,
      "step": 3068
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7083278894424438,
      "learning_rate": 0.00016086421916988117,
      "loss": 1.0048,
      "step": 3069
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6306576728820801,
      "learning_rate": 0.00016084036758411547,
      "loss": 0.9398,
      "step": 3070
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7330958247184753,
      "learning_rate": 0.00016081651050176704,
      "loss": 0.9751,
      "step": 3071
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.901148796081543,
      "learning_rate": 0.00016079264792499124,
      "loss": 0.8909,
      "step": 3072
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7629517316818237,
      "learning_rate": 0.0001607687798559439,
      "loss": 1.0978,
      "step": 3073
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8078415393829346,
      "learning_rate": 0.00016074490629678137,
      "loss": 0.9588,
      "step": 3074
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8237608671188354,
      "learning_rate": 0.00016072102724966046,
      "loss": 1.1306,
      "step": 3075
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6664127111434937,
      "learning_rate": 0.00016069714271673856,
      "loss": 0.954,
      "step": 3076
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6480268239974976,
      "learning_rate": 0.00016067325270017352,
      "loss": 0.8522,
      "step": 3077
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7148239612579346,
      "learning_rate": 0.0001606493572021236,
      "loss": 0.9399,
      "step": 3078
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6466790437698364,
      "learning_rate": 0.00016062545622474764,
      "loss": 0.7129,
      "step": 3079
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6735865473747253,
      "learning_rate": 0.00016060154977020502,
      "loss": 0.8239,
      "step": 3080
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7058208584785461,
      "learning_rate": 0.00016057763784065542,
      "loss": 0.9673,
      "step": 3081
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6586633324623108,
      "learning_rate": 0.0001605537204382593,
      "loss": 1.0528,
      "step": 3082
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6823360323905945,
      "learning_rate": 0.0001605297975651774,
      "loss": 1.0073,
      "step": 3083
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7752652168273926,
      "learning_rate": 0.00016050586922357094,
      "loss": 1.0964,
      "step": 3084
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7399365305900574,
      "learning_rate": 0.00016048193541560183,
      "loss": 1.0044,
      "step": 3085
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.689049482345581,
      "learning_rate": 0.00016045799614343228,
      "loss": 0.8952,
      "step": 3086
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7327082753181458,
      "learning_rate": 0.00016043405140922512,
      "loss": 1.0547,
      "step": 3087
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7117489576339722,
      "learning_rate": 0.00016041010121514357,
      "loss": 1.0172,
      "step": 3088
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7927287817001343,
      "learning_rate": 0.00016038614556335143,
      "loss": 0.9599,
      "step": 3089
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6660590767860413,
      "learning_rate": 0.00016036218445601293,
      "loss": 0.9384,
      "step": 3090
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6778466105461121,
      "learning_rate": 0.00016033821789529284,
      "loss": 0.8565,
      "step": 3091
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.694474995136261,
      "learning_rate": 0.0001603142458833564,
      "loss": 0.9739,
      "step": 3092
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8222754001617432,
      "learning_rate": 0.00016029026842236935,
      "loss": 1.0106,
      "step": 3093
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7755143642425537,
      "learning_rate": 0.0001602662855144979,
      "loss": 0.9953,
      "step": 3094
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.6970244646072388,
      "learning_rate": 0.00016024229716190877,
      "loss": 0.8954,
      "step": 3095
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9485830068588257,
      "learning_rate": 0.0001602183033667692,
      "loss": 1.0471,
      "step": 3096
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7483344078063965,
      "learning_rate": 0.00016019430413124686,
      "loss": 1.123,
      "step": 3097
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7733362913131714,
      "learning_rate": 0.00016017029945750996,
      "loss": 0.9967,
      "step": 3098
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.857623815536499,
      "learning_rate": 0.0001601462893477272,
      "loss": 1.0583,
      "step": 3099
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7281124591827393,
      "learning_rate": 0.00016012227380406775,
      "loss": 1.0528,
      "step": 3100
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.673368513584137,
      "learning_rate": 0.00016009825282870126,
      "loss": 1.1355,
      "step": 3101
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7686245441436768,
      "learning_rate": 0.0001600742264237979,
      "loss": 1.1185,
      "step": 3102
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7917776703834534,
      "learning_rate": 0.00016005019459152834,
      "loss": 1.0218,
      "step": 3103
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.626288652420044,
      "learning_rate": 0.0001600261573340637,
      "loss": 0.8897,
      "step": 3104
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7067705392837524,
      "learning_rate": 0.0001600021146535756,
      "loss": 1.0538,
      "step": 3105
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.6985976099967957,
      "learning_rate": 0.00015997806655223618,
      "loss": 0.9316,
      "step": 3106
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.6789809465408325,
      "learning_rate": 0.000159954013032218,
      "loss": 0.8942,
      "step": 3107
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7024329900741577,
      "learning_rate": 0.00015992995409569428,
      "loss": 0.8517,
      "step": 3108
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7696155309677124,
      "learning_rate": 0.0001599058897448385,
      "loss": 1.0287,
      "step": 3109
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.697361946105957,
      "learning_rate": 0.00015988181998182475,
      "loss": 1.0239,
      "step": 3110
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.6460540294647217,
      "learning_rate": 0.00015985774480882758,
      "loss": 0.8986,
      "step": 3111
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7324708104133606,
      "learning_rate": 0.00015983366422802213,
      "loss": 1.0674,
      "step": 3112
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8241015076637268,
      "learning_rate": 0.0001598095782415839,
      "loss": 0.9858,
      "step": 3113
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.6638197302818298,
      "learning_rate": 0.00015978548685168892,
      "loss": 0.8558,
      "step": 3114
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7642309069633484,
      "learning_rate": 0.00015976139006051366,
      "loss": 0.9775,
      "step": 3115
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7588913440704346,
      "learning_rate": 0.0001597372878702352,
      "loss": 0.9801,
      "step": 3116
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.6915333271026611,
      "learning_rate": 0.00015971318028303104,
      "loss": 0.9893,
      "step": 3117
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7214621305465698,
      "learning_rate": 0.0001596890673010791,
      "loss": 0.8739,
      "step": 3118
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.6884421706199646,
      "learning_rate": 0.0001596649489265579,
      "loss": 0.9756,
      "step": 3119
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8056734800338745,
      "learning_rate": 0.0001596408251616464,
      "loss": 1.0987,
      "step": 3120
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7120291590690613,
      "learning_rate": 0.000159616696008524,
      "loss": 1.0054,
      "step": 3121
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7776278853416443,
      "learning_rate": 0.00015959256146937068,
      "loss": 1.0415,
      "step": 3122
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7185929417610168,
      "learning_rate": 0.00015956842154636687,
      "loss": 0.9976,
      "step": 3123
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7339257001876831,
      "learning_rate": 0.00015954427624169344,
      "loss": 0.9121,
      "step": 3124
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8103228807449341,
      "learning_rate": 0.00015952012555753178,
      "loss": 0.9984,
      "step": 3125
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7015619277954102,
      "learning_rate": 0.00015949596949606376,
      "loss": 0.9733,
      "step": 3126
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9180814027786255,
      "learning_rate": 0.0001594718080594718,
      "loss": 1.0364,
      "step": 3127
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7444248199462891,
      "learning_rate": 0.00015944764124993868,
      "loss": 0.9475,
      "step": 3128
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.6662657856941223,
      "learning_rate": 0.00015942346906964777,
      "loss": 0.7493,
      "step": 3129
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.6935024261474609,
      "learning_rate": 0.00015939929152078287,
      "loss": 1.0452,
      "step": 3130
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7142334580421448,
      "learning_rate": 0.0001593751086055283,
      "loss": 0.7778,
      "step": 3131
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7099422812461853,
      "learning_rate": 0.00015935092032606888,
      "loss": 0.9093,
      "step": 3132
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7817436456680298,
      "learning_rate": 0.00015932672668458986,
      "loss": 1.115,
      "step": 3133
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7280943393707275,
      "learning_rate": 0.00015930252768327696,
      "loss": 0.998,
      "step": 3134
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7231597900390625,
      "learning_rate": 0.00015927832332431646,
      "loss": 0.92,
      "step": 3135
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7060204744338989,
      "learning_rate": 0.0001592541136098951,
      "loss": 1.0271,
      "step": 3136
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.717904806137085,
      "learning_rate": 0.00015922989854220005,
      "loss": 0.981,
      "step": 3137
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7248113751411438,
      "learning_rate": 0.00015920567812341904,
      "loss": 0.9549,
      "step": 3138
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.6890550851821899,
      "learning_rate": 0.0001591814523557402,
      "loss": 1.0144,
      "step": 3139
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8595572710037231,
      "learning_rate": 0.00015915722124135227,
      "loss": 1.1475,
      "step": 3140
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7589665651321411,
      "learning_rate": 0.0001591329847824443,
      "loss": 0.9801,
      "step": 3141
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.6806226372718811,
      "learning_rate": 0.000159108742981206,
      "loss": 0.7727,
      "step": 3142
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9598560333251953,
      "learning_rate": 0.00015908449583982743,
      "loss": 1.101,
      "step": 3143
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.734533429145813,
      "learning_rate": 0.00015906024336049922,
      "loss": 0.9844,
      "step": 3144
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7020285725593567,
      "learning_rate": 0.0001590359855454124,
      "loss": 1.0375,
      "step": 3145
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8799060583114624,
      "learning_rate": 0.00015901172239675851,
      "loss": 0.9193,
      "step": 3146
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7654073238372803,
      "learning_rate": 0.00015898745391672967,
      "loss": 0.9879,
      "step": 3147
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.6300762295722961,
      "learning_rate": 0.00015896318010751837,
      "loss": 1.0095,
      "step": 3148
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7860686779022217,
      "learning_rate": 0.0001589389009713176,
      "loss": 1.0894,
      "step": 3149
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7179495692253113,
      "learning_rate": 0.0001589146165103208,
      "loss": 0.9753,
      "step": 3150
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7529764175415039,
      "learning_rate": 0.00015889032672672198,
      "loss": 0.9269,
      "step": 3151
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.6052923798561096,
      "learning_rate": 0.00015886603162271556,
      "loss": 0.7845,
      "step": 3152
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7489050626754761,
      "learning_rate": 0.00015884173120049653,
      "loss": 0.9919,
      "step": 3153
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7350334525108337,
      "learning_rate": 0.00015881742546226018,
      "loss": 1.0606,
      "step": 3154
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7081111669540405,
      "learning_rate": 0.0001587931144102025,
      "loss": 0.8885,
      "step": 3155
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.775686502456665,
      "learning_rate": 0.00015876879804651984,
      "loss": 0.9711,
      "step": 3156
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7543262839317322,
      "learning_rate": 0.00015874447637340898,
      "loss": 1.1627,
      "step": 3157
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7249549627304077,
      "learning_rate": 0.00015872014939306733,
      "loss": 0.8961,
      "step": 3158
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7103540897369385,
      "learning_rate": 0.00015869581710769266,
      "loss": 0.9714,
      "step": 3159
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7673888802528381,
      "learning_rate": 0.00015867147951948324,
      "loss": 1.08,
      "step": 3160
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7711588144302368,
      "learning_rate": 0.0001586471366306378,
      "loss": 1.1145,
      "step": 3161
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.616401195526123,
      "learning_rate": 0.00015862278844335566,
      "loss": 0.935,
      "step": 3162
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.6484968066215515,
      "learning_rate": 0.00015859843495983651,
      "loss": 0.9124,
      "step": 3163
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.6611524224281311,
      "learning_rate": 0.00015857407618228054,
      "loss": 0.7792,
      "step": 3164
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.6965488195419312,
      "learning_rate": 0.00015854971211288845,
      "loss": 0.7451,
      "step": 3165
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.638028085231781,
      "learning_rate": 0.00015852534275386134,
      "loss": 0.9095,
      "step": 3166
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.740543007850647,
      "learning_rate": 0.0001585009681074009,
      "loss": 1.0657,
      "step": 3167
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.6465707421302795,
      "learning_rate": 0.00015847658817570923,
      "loss": 0.9527,
      "step": 3168
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.692480742931366,
      "learning_rate": 0.0001584522029609889,
      "loss": 0.961,
      "step": 3169
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7205979824066162,
      "learning_rate": 0.00015842781246544298,
      "loss": 1.0782,
      "step": 3170
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.6649343967437744,
      "learning_rate": 0.00015840341669127504,
      "loss": 0.8621,
      "step": 3171
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.6777862906455994,
      "learning_rate": 0.00015837901564068907,
      "loss": 0.8381,
      "step": 3172
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7257649898529053,
      "learning_rate": 0.00015835460931588958,
      "loss": 1.0025,
      "step": 3173
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.764773964881897,
      "learning_rate": 0.00015833019771908154,
      "loss": 0.9403,
      "step": 3174
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7970273494720459,
      "learning_rate": 0.00015830578085247043,
      "loss": 1.1609,
      "step": 3175
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7514870762825012,
      "learning_rate": 0.0001582813587182621,
      "loss": 0.8338,
      "step": 3176
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.6733028888702393,
      "learning_rate": 0.00015825693131866304,
      "loss": 0.8313,
      "step": 3177
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7544176578521729,
      "learning_rate": 0.00015823249865588003,
      "loss": 0.9311,
      "step": 3178
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7014653086662292,
      "learning_rate": 0.00015820806073212055,
      "loss": 0.9614,
      "step": 3179
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7118635177612305,
      "learning_rate": 0.00015818361754959233,
      "loss": 1.0199,
      "step": 3180
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7127907872200012,
      "learning_rate": 0.00015815916911050365,
      "loss": 1.1185,
      "step": 3181
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.6535012722015381,
      "learning_rate": 0.00015813471541706336,
      "loss": 0.8695,
      "step": 3182
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8021343946456909,
      "learning_rate": 0.00015811025647148073,
      "loss": 0.8705,
      "step": 3183
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7726711630821228,
      "learning_rate": 0.00015808579227596543,
      "loss": 0.8476,
      "step": 3184
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7215268015861511,
      "learning_rate": 0.00015806132283272768,
      "loss": 0.9388,
      "step": 3185
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7374808192253113,
      "learning_rate": 0.00015803684814397813,
      "loss": 1.0008,
      "step": 3186
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7936142683029175,
      "learning_rate": 0.00015801236821192798,
      "loss": 0.9965,
      "step": 3187
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.6195435523986816,
      "learning_rate": 0.00015798788303878882,
      "loss": 0.9651,
      "step": 3188
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7308238744735718,
      "learning_rate": 0.00015796339262677273,
      "loss": 0.8503,
      "step": 3189
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.672542154788971,
      "learning_rate": 0.00015793889697809235,
      "loss": 0.9087,
      "step": 3190
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.6192045211791992,
      "learning_rate": 0.00015791439609496064,
      "loss": 0.9253,
      "step": 3191
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7407450079917908,
      "learning_rate": 0.00015788988997959114,
      "loss": 1.1328,
      "step": 3192
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.6472103595733643,
      "learning_rate": 0.00015786537863419787,
      "loss": 0.8499,
      "step": 3193
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.765039324760437,
      "learning_rate": 0.0001578408620609953,
      "loss": 0.9891,
      "step": 3194
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.6693131923675537,
      "learning_rate": 0.00015781634026219835,
      "loss": 0.8916,
      "step": 3195
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.6833876371383667,
      "learning_rate": 0.00015779181324002234,
      "loss": 0.9281,
      "step": 3196
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7620227336883545,
      "learning_rate": 0.00015776728099668326,
      "loss": 0.9676,
      "step": 3197
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.6861891150474548,
      "learning_rate": 0.00015774274353439743,
      "loss": 1.0581,
      "step": 3198
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.6912551522254944,
      "learning_rate": 0.00015771820085538164,
      "loss": 0.889,
      "step": 3199
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7697704434394836,
      "learning_rate": 0.00015769365296185324,
      "loss": 1.0187,
      "step": 3200
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7657372355461121,
      "learning_rate": 0.00015766909985602992,
      "loss": 1.1127,
      "step": 3201
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7877035140991211,
      "learning_rate": 0.00015764454154012993,
      "loss": 0.9326,
      "step": 3202
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7275406122207642,
      "learning_rate": 0.00015761997801637203,
      "loss": 0.9775,
      "step": 3203
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.6876859068870544,
      "learning_rate": 0.00015759540928697536,
      "loss": 0.9551,
      "step": 3204
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.6654580235481262,
      "learning_rate": 0.00015757083535415957,
      "loss": 0.8806,
      "step": 3205
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7412167191505432,
      "learning_rate": 0.00015754625622014474,
      "loss": 0.9021,
      "step": 3206
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7922194004058838,
      "learning_rate": 0.0001575216718871515,
      "loss": 0.9801,
      "step": 3207
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.6838287115097046,
      "learning_rate": 0.00015749708235740091,
      "loss": 1.0322,
      "step": 3208
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8312490582466125,
      "learning_rate": 0.00015747248763311444,
      "loss": 0.928,
      "step": 3209
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7289003729820251,
      "learning_rate": 0.00015744788771651416,
      "loss": 0.9353,
      "step": 3210
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7136335968971252,
      "learning_rate": 0.00015742328260982247,
      "loss": 1.0114,
      "step": 3211
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.6795276999473572,
      "learning_rate": 0.00015739867231526233,
      "loss": 0.9974,
      "step": 3212
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.6614225506782532,
      "learning_rate": 0.00015737405683505714,
      "loss": 0.8994,
      "step": 3213
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.748741865158081,
      "learning_rate": 0.00015734943617143077,
      "loss": 0.9645,
      "step": 3214
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.657394289970398,
      "learning_rate": 0.00015732481032660755,
      "loss": 0.9442,
      "step": 3215
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7445186376571655,
      "learning_rate": 0.0001573001793028123,
      "loss": 0.9267,
      "step": 3216
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7444228529930115,
      "learning_rate": 0.00015727554310227023,
      "loss": 0.7558,
      "step": 3217
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7758005261421204,
      "learning_rate": 0.0001572509017272072,
      "loss": 0.8827,
      "step": 3218
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7533059120178223,
      "learning_rate": 0.00015722625517984934,
      "loss": 0.8436,
      "step": 3219
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7460200190544128,
      "learning_rate": 0.00015720160346242333,
      "loss": 0.9243,
      "step": 3220
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.6741672158241272,
      "learning_rate": 0.00015717694657715629,
      "loss": 0.9178,
      "step": 3221
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.744194746017456,
      "learning_rate": 0.0001571522845262759,
      "loss": 0.935,
      "step": 3222
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.6983630657196045,
      "learning_rate": 0.0001571276173120102,
      "loss": 0.8223,
      "step": 3223
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7060294151306152,
      "learning_rate": 0.00015710294493658772,
      "loss": 0.972,
      "step": 3224
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7305655479431152,
      "learning_rate": 0.00015707826740223748,
      "loss": 0.7972,
      "step": 3225
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7339190244674683,
      "learning_rate": 0.00015705358471118896,
      "loss": 1.0512,
      "step": 3226
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8077118992805481,
      "learning_rate": 0.0001570288968656721,
      "loss": 0.8467,
      "step": 3227
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8146405816078186,
      "learning_rate": 0.0001570042038679173,
      "loss": 0.9306,
      "step": 3228
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.628921627998352,
      "learning_rate": 0.00015697950572015544,
      "loss": 0.8804,
      "step": 3229
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.6738828420639038,
      "learning_rate": 0.00015695480242461783,
      "loss": 0.6829,
      "step": 3230
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.720912754535675,
      "learning_rate": 0.00015693009398353634,
      "loss": 1.0782,
      "step": 3231
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7239159941673279,
      "learning_rate": 0.00015690538039914316,
      "loss": 0.9751,
      "step": 3232
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.6557945609092712,
      "learning_rate": 0.00015688066167367107,
      "loss": 0.9095,
      "step": 3233
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7212446331977844,
      "learning_rate": 0.00015685593780935325,
      "loss": 1.041,
      "step": 3234
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7091822624206543,
      "learning_rate": 0.00015683120880842342,
      "loss": 1.034,
      "step": 3235
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.755889356136322,
      "learning_rate": 0.00015680647467311557,
      "loss": 1.0469,
      "step": 3236
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7285449504852295,
      "learning_rate": 0.00015678173540566444,
      "loss": 0.9419,
      "step": 3237
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7696725726127625,
      "learning_rate": 0.00015675699100830496,
      "loss": 1.1016,
      "step": 3238
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.6976828575134277,
      "learning_rate": 0.00015673224148327275,
      "loss": 0.9678,
      "step": 3239
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.6844419836997986,
      "learning_rate": 0.0001567074868328037,
      "loss": 0.893,
      "step": 3240
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7798510193824768,
      "learning_rate": 0.00015668272705913428,
      "loss": 0.953,
      "step": 3241
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.6004636287689209,
      "learning_rate": 0.00015665796216450146,
      "loss": 0.7275,
      "step": 3242
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8678481578826904,
      "learning_rate": 0.0001566331921511425,
      "loss": 1.0891,
      "step": 3243
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7131955027580261,
      "learning_rate": 0.00015660841702129532,
      "loss": 1.0129,
      "step": 3244
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7723953723907471,
      "learning_rate": 0.00015658363677719818,
      "loss": 1.0193,
      "step": 3245
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7669752836227417,
      "learning_rate": 0.0001565588514210898,
      "loss": 0.8838,
      "step": 3246
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7441872954368591,
      "learning_rate": 0.00015653406095520947,
      "loss": 1.0323,
      "step": 3247
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7313008308410645,
      "learning_rate": 0.00015650926538179682,
      "loss": 1.075,
      "step": 3248
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8604608178138733,
      "learning_rate": 0.00015648446470309198,
      "loss": 0.8467,
      "step": 3249
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7757964134216309,
      "learning_rate": 0.00015645965892133559,
      "loss": 0.984,
      "step": 3250
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.6894923448562622,
      "learning_rate": 0.00015643484803876866,
      "loss": 0.7665,
      "step": 3251
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7314807176589966,
      "learning_rate": 0.00015641003205763275,
      "loss": 1.0821,
      "step": 3252
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.6835300922393799,
      "learning_rate": 0.00015638521098016982,
      "loss": 0.7995,
      "step": 3253
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7496686577796936,
      "learning_rate": 0.00015636038480862239,
      "loss": 0.9026,
      "step": 3254
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7328741550445557,
      "learning_rate": 0.0001563355535452333,
      "loss": 0.8307,
      "step": 3255
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8298733234405518,
      "learning_rate": 0.00015631071719224583,
      "loss": 0.9103,
      "step": 3256
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8060025572776794,
      "learning_rate": 0.00015628587575190395,
      "loss": 1.0579,
      "step": 3257
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7077762484550476,
      "learning_rate": 0.00015626102922645192,
      "loss": 0.8934,
      "step": 3258
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7120048403739929,
      "learning_rate": 0.0001562361776181344,
      "loss": 0.9242,
      "step": 3259
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8317088484764099,
      "learning_rate": 0.00015621132092919667,
      "loss": 0.94,
      "step": 3260
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7645663022994995,
      "learning_rate": 0.00015618645916188432,
      "loss": 0.9364,
      "step": 3261
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7070802450180054,
      "learning_rate": 0.00015616159231844354,
      "loss": 0.84,
      "step": 3262
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7429453134536743,
      "learning_rate": 0.00015613672040112086,
      "loss": 0.8429,
      "step": 3263
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7282542586326599,
      "learning_rate": 0.0001561118434121634,
      "loss": 0.8897,
      "step": 3264
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7371309995651245,
      "learning_rate": 0.00015608696135381851,
      "loss": 0.9725,
      "step": 3265
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.6809751391410828,
      "learning_rate": 0.00015606207422833425,
      "loss": 0.9137,
      "step": 3266
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.6739715933799744,
      "learning_rate": 0.00015603718203795902,
      "loss": 1.0297,
      "step": 3267
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7235797047615051,
      "learning_rate": 0.00015601228478494166,
      "loss": 0.8851,
      "step": 3268
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7177626490592957,
      "learning_rate": 0.00015598738247153156,
      "loss": 0.8884,
      "step": 3269
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7195912599563599,
      "learning_rate": 0.00015596247509997842,
      "loss": 0.992,
      "step": 3270
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.6846754550933838,
      "learning_rate": 0.0001559375626725325,
      "loss": 1.073,
      "step": 3271
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8329934477806091,
      "learning_rate": 0.0001559126451914445,
      "loss": 1.0911,
      "step": 3272
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7074479460716248,
      "learning_rate": 0.0001558877226589656,
      "loss": 0.991,
      "step": 3273
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7118126749992371,
      "learning_rate": 0.00015586279507734743,
      "loss": 1.0068,
      "step": 3274
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7480494379997253,
      "learning_rate": 0.00015583786244884198,
      "loss": 1.0171,
      "step": 3275
    },
    {
      "epoch": 1.23,
      "eval_loss": 0.9602957367897034,
      "eval_runtime": 17.812,
      "eval_samples_per_second": 153.211,
      "eval_steps_per_second": 38.345,
      "step": 3275
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.6866698265075684,
      "learning_rate": 0.0001558129247757018,
      "loss": 1.0152,
      "step": 3276
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7157483696937561,
      "learning_rate": 0.00015578798206017987,
      "loss": 0.9652,
      "step": 3277
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8190251588821411,
      "learning_rate": 0.00015576303430452968,
      "loss": 1.0241,
      "step": 3278
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7425306439399719,
      "learning_rate": 0.00015573808151100505,
      "loss": 1.0053,
      "step": 3279
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.6862152218818665,
      "learning_rate": 0.00015571312368186036,
      "loss": 0.9581,
      "step": 3280
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7101152539253235,
      "learning_rate": 0.00015568816081935034,
      "loss": 0.8594,
      "step": 3281
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.6727707386016846,
      "learning_rate": 0.00015566319292573035,
      "loss": 0.9591,
      "step": 3282
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.6215928792953491,
      "learning_rate": 0.000155638220003256,
      "loss": 0.8791,
      "step": 3283
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8467756509780884,
      "learning_rate": 0.00015561324205418353,
      "loss": 1.0264,
      "step": 3284
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7936529517173767,
      "learning_rate": 0.0001555882590807695,
      "loss": 0.8712,
      "step": 3285
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7405680418014526,
      "learning_rate": 0.000155563271085271,
      "loss": 0.8966,
      "step": 3286
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7553184032440186,
      "learning_rate": 0.00015553827806994556,
      "loss": 0.9128,
      "step": 3287
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7074669599533081,
      "learning_rate": 0.00015551328003705117,
      "loss": 0.7814,
      "step": 3288
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7011464834213257,
      "learning_rate": 0.00015548827698884626,
      "loss": 1.133,
      "step": 3289
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.71796715259552,
      "learning_rate": 0.00015546326892758964,
      "loss": 0.919,
      "step": 3290
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.6808836460113525,
      "learning_rate": 0.00015543825585554073,
      "loss": 0.835,
      "step": 3291
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7186148166656494,
      "learning_rate": 0.0001554132377749593,
      "loss": 0.8625,
      "step": 3292
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7094897031784058,
      "learning_rate": 0.0001553882146881056,
      "loss": 0.8783,
      "step": 3293
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7546431422233582,
      "learning_rate": 0.0001553631865972403,
      "loss": 0.9686,
      "step": 3294
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.6863462924957275,
      "learning_rate": 0.00015533815350462455,
      "loss": 1.0179,
      "step": 3295
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.6790794134140015,
      "learning_rate": 0.00015531311541251995,
      "loss": 0.9417,
      "step": 3296
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7436459064483643,
      "learning_rate": 0.00015528807232318857,
      "loss": 0.9276,
      "step": 3297
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.732758641242981,
      "learning_rate": 0.00015526302423889287,
      "loss": 1.0102,
      "step": 3298
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7761924266815186,
      "learning_rate": 0.00015523797116189587,
      "loss": 1.1556,
      "step": 3299
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7570998668670654,
      "learning_rate": 0.00015521291309446089,
      "loss": 0.8907,
      "step": 3300
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7217000722885132,
      "learning_rate": 0.00015518785003885185,
      "loss": 0.9827,
      "step": 3301
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9155686497688293,
      "learning_rate": 0.00015516278199733302,
      "loss": 0.8761,
      "step": 3302
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8354843854904175,
      "learning_rate": 0.00015513770897216918,
      "loss": 1.1249,
      "step": 3303
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7380690574645996,
      "learning_rate": 0.0001551126309656255,
      "loss": 0.8494,
      "step": 3304
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.73750239610672,
      "learning_rate": 0.0001550875479799677,
      "loss": 1.0285,
      "step": 3305
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.793064296245575,
      "learning_rate": 0.0001550624600174618,
      "loss": 0.8748,
      "step": 3306
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7800728678703308,
      "learning_rate": 0.0001550373670803744,
      "loss": 0.9082,
      "step": 3307
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.6863796710968018,
      "learning_rate": 0.00015501226917097257,
      "loss": 0.9241,
      "step": 3308
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8050414323806763,
      "learning_rate": 0.00015498716629152367,
      "loss": 0.9243,
      "step": 3309
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6615169048309326,
      "learning_rate": 0.0001549620584442956,
      "loss": 0.8398,
      "step": 3310
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7062418460845947,
      "learning_rate": 0.00015493694563155675,
      "loss": 1.0246,
      "step": 3311
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6960569024085999,
      "learning_rate": 0.00015491182785557593,
      "loss": 0.9849,
      "step": 3312
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7825234532356262,
      "learning_rate": 0.00015488670511862237,
      "loss": 1.0409,
      "step": 3313
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7394260764122009,
      "learning_rate": 0.00015486157742296576,
      "loss": 0.8847,
      "step": 3314
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7210456728935242,
      "learning_rate": 0.00015483644477087626,
      "loss": 1.0684,
      "step": 3315
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.786108672618866,
      "learning_rate": 0.00015481130716462447,
      "loss": 0.7895,
      "step": 3316
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.639926552772522,
      "learning_rate": 0.00015478616460648143,
      "loss": 0.86,
      "step": 3317
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7433844208717346,
      "learning_rate": 0.0001547610170987186,
      "loss": 0.7851,
      "step": 3318
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7296815514564514,
      "learning_rate": 0.0001547358646436079,
      "loss": 0.9552,
      "step": 3319
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6938995122909546,
      "learning_rate": 0.00015471070724342177,
      "loss": 1.0096,
      "step": 3320
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8119737505912781,
      "learning_rate": 0.00015468554490043302,
      "loss": 1.0975,
      "step": 3321
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6842712759971619,
      "learning_rate": 0.0001546603776169149,
      "loss": 1.0019,
      "step": 3322
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7045173048973083,
      "learning_rate": 0.00015463520539514116,
      "loss": 0.9034,
      "step": 3323
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6672970652580261,
      "learning_rate": 0.00015461002823738597,
      "loss": 0.8547,
      "step": 3324
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7053340673446655,
      "learning_rate": 0.00015458484614592393,
      "loss": 0.9211,
      "step": 3325
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6476091146469116,
      "learning_rate": 0.00015455965912303005,
      "loss": 0.8302,
      "step": 3326
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7886658310890198,
      "learning_rate": 0.00015453446717097994,
      "loss": 1.0586,
      "step": 3327
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6936331987380981,
      "learning_rate": 0.00015450927029204948,
      "loss": 0.9953,
      "step": 3328
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8038539290428162,
      "learning_rate": 0.00015448406848851508,
      "loss": 1.0148,
      "step": 3329
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7686813473701477,
      "learning_rate": 0.00015445886176265354,
      "loss": 0.9441,
      "step": 3330
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7950506210327148,
      "learning_rate": 0.00015443365011674222,
      "loss": 1.0075,
      "step": 3331
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7895020246505737,
      "learning_rate": 0.0001544084335530588,
      "loss": 0.9223,
      "step": 3332
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7187060713768005,
      "learning_rate": 0.0001543832120738815,
      "loss": 0.9758,
      "step": 3333
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.771331787109375,
      "learning_rate": 0.0001543579856814889,
      "loss": 1.0886,
      "step": 3334
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7543261051177979,
      "learning_rate": 0.00015433275437816004,
      "loss": 0.9031,
      "step": 3335
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7562270164489746,
      "learning_rate": 0.00015430751816617445,
      "loss": 1.0196,
      "step": 3336
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.688003659248352,
      "learning_rate": 0.0001542822770478121,
      "loss": 1.0277,
      "step": 3337
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7344198226928711,
      "learning_rate": 0.00015425703102535339,
      "loss": 1.0609,
      "step": 3338
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8227985501289368,
      "learning_rate": 0.0001542317801010791,
      "loss": 0.9561,
      "step": 3339
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.6553336977958679,
      "learning_rate": 0.00015420652427727054,
      "loss": 0.9462,
      "step": 3340
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7725002765655518,
      "learning_rate": 0.00015418126355620944,
      "loss": 0.8813,
      "step": 3341
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.687086284160614,
      "learning_rate": 0.00015415599794017793,
      "loss": 0.9903,
      "step": 3342
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.735305905342102,
      "learning_rate": 0.00015413072743145865,
      "loss": 1.0012,
      "step": 3343
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.800947368144989,
      "learning_rate": 0.0001541054520323346,
      "loss": 0.7199,
      "step": 3344
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7069570422172546,
      "learning_rate": 0.00015408017174508936,
      "loss": 0.9401,
      "step": 3345
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7075111865997314,
      "learning_rate": 0.00015405488657200674,
      "loss": 0.8181,
      "step": 3346
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7980069518089294,
      "learning_rate": 0.00015402959651537125,
      "loss": 1.0082,
      "step": 3347
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7631432414054871,
      "learning_rate": 0.00015400430157746757,
      "loss": 0.9643,
      "step": 3348
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8499820828437805,
      "learning_rate": 0.00015397900176058107,
      "loss": 0.967,
      "step": 3349
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7591486573219299,
      "learning_rate": 0.00015395369706699733,
      "loss": 0.981,
      "step": 3350
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7172725200653076,
      "learning_rate": 0.00015392838749900257,
      "loss": 1.0516,
      "step": 3351
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7486289739608765,
      "learning_rate": 0.00015390307305888335,
      "loss": 0.9728,
      "step": 3352
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7924482822418213,
      "learning_rate": 0.00015387775374892665,
      "loss": 0.7647,
      "step": 3353
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9388603568077087,
      "learning_rate": 0.00015385242957142,
      "loss": 1.1205,
      "step": 3354
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.6928844451904297,
      "learning_rate": 0.0001538271005286512,
      "loss": 0.8248,
      "step": 3355
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7360419034957886,
      "learning_rate": 0.00015380176662290867,
      "loss": 1.0013,
      "step": 3356
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.706224262714386,
      "learning_rate": 0.00015377642785648115,
      "loss": 1.0004,
      "step": 3357
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7525461316108704,
      "learning_rate": 0.00015375108423165785,
      "loss": 1.0026,
      "step": 3358
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.732796847820282,
      "learning_rate": 0.00015372573575072842,
      "loss": 0.9005,
      "step": 3359
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.6761058568954468,
      "learning_rate": 0.00015370038241598294,
      "loss": 0.9114,
      "step": 3360
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7725770473480225,
      "learning_rate": 0.000153675024229712,
      "loss": 0.9592,
      "step": 3361
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7360583543777466,
      "learning_rate": 0.0001536496611942065,
      "loss": 1.0003,
      "step": 3362
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7047659754753113,
      "learning_rate": 0.0001536242933117579,
      "loss": 0.8064,
      "step": 3363
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8691027760505676,
      "learning_rate": 0.000153598920584658,
      "loss": 1.0726,
      "step": 3364
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7493014931678772,
      "learning_rate": 0.00015357354301519908,
      "loss": 0.9652,
      "step": 3365
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.6685104370117188,
      "learning_rate": 0.00015354816060567393,
      "loss": 0.9319,
      "step": 3366
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7175832390785217,
      "learning_rate": 0.00015352277335837561,
      "loss": 1.0431,
      "step": 3367
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7429420948028564,
      "learning_rate": 0.00015349738127559782,
      "loss": 0.8698,
      "step": 3368
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7226133346557617,
      "learning_rate": 0.0001534719843596345,
      "loss": 1.0228,
      "step": 3369
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8891973495483398,
      "learning_rate": 0.0001534465826127801,
      "loss": 0.9278,
      "step": 3370
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7273642420768738,
      "learning_rate": 0.00015342117603732965,
      "loss": 0.9188,
      "step": 3371
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7487305402755737,
      "learning_rate": 0.00015339576463557837,
      "loss": 0.9189,
      "step": 3372
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7615191340446472,
      "learning_rate": 0.00015337034840982213,
      "loss": 1.0485,
      "step": 3373
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8446441292762756,
      "learning_rate": 0.00015334492736235705,
      "loss": 1.0208,
      "step": 3374
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7321614623069763,
      "learning_rate": 0.0001533195014954798,
      "loss": 0.9412,
      "step": 3375
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7862521409988403,
      "learning_rate": 0.00015329407081148753,
      "loss": 0.8936,
      "step": 3376
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8918995261192322,
      "learning_rate": 0.00015326863531267773,
      "loss": 0.978,
      "step": 3377
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.6822420358657837,
      "learning_rate": 0.0001532431950013483,
      "loss": 0.9463,
      "step": 3378
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7396413087844849,
      "learning_rate": 0.00015321774987979767,
      "loss": 0.9282,
      "step": 3379
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7108849883079529,
      "learning_rate": 0.00015319229995032464,
      "loss": 0.8528,
      "step": 3380
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.813988983631134,
      "learning_rate": 0.0001531668452152285,
      "loss": 0.9853,
      "step": 3381
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.717553973197937,
      "learning_rate": 0.00015314138567680896,
      "loss": 0.8755,
      "step": 3382
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.6788370609283447,
      "learning_rate": 0.00015311592133736605,
      "loss": 0.835,
      "step": 3383
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7784591913223267,
      "learning_rate": 0.0001530904521992004,
      "loss": 1.0843,
      "step": 3384
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9008314609527588,
      "learning_rate": 0.00015306497826461302,
      "loss": 0.9313,
      "step": 3385
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8545159101486206,
      "learning_rate": 0.00015303949953590529,
      "loss": 0.9582,
      "step": 3386
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8266891241073608,
      "learning_rate": 0.0001530140160153791,
      "loss": 0.9513,
      "step": 3387
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7323997616767883,
      "learning_rate": 0.00015298852770533668,
      "loss": 0.9853,
      "step": 3388
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7101405262947083,
      "learning_rate": 0.00015296303460808085,
      "loss": 0.9666,
      "step": 3389
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.6381584405899048,
      "learning_rate": 0.0001529375367259147,
      "loss": 0.8627,
      "step": 3390
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8005196452140808,
      "learning_rate": 0.00015291203406114185,
      "loss": 1.1132,
      "step": 3391
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7641215324401855,
      "learning_rate": 0.00015288652661606632,
      "loss": 1.067,
      "step": 3392
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7402655482292175,
      "learning_rate": 0.0001528610143929925,
      "loss": 1.0194,
      "step": 3393
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.088860273361206,
      "learning_rate": 0.0001528354973942254,
      "loss": 0.8712,
      "step": 3394
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7720997333526611,
      "learning_rate": 0.00015280997562207023,
      "loss": 0.9561,
      "step": 3395
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7261041402816772,
      "learning_rate": 0.00015278444907883273,
      "loss": 0.8256,
      "step": 3396
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.6564508676528931,
      "learning_rate": 0.00015275891776681918,
      "loss": 0.9009,
      "step": 3397
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7378062009811401,
      "learning_rate": 0.00015273338168833613,
      "loss": 1.0495,
      "step": 3398
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8059569597244263,
      "learning_rate": 0.0001527078408456906,
      "loss": 0.8445,
      "step": 3399
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.649719774723053,
      "learning_rate": 0.00015268229524119007,
      "loss": 0.8538,
      "step": 3400
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7831336259841919,
      "learning_rate": 0.00015265674487714245,
      "loss": 0.9864,
      "step": 3401
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7160272002220154,
      "learning_rate": 0.0001526311897558561,
      "loss": 0.9808,
      "step": 3402
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7628503441810608,
      "learning_rate": 0.00015260562987963976,
      "loss": 0.9573,
      "step": 3403
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.760503888130188,
      "learning_rate": 0.00015258006525080257,
      "loss": 0.9998,
      "step": 3404
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.6774016618728638,
      "learning_rate": 0.00015255449587165423,
      "loss": 0.9838,
      "step": 3405
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7273324728012085,
      "learning_rate": 0.00015252892174450473,
      "loss": 0.9831,
      "step": 3406
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7090844511985779,
      "learning_rate": 0.00015250334287166453,
      "loss": 1.0137,
      "step": 3407
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7482985258102417,
      "learning_rate": 0.00015247775925544466,
      "loss": 0.8267,
      "step": 3408
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.6855625510215759,
      "learning_rate": 0.0001524521708981563,
      "loss": 0.8884,
      "step": 3409
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.720038890838623,
      "learning_rate": 0.00015242657780211133,
      "loss": 1.009,
      "step": 3410
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8395721912384033,
      "learning_rate": 0.00015240097996962185,
      "loss": 0.8667,
      "step": 3411
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7716408967971802,
      "learning_rate": 0.00015237537740300056,
      "loss": 0.9672,
      "step": 3412
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.6766606569290161,
      "learning_rate": 0.00015234977010456047,
      "loss": 0.8397,
      "step": 3413
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8159101605415344,
      "learning_rate": 0.00015232415807661504,
      "loss": 0.9371,
      "step": 3414
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.797480046749115,
      "learning_rate": 0.00015229854132147816,
      "loss": 1.0038,
      "step": 3415
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7471280097961426,
      "learning_rate": 0.0001522729198414642,
      "loss": 1.0894,
      "step": 3416
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7765840291976929,
      "learning_rate": 0.00015224729363888794,
      "loss": 0.9695,
      "step": 3417
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.674380362033844,
      "learning_rate": 0.0001522216627160645,
      "loss": 1.0249,
      "step": 3418
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7781017422676086,
      "learning_rate": 0.0001521960270753095,
      "loss": 1.0102,
      "step": 3419
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8835538625717163,
      "learning_rate": 0.00015217038671893896,
      "loss": 1.1093,
      "step": 3420
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7141196131706238,
      "learning_rate": 0.00015214474164926945,
      "loss": 0.894,
      "step": 3421
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7608664631843567,
      "learning_rate": 0.00015211909186861773,
      "loss": 0.9484,
      "step": 3422
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7789210677146912,
      "learning_rate": 0.00015209343737930116,
      "loss": 1.1107,
      "step": 3423
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.701990008354187,
      "learning_rate": 0.00015206777818363746,
      "loss": 0.7809,
      "step": 3424
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8689419031143188,
      "learning_rate": 0.00015204211428394482,
      "loss": 1.1252,
      "step": 3425
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.6336739659309387,
      "learning_rate": 0.0001520164456825418,
      "loss": 0.8921,
      "step": 3426
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.6421981453895569,
      "learning_rate": 0.00015199077238174747,
      "loss": 0.8899,
      "step": 3427
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8911098837852478,
      "learning_rate": 0.0001519650943838812,
      "loss": 0.9314,
      "step": 3428
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8368533253669739,
      "learning_rate": 0.00015193941169126286,
      "loss": 0.9066,
      "step": 3429
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.6978997588157654,
      "learning_rate": 0.00015191372430621277,
      "loss": 0.9714,
      "step": 3430
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7743934988975525,
      "learning_rate": 0.0001518880322310516,
      "loss": 1.0117,
      "step": 3431
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7560721635818481,
      "learning_rate": 0.00015186233546810057,
      "loss": 0.9588,
      "step": 3432
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.6989509463310242,
      "learning_rate": 0.0001518366340196811,
      "loss": 0.9755,
      "step": 3433
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7402550578117371,
      "learning_rate": 0.00015181092788811533,
      "loss": 0.9117,
      "step": 3434
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8155473470687866,
      "learning_rate": 0.00015178521707572552,
      "loss": 0.9887,
      "step": 3435
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7658118009567261,
      "learning_rate": 0.00015175950158483455,
      "loss": 0.9122,
      "step": 3436
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7669534087181091,
      "learning_rate": 0.00015173378141776568,
      "loss": 0.9987,
      "step": 3437
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8077985048294067,
      "learning_rate": 0.0001517080565768426,
      "loss": 0.8834,
      "step": 3438
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.717319667339325,
      "learning_rate": 0.00015168232706438937,
      "loss": 0.9681,
      "step": 3439
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.802055299282074,
      "learning_rate": 0.0001516565928827305,
      "loss": 1.1436,
      "step": 3440
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7889139652252197,
      "learning_rate": 0.00015163085403419095,
      "loss": 0.9313,
      "step": 3441
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7818065285682678,
      "learning_rate": 0.00015160511052109604,
      "loss": 1.1377,
      "step": 3442
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7210289239883423,
      "learning_rate": 0.00015157936234577163,
      "loss": 0.8287,
      "step": 3443
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7757744193077087,
      "learning_rate": 0.00015155360951054383,
      "loss": 0.8272,
      "step": 3444
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7870873212814331,
      "learning_rate": 0.0001515278520177393,
      "loss": 0.902,
      "step": 3445
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7317755818367004,
      "learning_rate": 0.00015150208986968515,
      "loss": 1.0121,
      "step": 3446
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7527960538864136,
      "learning_rate": 0.00015147632306870875,
      "loss": 0.9836,
      "step": 3447
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7288510203361511,
      "learning_rate": 0.00015145055161713803,
      "loss": 0.9623,
      "step": 3448
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7237223386764526,
      "learning_rate": 0.00015142477551730128,
      "loss": 1.0682,
      "step": 3449
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.6942408680915833,
      "learning_rate": 0.0001513989947715272,
      "loss": 0.8924,
      "step": 3450
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7724366188049316,
      "learning_rate": 0.000151373209382145,
      "loss": 1.0604,
      "step": 3451
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7728193402290344,
      "learning_rate": 0.0001513474193514842,
      "loss": 0.9177,
      "step": 3452
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8020523190498352,
      "learning_rate": 0.0001513216246818748,
      "loss": 1.1552,
      "step": 3453
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7387242913246155,
      "learning_rate": 0.00015129582537564716,
      "loss": 0.9259,
      "step": 3454
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7924624681472778,
      "learning_rate": 0.00015127002143513213,
      "loss": 0.9954,
      "step": 3455
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7585887312889099,
      "learning_rate": 0.000151244212862661,
      "loss": 0.7944,
      "step": 3456
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7654167413711548,
      "learning_rate": 0.00015121839966056536,
      "loss": 1.0601,
      "step": 3457
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8087634444236755,
      "learning_rate": 0.00015119258183117728,
      "loss": 0.8418,
      "step": 3458
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7145108580589294,
      "learning_rate": 0.00015116675937682934,
      "loss": 0.9512,
      "step": 3459
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.755843997001648,
      "learning_rate": 0.00015114093229985436,
      "loss": 0.9324,
      "step": 3460
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.744723379611969,
      "learning_rate": 0.0001511151006025857,
      "loss": 1.0098,
      "step": 3461
    },
    {
      "epoch": 1.3,
      "grad_norm": NaN,
      "learning_rate": 0.0001511151006025857,
      "loss": 0.9791,
      "step": 3462
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7100603580474854,
      "learning_rate": 0.0001510892642873571,
      "loss": 1.0333,
      "step": 3463
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7299081683158875,
      "learning_rate": 0.0001510634233565028,
      "loss": 0.8616,
      "step": 3464
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7373127937316895,
      "learning_rate": 0.00015103757781235726,
      "loss": 0.9998,
      "step": 3465
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7837493419647217,
      "learning_rate": 0.00015101172765725554,
      "loss": 1.0791,
      "step": 3466
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.722700297832489,
      "learning_rate": 0.00015098587289353308,
      "loss": 0.856,
      "step": 3467
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.6745336055755615,
      "learning_rate": 0.0001509600135235257,
      "loss": 0.8963,
      "step": 3468
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.6792166233062744,
      "learning_rate": 0.0001509341495495696,
      "loss": 0.991,
      "step": 3469
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7579882740974426,
      "learning_rate": 0.0001509082809740015,
      "loss": 0.9236,
      "step": 3470
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.6892275214195251,
      "learning_rate": 0.00015088240779915842,
      "loss": 0.8991,
      "step": 3471
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7705352306365967,
      "learning_rate": 0.0001508565300273779,
      "loss": 1.0342,
      "step": 3472
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.811631977558136,
      "learning_rate": 0.00015083064766099788,
      "loss": 0.8619,
      "step": 3473
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8345324397087097,
      "learning_rate": 0.00015080476070235662,
      "loss": 1.0333,
      "step": 3474
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.6630562543869019,
      "learning_rate": 0.0001507788691537929,
      "loss": 0.9507,
      "step": 3475
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7931245565414429,
      "learning_rate": 0.00015075297301764585,
      "loss": 0.9996,
      "step": 3476
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7066725492477417,
      "learning_rate": 0.00015072707229625505,
      "loss": 0.9775,
      "step": 3477
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7714331746101379,
      "learning_rate": 0.00015070116699196045,
      "loss": 0.972,
      "step": 3478
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.854289174079895,
      "learning_rate": 0.00015067525710710252,
      "loss": 1.0122,
      "step": 3479
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.6653081178665161,
      "learning_rate": 0.00015064934264402203,
      "loss": 0.9286,
      "step": 3480
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.698539137840271,
      "learning_rate": 0.00015062342360506022,
      "loss": 0.8626,
      "step": 3481
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.84158855676651,
      "learning_rate": 0.0001505974999925587,
      "loss": 0.8943,
      "step": 3482
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8350929021835327,
      "learning_rate": 0.00015057157180885952,
      "loss": 0.8051,
      "step": 3483
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8504432439804077,
      "learning_rate": 0.00015054563905630522,
      "loss": 1.1219,
      "step": 3484
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7879734039306641,
      "learning_rate": 0.00015051970173723858,
      "loss": 0.8582,
      "step": 3485
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7625810503959656,
      "learning_rate": 0.0001504937598540029,
      "loss": 1.0164,
      "step": 3486
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7224019765853882,
      "learning_rate": 0.00015046781340894195,
      "loss": 1.0259,
      "step": 3487
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.738791286945343,
      "learning_rate": 0.0001504418624043998,
      "loss": 1.0126,
      "step": 3488
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.6853582262992859,
      "learning_rate": 0.000150415906842721,
      "loss": 0.965,
      "step": 3489
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7412823438644409,
      "learning_rate": 0.00015038994672625047,
      "loss": 0.878,
      "step": 3490
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7869895100593567,
      "learning_rate": 0.0001503639820573335,
      "loss": 0.7748,
      "step": 3491
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8981345891952515,
      "learning_rate": 0.00015033801283831596,
      "loss": 0.8713,
      "step": 3492
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9000715613365173,
      "learning_rate": 0.00015031203907154395,
      "loss": 1.0349,
      "step": 3493
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7573916912078857,
      "learning_rate": 0.00015028606075936414,
      "loss": 1.0724,
      "step": 3494
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.6647499799728394,
      "learning_rate": 0.0001502600779041234,
      "loss": 0.9733,
      "step": 3495
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8138927817344666,
      "learning_rate": 0.00015023409050816922,
      "loss": 1.1055,
      "step": 3496
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7094717621803284,
      "learning_rate": 0.00015020809857384937,
      "loss": 0.9432,
      "step": 3497
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7191406488418579,
      "learning_rate": 0.0001501821021035121,
      "loss": 1.1021,
      "step": 3498
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8311622142791748,
      "learning_rate": 0.00015015610109950602,
      "loss": 0.9388,
      "step": 3499
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7128794193267822,
      "learning_rate": 0.00015013009556418024,
      "loss": 0.8522,
      "step": 3500
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7016494274139404,
      "learning_rate": 0.00015010408549988408,
      "loss": 0.9804,
      "step": 3501
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7902352809906006,
      "learning_rate": 0.00015007807090896752,
      "loss": 0.99,
      "step": 3502
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7388083934783936,
      "learning_rate": 0.00015005205179378082,
      "loss": 1.0261,
      "step": 3503
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8480693101882935,
      "learning_rate": 0.00015002602815667463,
      "loss": 1.006,
      "step": 3504
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7673492431640625,
      "learning_rate": 0.00015000000000000001,
      "loss": 1.026,
      "step": 3505
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8567508459091187,
      "learning_rate": 0.0001499739673261085,
      "loss": 0.8649,
      "step": 3506
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8326045274734497,
      "learning_rate": 0.000149947930137352,
      "loss": 1.001,
      "step": 3507
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7344638705253601,
      "learning_rate": 0.0001499218884360828,
      "loss": 0.9197,
      "step": 3508
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.816310703754425,
      "learning_rate": 0.00014989584222465367,
      "loss": 1.0347,
      "step": 3509
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8038501143455505,
      "learning_rate": 0.0001498697915054177,
      "loss": 1.0826,
      "step": 3510
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7537353038787842,
      "learning_rate": 0.0001498437362807284,
      "loss": 1.0289,
      "step": 3511
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7652748823165894,
      "learning_rate": 0.00014981767655293976,
      "loss": 0.7611,
      "step": 3512
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.732870876789093,
      "learning_rate": 0.00014979161232440613,
      "loss": 1.0268,
      "step": 3513
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.725989818572998,
      "learning_rate": 0.00014976554359748224,
      "loss": 0.9618,
      "step": 3514
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8126868009567261,
      "learning_rate": 0.00014973947037452324,
      "loss": 0.9914,
      "step": 3515
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8361872434616089,
      "learning_rate": 0.0001497133926578847,
      "loss": 1.1645,
      "step": 3516
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8904275298118591,
      "learning_rate": 0.00014968731044992262,
      "loss": 0.9246,
      "step": 3517
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7571832537651062,
      "learning_rate": 0.0001496612237529934,
      "loss": 0.9037,
      "step": 3518
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8950560092926025,
      "learning_rate": 0.00014963513256945377,
      "loss": 1.1312,
      "step": 3519
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7928481101989746,
      "learning_rate": 0.00014960903690166094,
      "loss": 0.9981,
      "step": 3520
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.676289439201355,
      "learning_rate": 0.0001495829367519725,
      "loss": 0.9821,
      "step": 3521
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7975377440452576,
      "learning_rate": 0.0001495568321227465,
      "loss": 1.0974,
      "step": 3522
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7216330766677856,
      "learning_rate": 0.00014953072301634128,
      "loss": 1.0396,
      "step": 3523
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7259735465049744,
      "learning_rate": 0.00014950460943511572,
      "loss": 0.958,
      "step": 3524
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.6637389659881592,
      "learning_rate": 0.00014947849138142894,
      "loss": 0.8469,
      "step": 3525
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.6828493475914001,
      "learning_rate": 0.00014945236885764062,
      "loss": 0.9702,
      "step": 3526
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9553461074829102,
      "learning_rate": 0.0001494262418661108,
      "loss": 1.12,
      "step": 3527
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.6893486976623535,
      "learning_rate": 0.00014940011040919985,
      "loss": 0.8297,
      "step": 3528
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.6896955370903015,
      "learning_rate": 0.00014937397448926864,
      "loss": 0.8091,
      "step": 3529
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.769710898399353,
      "learning_rate": 0.0001493478341086784,
      "loss": 0.9944,
      "step": 3530
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.6986170411109924,
      "learning_rate": 0.00014932168926979074,
      "loss": 0.8848,
      "step": 3531
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7402127981185913,
      "learning_rate": 0.00014929553997496773,
      "loss": 0.9583,
      "step": 3532
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8498134016990662,
      "learning_rate": 0.00014926938622657177,
      "loss": 0.9529,
      "step": 3533
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7263255715370178,
      "learning_rate": 0.00014924322802696574,
      "loss": 0.966,
      "step": 3534
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.6980369091033936,
      "learning_rate": 0.00014921706537851287,
      "loss": 0.9146,
      "step": 3535
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7615999579429626,
      "learning_rate": 0.0001491908982835768,
      "loss": 0.9918,
      "step": 3536
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7459874153137207,
      "learning_rate": 0.0001491647267445216,
      "loss": 0.8988,
      "step": 3537
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8014723062515259,
      "learning_rate": 0.0001491385507637117,
      "loss": 1.0638,
      "step": 3538
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7640689611434937,
      "learning_rate": 0.00014911237034351195,
      "loss": 1.0857,
      "step": 3539
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7542036771774292,
      "learning_rate": 0.00014908618548628767,
      "loss": 0.8638,
      "step": 3540
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8518773317337036,
      "learning_rate": 0.00014905999619440442,
      "loss": 0.9615,
      "step": 3541
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7619187831878662,
      "learning_rate": 0.00014903380247022826,
      "loss": 0.9976,
      "step": 3542
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8794530034065247,
      "learning_rate": 0.0001490076043161257,
      "loss": 0.7971,
      "step": 3543
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8578647375106812,
      "learning_rate": 0.00014898140173446356,
      "loss": 1.0534,
      "step": 3544
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7493592500686646,
      "learning_rate": 0.00014895519472760913,
      "loss": 1.0227,
      "step": 3545
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8536798357963562,
      "learning_rate": 0.00014892898329793003,
      "loss": 0.9422,
      "step": 3546
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7785354256629944,
      "learning_rate": 0.0001489027674477943,
      "loss": 1.0192,
      "step": 3547
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7300075888633728,
      "learning_rate": 0.00014887654717957044,
      "loss": 0.9542,
      "step": 3548
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7428839802742004,
      "learning_rate": 0.0001488503224956273,
      "loss": 1.0701,
      "step": 3549
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7662368416786194,
      "learning_rate": 0.0001488240933983341,
      "loss": 1.0985,
      "step": 3550
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7324904799461365,
      "learning_rate": 0.0001487978598900605,
      "loss": 0.8729,
      "step": 3551
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7264313697814941,
      "learning_rate": 0.00014877162197317654,
      "loss": 0.9747,
      "step": 3552
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7613300085067749,
      "learning_rate": 0.00014874537965005267,
      "loss": 1.1532,
      "step": 3553
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7279550433158875,
      "learning_rate": 0.00014871913292305978,
      "loss": 1.0743,
      "step": 3554
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7823167443275452,
      "learning_rate": 0.00014869288179456905,
      "loss": 0.9377,
      "step": 3555
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.854567289352417,
      "learning_rate": 0.00014866662626695213,
      "loss": 0.9287,
      "step": 3556
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7282840609550476,
      "learning_rate": 0.00014864036634258112,
      "loss": 0.9809,
      "step": 3557
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.6796576976776123,
      "learning_rate": 0.0001486141020238284,
      "loss": 0.803,
      "step": 3558
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7584293484687805,
      "learning_rate": 0.00014858783331306682,
      "loss": 1.0061,
      "step": 3559
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7804273366928101,
      "learning_rate": 0.00014856156021266956,
      "loss": 1.0346,
      "step": 3560
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7761217951774597,
      "learning_rate": 0.00014853528272501033,
      "loss": 0.8941,
      "step": 3561
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7115616798400879,
      "learning_rate": 0.0001485090008524631,
      "loss": 0.9078,
      "step": 3562
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8151125311851501,
      "learning_rate": 0.0001484827145974023,
      "loss": 1.135,
      "step": 3563
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7691593170166016,
      "learning_rate": 0.00014845642396220276,
      "loss": 0.995,
      "step": 3564
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7243972420692444,
      "learning_rate": 0.00014843012894923967,
      "loss": 0.9317,
      "step": 3565
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.714561402797699,
      "learning_rate": 0.00014840382956088862,
      "loss": 1.0863,
      "step": 3566
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7968424558639526,
      "learning_rate": 0.00014837752579952563,
      "loss": 0.9808,
      "step": 3567
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.6983405351638794,
      "learning_rate": 0.00014835121766752714,
      "loss": 0.9849,
      "step": 3568
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.6801846027374268,
      "learning_rate": 0.00014832490516726986,
      "loss": 0.9781,
      "step": 3569
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7692368030548096,
      "learning_rate": 0.000148298588301131,
      "loss": 0.8748,
      "step": 3570
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.885966956615448,
      "learning_rate": 0.00014827226707148818,
      "loss": 1.1406,
      "step": 3571
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.6834810972213745,
      "learning_rate": 0.00014824594148071934,
      "loss": 0.8532,
      "step": 3572
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8407056331634521,
      "learning_rate": 0.00014821961153120287,
      "loss": 0.923,
      "step": 3573
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8173869848251343,
      "learning_rate": 0.00014819327722531749,
      "loss": 0.9967,
      "step": 3574
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8253581523895264,
      "learning_rate": 0.0001481669385654424,
      "loss": 1.1102,
      "step": 3575
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7063455581665039,
      "learning_rate": 0.00014814059555395712,
      "loss": 0.9345,
      "step": 3576
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7028665542602539,
      "learning_rate": 0.0001481142481932416,
      "loss": 0.9175,
      "step": 3577
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7196552753448486,
      "learning_rate": 0.0001480878964856762,
      "loss": 0.9596,
      "step": 3578
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9650554656982422,
      "learning_rate": 0.0001480615404336416,
      "loss": 1.0165,
      "step": 3579
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7212446928024292,
      "learning_rate": 0.00014803518003951897,
      "loss": 0.9302,
      "step": 3580
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8186121582984924,
      "learning_rate": 0.00014800881530568978,
      "loss": 0.7887,
      "step": 3581
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7647170424461365,
      "learning_rate": 0.00014798244623453596,
      "loss": 1.0245,
      "step": 3582
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7086816430091858,
      "learning_rate": 0.0001479560728284398,
      "loss": 0.969,
      "step": 3583
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7693114280700684,
      "learning_rate": 0.00014792969508978398,
      "loss": 1.0047,
      "step": 3584
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7128672003746033,
      "learning_rate": 0.0001479033130209516,
      "loss": 0.7372,
      "step": 3585
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7224681973457336,
      "learning_rate": 0.00014787692662432612,
      "loss": 0.9198,
      "step": 3586
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7356724739074707,
      "learning_rate": 0.00014785053590229134,
      "loss": 0.812,
      "step": 3587
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7544524669647217,
      "learning_rate": 0.00014782414085723165,
      "loss": 0.9964,
      "step": 3588
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7893449664115906,
      "learning_rate": 0.00014779774149153155,
      "loss": 1.086,
      "step": 3589
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7163593769073486,
      "learning_rate": 0.00014777133780757617,
      "loss": 0.9591,
      "step": 3590
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7546105980873108,
      "learning_rate": 0.0001477449298077509,
      "loss": 0.9707,
      "step": 3591
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7793951630592346,
      "learning_rate": 0.00014771851749444156,
      "loss": 1.0296,
      "step": 3592
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7393033504486084,
      "learning_rate": 0.0001476921008700343,
      "loss": 0.9886,
      "step": 3593
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7126401662826538,
      "learning_rate": 0.00014766567993691583,
      "loss": 0.9804,
      "step": 3594
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.6830233931541443,
      "learning_rate": 0.000147639254697473,
      "loss": 0.8706,
      "step": 3595
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8743071556091309,
      "learning_rate": 0.00014761282515409326,
      "loss": 0.9761,
      "step": 3596
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7935216426849365,
      "learning_rate": 0.00014758639130916436,
      "loss": 1.0794,
      "step": 3597
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8059981465339661,
      "learning_rate": 0.00014755995316507442,
      "loss": 1.0446,
      "step": 3598
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7111148834228516,
      "learning_rate": 0.000147533510724212,
      "loss": 0.9263,
      "step": 3599
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7591990828514099,
      "learning_rate": 0.00014750706398896604,
      "loss": 1.0087,
      "step": 3600
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7582071423530579,
      "learning_rate": 0.0001474806129617258,
      "loss": 0.8304,
      "step": 3601
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8318824172019958,
      "learning_rate": 0.00014745415764488105,
      "loss": 0.9744,
      "step": 3602
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.6948772072792053,
      "learning_rate": 0.00014742769804082178,
      "loss": 1.0361,
      "step": 3603
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7453570365905762,
      "learning_rate": 0.0001474012341519386,
      "loss": 0.8178,
      "step": 3604
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7137439250946045,
      "learning_rate": 0.0001473747659806223,
      "loss": 0.9621,
      "step": 3605
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.860538899898529,
      "learning_rate": 0.00014734829352926405,
      "loss": 1.0802,
      "step": 3606
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7084441781044006,
      "learning_rate": 0.00014732181680025564,
      "loss": 0.9248,
      "step": 3607
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.6707698702812195,
      "learning_rate": 0.000147295335795989,
      "loss": 0.8148,
      "step": 3608
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.6666585803031921,
      "learning_rate": 0.00014726885051885653,
      "loss": 0.9949,
      "step": 3609
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7207210659980774,
      "learning_rate": 0.00014724236097125108,
      "loss": 0.964,
      "step": 3610
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.838354229927063,
      "learning_rate": 0.0001472158671555658,
      "loss": 0.9344,
      "step": 3611
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7047592401504517,
      "learning_rate": 0.00014718936907419424,
      "loss": 0.7781,
      "step": 3612
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8515235781669617,
      "learning_rate": 0.00014716286672953038,
      "loss": 0.9691,
      "step": 3613
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.6779474020004272,
      "learning_rate": 0.00014713636012396854,
      "loss": 0.8669,
      "step": 3614
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7808057069778442,
      "learning_rate": 0.00014710984925990345,
      "loss": 0.9782,
      "step": 3615
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.6925843954086304,
      "learning_rate": 0.0001470833341397302,
      "loss": 0.8911,
      "step": 3616
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7705125212669373,
      "learning_rate": 0.00014705681476584432,
      "loss": 1.0675,
      "step": 3617
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8819014430046082,
      "learning_rate": 0.00014703029114064166,
      "loss": 0.9161,
      "step": 3618
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7916482090950012,
      "learning_rate": 0.00014700376326651847,
      "loss": 0.8913,
      "step": 3619
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7155230641365051,
      "learning_rate": 0.00014697723114587142,
      "loss": 0.9563,
      "step": 3620
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7341004610061646,
      "learning_rate": 0.0001469506947810975,
      "loss": 1.0403,
      "step": 3621
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7908432483673096,
      "learning_rate": 0.00014692415417459414,
      "loss": 0.9515,
      "step": 3622
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7485032081604004,
      "learning_rate": 0.00014689760932875916,
      "loss": 0.9666,
      "step": 3623
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8104244470596313,
      "learning_rate": 0.00014687106024599067,
      "loss": 1.049,
      "step": 3624
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.6753594279289246,
      "learning_rate": 0.00014684450692868732,
      "loss": 0.9839,
      "step": 3625
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9177693724632263,
      "learning_rate": 0.00014681794937924795,
      "loss": 1.0646,
      "step": 3626
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7891800999641418,
      "learning_rate": 0.00014679138760007198,
      "loss": 0.9064,
      "step": 3627
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.728115439414978,
      "learning_rate": 0.00014676482159355905,
      "loss": 1.0293,
      "step": 3628
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7733248472213745,
      "learning_rate": 0.00014673825136210929,
      "loss": 0.9834,
      "step": 3629
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7536121606826782,
      "learning_rate": 0.0001467116769081231,
      "loss": 0.8856,
      "step": 3630
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7316053509712219,
      "learning_rate": 0.00014668509823400144,
      "loss": 0.9515,
      "step": 3631
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.6929289698600769,
      "learning_rate": 0.00014665851534214545,
      "loss": 1.1022,
      "step": 3632
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8146265745162964,
      "learning_rate": 0.0001466319282349568,
      "loss": 0.9602,
      "step": 3633
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7151821851730347,
      "learning_rate": 0.00014660533691483746,
      "loss": 0.995,
      "step": 3634
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7196069955825806,
      "learning_rate": 0.0001465787413841898,
      "loss": 0.981,
      "step": 3635
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0193686485290527,
      "learning_rate": 0.00014655214164541658,
      "loss": 1.1679,
      "step": 3636
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.6898132562637329,
      "learning_rate": 0.00014652553770092095,
      "loss": 0.9004,
      "step": 3637
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7551093697547913,
      "learning_rate": 0.0001464989295531064,
      "loss": 1.0437,
      "step": 3638
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7888858914375305,
      "learning_rate": 0.00014647231720437686,
      "loss": 0.9664,
      "step": 3639
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7758256196975708,
      "learning_rate": 0.0001464457006571366,
      "loss": 0.864,
      "step": 3640
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7542037963867188,
      "learning_rate": 0.0001464190799137902,
      "loss": 1.0199,
      "step": 3641
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7568319439888,
      "learning_rate": 0.00014639245497674282,
      "loss": 0.9234,
      "step": 3642
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.776556134223938,
      "learning_rate": 0.00014636582584839978,
      "loss": 1.0185,
      "step": 3643
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7782880067825317,
      "learning_rate": 0.0001463391925311669,
      "loss": 1.205,
      "step": 3644
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8340494632720947,
      "learning_rate": 0.00014631255502745036,
      "loss": 1.0501,
      "step": 3645
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7521584630012512,
      "learning_rate": 0.00014628591333965663,
      "loss": 1.0301,
      "step": 3646
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.80910724401474,
      "learning_rate": 0.00014625926747019276,
      "loss": 1.1471,
      "step": 3647
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7709221243858337,
      "learning_rate": 0.00014623261742146602,
      "loss": 0.9235,
      "step": 3648
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8370350003242493,
      "learning_rate": 0.000146205963195884,
      "loss": 0.8984,
      "step": 3649
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8251968026161194,
      "learning_rate": 0.00014617930479585487,
      "loss": 1.0234,
      "step": 3650
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8307152986526489,
      "learning_rate": 0.00014615264222378696,
      "loss": 1.0159,
      "step": 3651
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8258378505706787,
      "learning_rate": 0.0001461259754820892,
      "loss": 1.1309,
      "step": 3652
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.661595344543457,
      "learning_rate": 0.0001460993045731707,
      "loss": 0.7346,
      "step": 3653
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8245034217834473,
      "learning_rate": 0.00014607262949944104,
      "loss": 1.0182,
      "step": 3654
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7197308540344238,
      "learning_rate": 0.00014604595026331015,
      "loss": 0.9962,
      "step": 3655
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8030989170074463,
      "learning_rate": 0.0001460192668671884,
      "loss": 0.9241,
      "step": 3656
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8018397688865662,
      "learning_rate": 0.00014599257931348643,
      "loss": 0.9935,
      "step": 3657
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7212077975273132,
      "learning_rate": 0.00014596588760461534,
      "loss": 0.8948,
      "step": 3658
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7394177317619324,
      "learning_rate": 0.00014593919174298655,
      "loss": 1.0604,
      "step": 3659
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7774109244346619,
      "learning_rate": 0.00014591249173101192,
      "loss": 0.8841,
      "step": 3660
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.781120240688324,
      "learning_rate": 0.00014588578757110358,
      "loss": 1.0129,
      "step": 3661
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8257413506507874,
      "learning_rate": 0.00014585907926567416,
      "loss": 1.0161,
      "step": 3662
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7354034185409546,
      "learning_rate": 0.00014583236681713657,
      "loss": 0.9752,
      "step": 3663
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7330094575881958,
      "learning_rate": 0.00014580565022790414,
      "loss": 0.9434,
      "step": 3664
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7110805511474609,
      "learning_rate": 0.00014577892950039058,
      "loss": 1.0484,
      "step": 3665
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8536974191665649,
      "learning_rate": 0.00014575220463700992,
      "loss": 0.903,
      "step": 3666
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7048803567886353,
      "learning_rate": 0.00014572547564017664,
      "loss": 0.8204,
      "step": 3667
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7898951172828674,
      "learning_rate": 0.0001456987425123055,
      "loss": 0.9791,
      "step": 3668
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8470849394798279,
      "learning_rate": 0.00014567200525581175,
      "loss": 1.0017,
      "step": 3669
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8138737082481384,
      "learning_rate": 0.0001456452638731109,
      "loss": 0.9451,
      "step": 3670
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7108284831047058,
      "learning_rate": 0.00014561851836661892,
      "loss": 0.83,
      "step": 3671
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8926004767417908,
      "learning_rate": 0.00014559176873875208,
      "loss": 0.8786,
      "step": 3672
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7158498764038086,
      "learning_rate": 0.00014556501499192709,
      "loss": 1.097,
      "step": 3673
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7436313033103943,
      "learning_rate": 0.00014553825712856097,
      "loss": 0.8637,
      "step": 3674
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.723731517791748,
      "learning_rate": 0.00014551149515107116,
      "loss": 0.947,
      "step": 3675
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.784081757068634,
      "learning_rate": 0.00014548472906187544,
      "loss": 0.9916,
      "step": 3676
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.789717435836792,
      "learning_rate": 0.00014545795886339202,
      "loss": 1.1036,
      "step": 3677
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7132088541984558,
      "learning_rate": 0.00014543118455803935,
      "loss": 0.9898,
      "step": 3678
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0612138509750366,
      "learning_rate": 0.00014540440614823643,
      "loss": 1.0354,
      "step": 3679
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7618119120597839,
      "learning_rate": 0.00014537762363640248,
      "loss": 1.0126,
      "step": 3680
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.6965414881706238,
      "learning_rate": 0.00014535083702495714,
      "loss": 1.0952,
      "step": 3681
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7453968524932861,
      "learning_rate": 0.00014532404631632047,
      "loss": 1.3152,
      "step": 3682
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8747276663780212,
      "learning_rate": 0.00014529725151291283,
      "loss": 1.0533,
      "step": 3683
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7281939387321472,
      "learning_rate": 0.000145270452617155,
      "loss": 0.9864,
      "step": 3684
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7459536790847778,
      "learning_rate": 0.0001452436496314681,
      "loss": 1.047,
      "step": 3685
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8038800358772278,
      "learning_rate": 0.0001452168425582736,
      "loss": 0.965,
      "step": 3686
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8129239678382874,
      "learning_rate": 0.00014519003139999337,
      "loss": 0.8938,
      "step": 3687
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.784221887588501,
      "learning_rate": 0.00014516321615904972,
      "loss": 1.0522,
      "step": 3688
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7523348331451416,
      "learning_rate": 0.0001451363968378652,
      "loss": 0.9661,
      "step": 3689
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7557814717292786,
      "learning_rate": 0.00014510957343886278,
      "loss": 0.9508,
      "step": 3690
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7756308913230896,
      "learning_rate": 0.00014508274596446578,
      "loss": 0.9669,
      "step": 3691
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.799447774887085,
      "learning_rate": 0.00014505591441709795,
      "loss": 0.9815,
      "step": 3692
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7798731327056885,
      "learning_rate": 0.0001450290787991834,
      "loss": 1.0101,
      "step": 3693
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7313754558563232,
      "learning_rate": 0.0001450022391131465,
      "loss": 0.8448,
      "step": 3694
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7849476933479309,
      "learning_rate": 0.0001449753953614121,
      "loss": 1.0423,
      "step": 3695
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8375855088233948,
      "learning_rate": 0.0001449485475464054,
      "loss": 1.1203,
      "step": 3696
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.79655522108078,
      "learning_rate": 0.0001449216956705519,
      "loss": 1.0499,
      "step": 3697
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7769049406051636,
      "learning_rate": 0.00014489483973627754,
      "loss": 1.0213,
      "step": 3698
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7109012603759766,
      "learning_rate": 0.00014486797974600865,
      "loss": 0.9309,
      "step": 3699
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7079145908355713,
      "learning_rate": 0.0001448411157021718,
      "loss": 0.8,
      "step": 3700
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7436159253120422,
      "learning_rate": 0.000144814247607194,
      "loss": 1.0672,
      "step": 3701
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7641329169273376,
      "learning_rate": 0.00014478737546350272,
      "loss": 0.9365,
      "step": 3702
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9353510737419128,
      "learning_rate": 0.00014476049927352564,
      "loss": 1.0433,
      "step": 3703
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.757537305355072,
      "learning_rate": 0.00014473361903969087,
      "loss": 1.058,
      "step": 3704
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.6940644383430481,
      "learning_rate": 0.0001447067347644269,
      "loss": 0.8961,
      "step": 3705
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7819743752479553,
      "learning_rate": 0.00014467984645016258,
      "loss": 0.9163,
      "step": 3706
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7648773193359375,
      "learning_rate": 0.0001446529540993271,
      "loss": 0.8864,
      "step": 3707
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.770237386226654,
      "learning_rate": 0.00014462605771435008,
      "loss": 0.8647,
      "step": 3708
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7355459332466125,
      "learning_rate": 0.0001445991572976614,
      "loss": 0.9652,
      "step": 3709
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7374024391174316,
      "learning_rate": 0.00014457225285169134,
      "loss": 0.906,
      "step": 3710
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8194604516029358,
      "learning_rate": 0.00014454534437887063,
      "loss": 1.0381,
      "step": 3711
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.762771487236023,
      "learning_rate": 0.00014451843188163029,
      "loss": 0.9989,
      "step": 3712
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.881404459476471,
      "learning_rate": 0.00014449151536240166,
      "loss": 0.9936,
      "step": 3713
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8181387782096863,
      "learning_rate": 0.00014446459482361654,
      "loss": 0.9665,
      "step": 3714
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7393097877502441,
      "learning_rate": 0.00014443767026770706,
      "loss": 0.9013,
      "step": 3715
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7258290648460388,
      "learning_rate": 0.00014441074169710562,
      "loss": 0.94,
      "step": 3716
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.738956868648529,
      "learning_rate": 0.00014438380911424516,
      "loss": 0.8634,
      "step": 3717
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8704954981803894,
      "learning_rate": 0.00014435687252155885,
      "loss": 1.1659,
      "step": 3718
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7542216777801514,
      "learning_rate": 0.00014432993192148023,
      "loss": 0.9391,
      "step": 3719
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.6866111755371094,
      "learning_rate": 0.0001443029873164433,
      "loss": 0.8937,
      "step": 3720
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.715201735496521,
      "learning_rate": 0.00014427603870888225,
      "loss": 0.9683,
      "step": 3721
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9084267020225525,
      "learning_rate": 0.00014424908610123182,
      "loss": 0.9476,
      "step": 3722
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7752277851104736,
      "learning_rate": 0.000144222129495927,
      "loss": 0.8479,
      "step": 3723
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7288671731948853,
      "learning_rate": 0.00014419516889540315,
      "loss": 0.9883,
      "step": 3724
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8041513562202454,
      "learning_rate": 0.00014416820430209607,
      "loss": 0.9846,
      "step": 3725
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7570642828941345,
      "learning_rate": 0.00014414123571844178,
      "loss": 0.8865,
      "step": 3726
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8212285041809082,
      "learning_rate": 0.00014411426314687678,
      "loss": 1.0912,
      "step": 3727
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7754359245300293,
      "learning_rate": 0.00014408728658983784,
      "loss": 1.1325,
      "step": 3728
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8131769299507141,
      "learning_rate": 0.00014406030604976222,
      "loss": 1.0475,
      "step": 3729
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.6917926073074341,
      "learning_rate": 0.0001440333215290874,
      "loss": 0.8188,
      "step": 3730
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7929239273071289,
      "learning_rate": 0.00014400633303025131,
      "loss": 1.0119,
      "step": 3731
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7956200838088989,
      "learning_rate": 0.00014397934055569218,
      "loss": 1.1565,
      "step": 3732
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9268775582313538,
      "learning_rate": 0.00014395234410784867,
      "loss": 0.9625,
      "step": 3733
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7056459188461304,
      "learning_rate": 0.00014392534368915972,
      "loss": 0.9952,
      "step": 3734
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.6934985518455505,
      "learning_rate": 0.00014389833930206463,
      "loss": 0.9829,
      "step": 3735
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7250593304634094,
      "learning_rate": 0.00014387133094900315,
      "loss": 0.9692,
      "step": 3736
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7265465259552002,
      "learning_rate": 0.0001438443186324153,
      "loss": 1.033,
      "step": 3737
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7638846039772034,
      "learning_rate": 0.00014381730235474155,
      "loss": 0.95,
      "step": 3738
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8740837574005127,
      "learning_rate": 0.00014379028211842265,
      "loss": 1.1423,
      "step": 3739
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7578192949295044,
      "learning_rate": 0.00014376325792589964,
      "loss": 1.0836,
      "step": 3740
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7851478457450867,
      "learning_rate": 0.00014373622977961409,
      "loss": 1.0269,
      "step": 3741
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7320329546928406,
      "learning_rate": 0.0001437091976820078,
      "loss": 0.8852,
      "step": 3742
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7559787631034851,
      "learning_rate": 0.00014368216163552298,
      "loss": 1.0108,
      "step": 3743
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7940683960914612,
      "learning_rate": 0.0001436551216426022,
      "loss": 1.0229,
      "step": 3744
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7840447425842285,
      "learning_rate": 0.00014362807770568835,
      "loss": 1.049,
      "step": 3745
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.6953481435775757,
      "learning_rate": 0.0001436010298272247,
      "loss": 0.9457,
      "step": 3746
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7428427934646606,
      "learning_rate": 0.00014357397800965487,
      "loss": 0.7076,
      "step": 3747
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7457131147384644,
      "learning_rate": 0.00014354692225542284,
      "loss": 0.9413,
      "step": 3748
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.6950200200080872,
      "learning_rate": 0.00014351986256697297,
      "loss": 0.7335,
      "step": 3749
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7549350261688232,
      "learning_rate": 0.0001434927989467499,
      "loss": 0.9044,
      "step": 3750
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7664708495140076,
      "learning_rate": 0.00014346573139719867,
      "loss": 1.1704,
      "step": 3751
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.6717768907546997,
      "learning_rate": 0.00014343865992076478,
      "loss": 0.8295,
      "step": 3752
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.701734721660614,
      "learning_rate": 0.00014341158451989386,
      "loss": 0.9549,
      "step": 3753
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.932205080986023,
      "learning_rate": 0.00014338450519703214,
      "loss": 0.926,
      "step": 3754
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9582824110984802,
      "learning_rate": 0.00014335742195462596,
      "loss": 1.0883,
      "step": 3755
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7160861492156982,
      "learning_rate": 0.00014333033479512218,
      "loss": 0.9851,
      "step": 3756
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7445812821388245,
      "learning_rate": 0.00014330324372096803,
      "loss": 0.8497,
      "step": 3757
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.6866586804389954,
      "learning_rate": 0.000143276148734611,
      "loss": 0.9647,
      "step": 3758
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8147525191307068,
      "learning_rate": 0.0001432490498384989,
      "loss": 1.0837,
      "step": 3759
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7915096879005432,
      "learning_rate": 0.0001432219470350801,
      "loss": 0.975,
      "step": 3760
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7530489563941956,
      "learning_rate": 0.00014319484032680307,
      "loss": 0.7848,
      "step": 3761
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7870637774467468,
      "learning_rate": 0.00014316772971611675,
      "loss": 0.7446,
      "step": 3762
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7144870758056641,
      "learning_rate": 0.00014314061520547054,
      "loss": 1.0712,
      "step": 3763
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8082242608070374,
      "learning_rate": 0.00014311349679731395,
      "loss": 0.9206,
      "step": 3764
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7679134607315063,
      "learning_rate": 0.00014308637449409706,
      "loss": 0.909,
      "step": 3765
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7546840310096741,
      "learning_rate": 0.00014305924829827017,
      "loss": 1.0257,
      "step": 3766
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7072516679763794,
      "learning_rate": 0.000143032118212284,
      "loss": 1.0949,
      "step": 3767
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9368565082550049,
      "learning_rate": 0.00014300498423858962,
      "loss": 0.8724,
      "step": 3768
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7789929509162903,
      "learning_rate": 0.00014297784637963838,
      "loss": 1.0801,
      "step": 3769
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8462679982185364,
      "learning_rate": 0.00014295070463788207,
      "loss": 0.9488,
      "step": 3770
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7308972477912903,
      "learning_rate": 0.00014292355901577276,
      "loss": 1.1541,
      "step": 3771
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7259547710418701,
      "learning_rate": 0.00014289640951576296,
      "loss": 0.9442,
      "step": 3772
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7557168006896973,
      "learning_rate": 0.00014286925614030542,
      "loss": 1.1086,
      "step": 3773
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7561401128768921,
      "learning_rate": 0.00014284209889185335,
      "loss": 1.1196,
      "step": 3774
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8499369621276855,
      "learning_rate": 0.00014281493777286016,
      "loss": 0.8272,
      "step": 3775
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.6794196367263794,
      "learning_rate": 0.00014278777278577977,
      "loss": 1.0419,
      "step": 3776
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7229602336883545,
      "learning_rate": 0.0001427606039330664,
      "loss": 0.8417,
      "step": 3777
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7260411977767944,
      "learning_rate": 0.00014273343121717457,
      "loss": 0.9832,
      "step": 3778
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9786548614501953,
      "learning_rate": 0.00014270625464055917,
      "loss": 1.0393,
      "step": 3779
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.816220223903656,
      "learning_rate": 0.00014267907420567548,
      "loss": 0.9016,
      "step": 3780
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.738017201423645,
      "learning_rate": 0.00014265188991497907,
      "loss": 1.0055,
      "step": 3781
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7251632213592529,
      "learning_rate": 0.0001426247017709259,
      "loss": 0.9767,
      "step": 3782
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7959434986114502,
      "learning_rate": 0.00014259750977597228,
      "loss": 1.03,
      "step": 3783
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7960846424102783,
      "learning_rate": 0.00014257031393257485,
      "loss": 1.0113,
      "step": 3784
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8744489550590515,
      "learning_rate": 0.00014254311424319056,
      "loss": 1.0178,
      "step": 3785
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8908901214599609,
      "learning_rate": 0.0001425159107102768,
      "loss": 0.9478,
      "step": 3786
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7419928312301636,
      "learning_rate": 0.0001424887033362912,
      "loss": 0.8289,
      "step": 3787
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7056096792221069,
      "learning_rate": 0.00014246149212369184,
      "loss": 0.9822,
      "step": 3788
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8795596361160278,
      "learning_rate": 0.0001424342770749371,
      "loss": 1.1753,
      "step": 3789
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8885317444801331,
      "learning_rate": 0.00014240705819248568,
      "loss": 1.1536,
      "step": 3790
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7233941555023193,
      "learning_rate": 0.00014237983547879662,
      "loss": 0.9257,
      "step": 3791
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.823823869228363,
      "learning_rate": 0.00014235260893632944,
      "loss": 1.181,
      "step": 3792
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8890931606292725,
      "learning_rate": 0.00014232537856754383,
      "loss": 0.9468,
      "step": 3793
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7318471074104309,
      "learning_rate": 0.00014229814437489988,
      "loss": 0.9542,
      "step": 3794
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9147783517837524,
      "learning_rate": 0.0001422709063608581,
      "loss": 1.0679,
      "step": 3795
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.813987672328949,
      "learning_rate": 0.00014224366452787924,
      "loss": 1.0211,
      "step": 3796
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.6452633738517761,
      "learning_rate": 0.0001422164188784245,
      "loss": 0.8243,
      "step": 3797
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7558382153511047,
      "learning_rate": 0.00014218916941495535,
      "loss": 0.8988,
      "step": 3798
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7191858887672424,
      "learning_rate": 0.0001421619161399336,
      "loss": 0.8812,
      "step": 3799
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7513072490692139,
      "learning_rate": 0.00014213465905582145,
      "loss": 1.0562,
      "step": 3800
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.671654999256134,
      "learning_rate": 0.00014210739816508143,
      "loss": 0.9391,
      "step": 3801
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7372080683708191,
      "learning_rate": 0.00014208013347017637,
      "loss": 0.9444,
      "step": 3802
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8163162469863892,
      "learning_rate": 0.00014205286497356956,
      "loss": 1.0991,
      "step": 3803
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8253856301307678,
      "learning_rate": 0.00014202559267772444,
      "loss": 0.9762,
      "step": 3804
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7247931361198425,
      "learning_rate": 0.00014199831658510503,
      "loss": 0.9654,
      "step": 3805
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.76895672082901,
      "learning_rate": 0.00014197103669817546,
      "loss": 1.1187,
      "step": 3806
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.6907783150672913,
      "learning_rate": 0.00014194375301940037,
      "loss": 1.0392,
      "step": 3807
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7909299731254578,
      "learning_rate": 0.0001419164655512447,
      "loss": 0.9487,
      "step": 3808
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.804953396320343,
      "learning_rate": 0.0001418891742961737,
      "loss": 0.9633,
      "step": 3809
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7446818947792053,
      "learning_rate": 0.00014186187925665296,
      "loss": 0.8268,
      "step": 3810
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7549101114273071,
      "learning_rate": 0.00014183458043514848,
      "loss": 1.0275,
      "step": 3811
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.6599337458610535,
      "learning_rate": 0.00014180727783412648,
      "loss": 0.974,
      "step": 3812
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8352928757667542,
      "learning_rate": 0.00014177997145605367,
      "loss": 1.1111,
      "step": 3813
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7139683961868286,
      "learning_rate": 0.000141752661303397,
      "loss": 1.1222,
      "step": 3814
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8132135272026062,
      "learning_rate": 0.0001417253473786238,
      "loss": 0.9324,
      "step": 3815
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7724325060844421,
      "learning_rate": 0.00014169802968420168,
      "loss": 0.9621,
      "step": 3816
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.903314471244812,
      "learning_rate": 0.00014167070822259867,
      "loss": 0.9272,
      "step": 3817
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.809247612953186,
      "learning_rate": 0.00014164338299628312,
      "loss": 0.8033,
      "step": 3818
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8909575343132019,
      "learning_rate": 0.0001416160540077237,
      "loss": 1.0189,
      "step": 3819
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.689109206199646,
      "learning_rate": 0.00014158872125938943,
      "loss": 0.96,
      "step": 3820
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8286650776863098,
      "learning_rate": 0.00014156138475374968,
      "loss": 1.1713,
      "step": 3821
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7234490513801575,
      "learning_rate": 0.00014153404449327413,
      "loss": 0.8662,
      "step": 3822
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.6873946785926819,
      "learning_rate": 0.0001415067004804328,
      "loss": 0.8921,
      "step": 3823
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8717650175094604,
      "learning_rate": 0.00014147935271769613,
      "loss": 1.0844,
      "step": 3824
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9551481008529663,
      "learning_rate": 0.00014145200120753478,
      "loss": 0.9103,
      "step": 3825
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8124105930328369,
      "learning_rate": 0.00014142464595241983,
      "loss": 1.1163,
      "step": 3826
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7727875709533691,
      "learning_rate": 0.00014139728695482267,
      "loss": 0.8512,
      "step": 3827
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.707558274269104,
      "learning_rate": 0.000141369924217215,
      "loss": 1.0252,
      "step": 3828
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7345936298370361,
      "learning_rate": 0.00014134255774206892,
      "loss": 0.9271,
      "step": 3829
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.781466543674469,
      "learning_rate": 0.00014131518753185684,
      "loss": 0.9791,
      "step": 3830
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8418498039245605,
      "learning_rate": 0.00014128781358905147,
      "loss": 0.9273,
      "step": 3831
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7959011793136597,
      "learning_rate": 0.00014126043591612594,
      "loss": 0.9947,
      "step": 3832
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7734252214431763,
      "learning_rate": 0.00014123305451555363,
      "loss": 0.9747,
      "step": 3833
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7216122150421143,
      "learning_rate": 0.0001412056693898083,
      "loss": 0.968,
      "step": 3834
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7466599345207214,
      "learning_rate": 0.0001411782805413641,
      "loss": 0.8883,
      "step": 3835
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8146680593490601,
      "learning_rate": 0.0001411508879726953,
      "loss": 1.0624,
      "step": 3836
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.6900213360786438,
      "learning_rate": 0.00014112349168627684,
      "loss": 0.9781,
      "step": 3837
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.6831471920013428,
      "learning_rate": 0.00014109609168458373,
      "loss": 0.9465,
      "step": 3838
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7674217224121094,
      "learning_rate": 0.00014106868797009143,
      "loss": 0.9436,
      "step": 3839
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7712320685386658,
      "learning_rate": 0.0001410412805452757,
      "loss": 0.9189,
      "step": 3840
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.806429386138916,
      "learning_rate": 0.00014101386941261262,
      "loss": 0.8869,
      "step": 3841
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7063391208648682,
      "learning_rate": 0.0001409864545745787,
      "loss": 0.9804,
      "step": 3842
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.841386616230011,
      "learning_rate": 0.00014095903603365066,
      "loss": 1.1257,
      "step": 3843
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9420889616012573,
      "learning_rate": 0.0001409316137923056,
      "loss": 1.0279,
      "step": 3844
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9218207597732544,
      "learning_rate": 0.00014090418785302103,
      "loss": 1.1393,
      "step": 3845
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8026069402694702,
      "learning_rate": 0.00014087675821827464,
      "loss": 0.9735,
      "step": 3846
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7738460898399353,
      "learning_rate": 0.0001408493248905446,
      "loss": 0.9638,
      "step": 3847
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.6933304071426392,
      "learning_rate": 0.00014082188787230938,
      "loss": 0.8385,
      "step": 3848
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7282217741012573,
      "learning_rate": 0.0001407944471660477,
      "loss": 1.0095,
      "step": 3849
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7486565113067627,
      "learning_rate": 0.00014076700277423869,
      "loss": 0.9736,
      "step": 3850
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8711503148078918,
      "learning_rate": 0.0001407395546993618,
      "loss": 1.0308,
      "step": 3851
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7639390826225281,
      "learning_rate": 0.00014071210294389682,
      "loss": 1.0158,
      "step": 3852
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7631992101669312,
      "learning_rate": 0.00014068464751032384,
      "loss": 1.0516,
      "step": 3853
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7923541069030762,
      "learning_rate": 0.00014065718840112335,
      "loss": 0.9994,
      "step": 3854
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7509554028511047,
      "learning_rate": 0.00014062972561877603,
      "loss": 0.9913,
      "step": 3855
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7766771912574768,
      "learning_rate": 0.00014060225916576305,
      "loss": 0.984,
      "step": 3856
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7930575609207153,
      "learning_rate": 0.0001405747890445659,
      "loss": 0.9921,
      "step": 3857
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7377687692642212,
      "learning_rate": 0.00014054731525766628,
      "loss": 0.9928,
      "step": 3858
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8305283188819885,
      "learning_rate": 0.00014051983780754627,
      "loss": 1.0663,
      "step": 3859
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.838887095451355,
      "learning_rate": 0.00014049235669668834,
      "loss": 1.049,
      "step": 3860
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7463981509208679,
      "learning_rate": 0.00014046487192757528,
      "loss": 0.8271,
      "step": 3861
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7429585456848145,
      "learning_rate": 0.00014043738350269015,
      "loss": 1.0368,
      "step": 3862
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7435213327407837,
      "learning_rate": 0.00014040989142451632,
      "loss": 0.9739,
      "step": 3863
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.6512870192527771,
      "learning_rate": 0.00014038239569553765,
      "loss": 0.7547,
      "step": 3864
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7357031106948853,
      "learning_rate": 0.0001403548963182382,
      "loss": 0.8476,
      "step": 3865
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7592230439186096,
      "learning_rate": 0.00014032739329510228,
      "loss": 1.045,
      "step": 3866
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.751270592212677,
      "learning_rate": 0.00014029988662861477,
      "loss": 1.1283,
      "step": 3867
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8035612106323242,
      "learning_rate": 0.00014027237632126068,
      "loss": 0.9893,
      "step": 3868
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7580231428146362,
      "learning_rate": 0.00014024486237552538,
      "loss": 0.9774,
      "step": 3869
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7281249165534973,
      "learning_rate": 0.00014021734479389465,
      "loss": 0.95,
      "step": 3870
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7264341711997986,
      "learning_rate": 0.0001401898235788545,
      "loss": 1.0655,
      "step": 3871
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7836666703224182,
      "learning_rate": 0.00014016229873289136,
      "loss": 0.8045,
      "step": 3872
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8120614886283875,
      "learning_rate": 0.00014013477025849195,
      "loss": 0.9924,
      "step": 3873
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.6951356530189514,
      "learning_rate": 0.00014010723815814326,
      "loss": 1.031,
      "step": 3874
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7149147987365723,
      "learning_rate": 0.00014007970243433272,
      "loss": 0.8887,
      "step": 3875
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7184805274009705,
      "learning_rate": 0.00014005216308954796,
      "loss": 0.9276,
      "step": 3876
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.768484354019165,
      "learning_rate": 0.00014002462012627703,
      "loss": 1.0148,
      "step": 3877
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7808959484100342,
      "learning_rate": 0.00013999707354700835,
      "loss": 0.9402,
      "step": 3878
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8177810311317444,
      "learning_rate": 0.0001399695233542305,
      "loss": 1.0563,
      "step": 3879
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.800896942615509,
      "learning_rate": 0.00013994196955043254,
      "loss": 1.0257,
      "step": 3880
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8390647172927856,
      "learning_rate": 0.00013991441213810374,
      "loss": 1.0125,
      "step": 3881
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.6802325248718262,
      "learning_rate": 0.00013988685111973384,
      "loss": 0.8674,
      "step": 3882
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8133691549301147,
      "learning_rate": 0.00013985928649781277,
      "loss": 0.9342,
      "step": 3883
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.736500084400177,
      "learning_rate": 0.00013983171827483085,
      "loss": 0.9786,
      "step": 3884
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7998164892196655,
      "learning_rate": 0.0001398041464532787,
      "loss": 1.0348,
      "step": 3885
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8324884176254272,
      "learning_rate": 0.00013977657103564727,
      "loss": 1.1051,
      "step": 3886
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7612653374671936,
      "learning_rate": 0.0001397489920244279,
      "loss": 0.9836,
      "step": 3887
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8343334794044495,
      "learning_rate": 0.00013972140942211214,
      "loss": 1.0232,
      "step": 3888
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7189517617225647,
      "learning_rate": 0.00013969382323119196,
      "loss": 0.9571,
      "step": 3889
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7711721062660217,
      "learning_rate": 0.00013966623345415956,
      "loss": 0.9215,
      "step": 3890
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.769665002822876,
      "learning_rate": 0.00013963864009350758,
      "loss": 0.8627,
      "step": 3891
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7339462041854858,
      "learning_rate": 0.0001396110431517289,
      "loss": 0.9453,
      "step": 3892
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8329907655715942,
      "learning_rate": 0.00013958344263131673,
      "loss": 0.8773,
      "step": 3893
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7495461702346802,
      "learning_rate": 0.00013955583853476468,
      "loss": 0.9406,
      "step": 3894
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7086147665977478,
      "learning_rate": 0.00013952823086456654,
      "loss": 0.6748,
      "step": 3895
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7977877855300903,
      "learning_rate": 0.00013950061962321656,
      "loss": 0.9603,
      "step": 3896
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7677198052406311,
      "learning_rate": 0.00013947300481320925,
      "loss": 1.0029,
      "step": 3897
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7576024532318115,
      "learning_rate": 0.0001394453864370395,
      "loss": 1.1522,
      "step": 3898
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7466287612915039,
      "learning_rate": 0.00013941776449720236,
      "loss": 0.8189,
      "step": 3899
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.6966318488121033,
      "learning_rate": 0.0001393901389961934,
      "loss": 0.9375,
      "step": 3900
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7954654097557068,
      "learning_rate": 0.00013936250993650838,
      "loss": 1.0475,
      "step": 3901
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.6510365009307861,
      "learning_rate": 0.00013933487732064353,
      "loss": 0.8256,
      "step": 3902
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7718179225921631,
      "learning_rate": 0.00013930724115109516,
      "loss": 1.1221,
      "step": 3903
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.747000515460968,
      "learning_rate": 0.00013927960143036015,
      "loss": 0.9645,
      "step": 3904
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7126678824424744,
      "learning_rate": 0.00013925195816093553,
      "loss": 0.851,
      "step": 3905
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7508659362792969,
      "learning_rate": 0.00013922431134531874,
      "loss": 0.7579,
      "step": 3906
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8252602815628052,
      "learning_rate": 0.00013919666098600753,
      "loss": 0.9258,
      "step": 3907
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8027669787406921,
      "learning_rate": 0.00013916900708549988,
      "loss": 1.1709,
      "step": 3908
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7662482261657715,
      "learning_rate": 0.00013914134964629427,
      "loss": 0.7885,
      "step": 3909
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7759925127029419,
      "learning_rate": 0.00013911368867088928,
      "loss": 1.0471,
      "step": 3910
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.829074501991272,
      "learning_rate": 0.00013908602416178403,
      "loss": 1.0349,
      "step": 3911
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7995732426643372,
      "learning_rate": 0.00013905835612147778,
      "loss": 1.0179,
      "step": 3912
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.6526477932929993,
      "learning_rate": 0.0001390306845524702,
      "loss": 0.8083,
      "step": 3913
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.791499674320221,
      "learning_rate": 0.00013900300945726127,
      "loss": 1.0309,
      "step": 3914
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8595665693283081,
      "learning_rate": 0.0001389753308383513,
      "loss": 0.9313,
      "step": 3915
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.6814246773719788,
      "learning_rate": 0.0001389476486982408,
      "loss": 0.8139,
      "step": 3916
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8005412220954895,
      "learning_rate": 0.00013891996303943077,
      "loss": 0.8702,
      "step": 3917
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.819680392742157,
      "learning_rate": 0.00013889227386442248,
      "loss": 0.7957,
      "step": 3918
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7297448515892029,
      "learning_rate": 0.00013886458117571743,
      "loss": 0.9389,
      "step": 3919
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7080660462379456,
      "learning_rate": 0.00013883688497581755,
      "loss": 1.0576,
      "step": 3920
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9082624316215515,
      "learning_rate": 0.00013880918526722497,
      "loss": 1.178,
      "step": 3921
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8355306386947632,
      "learning_rate": 0.00013878148205244225,
      "loss": 0.9952,
      "step": 3922
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7118585705757141,
      "learning_rate": 0.00013875377533397221,
      "loss": 0.9223,
      "step": 3923
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8073044419288635,
      "learning_rate": 0.00013872606511431803,
      "loss": 0.941,
      "step": 3924
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7692440152168274,
      "learning_rate": 0.0001386983513959831,
      "loss": 0.9574,
      "step": 3925
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.798399806022644,
      "learning_rate": 0.0001386706341814712,
      "loss": 1.0244,
      "step": 3926
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9476202130317688,
      "learning_rate": 0.0001386429134732865,
      "loss": 0.9288,
      "step": 3927
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7193227410316467,
      "learning_rate": 0.00013861518927393333,
      "loss": 0.9651,
      "step": 3928
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0366324186325073,
      "learning_rate": 0.00013858746158591648,
      "loss": 1.0158,
      "step": 3929
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.6833821535110474,
      "learning_rate": 0.00013855973041174092,
      "loss": 0.7529,
      "step": 3930
    },
    {
      "epoch": 1.48,
      "eval_loss": 0.9561265707015991,
      "eval_runtime": 17.7845,
      "eval_samples_per_second": 153.448,
      "eval_steps_per_second": 38.404,
      "step": 3930
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7655655741691589,
      "learning_rate": 0.00013853199575391206,
      "loss": 0.9809,
      "step": 3931
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.702148973941803,
      "learning_rate": 0.00013850425761493558,
      "loss": 0.8815,
      "step": 3932
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8169935345649719,
      "learning_rate": 0.0001384765159973174,
      "loss": 1.0864,
      "step": 3933
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.6614169478416443,
      "learning_rate": 0.00013844877090356389,
      "loss": 0.8591,
      "step": 3934
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8012004494667053,
      "learning_rate": 0.00013842102233618158,
      "loss": 0.9003,
      "step": 3935
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7351922988891602,
      "learning_rate": 0.00013839327029767744,
      "loss": 1.0095,
      "step": 3936
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.792578399181366,
      "learning_rate": 0.00013836551479055875,
      "loss": 0.8775,
      "step": 3937
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7366359233856201,
      "learning_rate": 0.000138337755817333,
      "loss": 1.0098,
      "step": 3938
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7572357654571533,
      "learning_rate": 0.0001383099933805081,
      "loss": 1.0691,
      "step": 3939
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.792985200881958,
      "learning_rate": 0.00013828222748259219,
      "loss": 0.9662,
      "step": 3940
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.791918933391571,
      "learning_rate": 0.00013825445812609374,
      "loss": 1.1858,
      "step": 3941
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7129440903663635,
      "learning_rate": 0.00013822668531352164,
      "loss": 0.9818,
      "step": 3942
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7407063841819763,
      "learning_rate": 0.000138198909047385,
      "loss": 1.0905,
      "step": 3943
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7492382526397705,
      "learning_rate": 0.00013817112933019315,
      "loss": 1.1537,
      "step": 3944
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7281045913696289,
      "learning_rate": 0.0001381433461644559,
      "loss": 0.9421,
      "step": 3945
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7278364896774292,
      "learning_rate": 0.00013811555955268328,
      "loss": 0.8511,
      "step": 3946
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.734602689743042,
      "learning_rate": 0.00013808776949738569,
      "loss": 0.9815,
      "step": 3947
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8731375932693481,
      "learning_rate": 0.0001380599760010738,
      "loss": 1.0916,
      "step": 3948
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.718619704246521,
      "learning_rate": 0.00013803217906625852,
      "loss": 0.8546,
      "step": 3949
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8179419040679932,
      "learning_rate": 0.00013800437869545125,
      "loss": 0.9959,
      "step": 3950
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7585844993591309,
      "learning_rate": 0.00013797657489116348,
      "loss": 0.9139,
      "step": 3951
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8072126507759094,
      "learning_rate": 0.00013794876765590723,
      "loss": 1.0179,
      "step": 3952
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7540550827980042,
      "learning_rate": 0.0001379209569921947,
      "loss": 0.9438,
      "step": 3953
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7537872791290283,
      "learning_rate": 0.0001378931429025384,
      "loss": 0.8506,
      "step": 3954
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7491480112075806,
      "learning_rate": 0.00013786532538945118,
      "loss": 0.9545,
      "step": 3955
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7633241415023804,
      "learning_rate": 0.00013783750445544622,
      "loss": 0.9078,
      "step": 3956
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7701250910758972,
      "learning_rate": 0.00013780968010303695,
      "loss": 0.891,
      "step": 3957
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7452569007873535,
      "learning_rate": 0.00013778185233473717,
      "loss": 1.0401,
      "step": 3958
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8602326512336731,
      "learning_rate": 0.000137754021153061,
      "loss": 1.0892,
      "step": 3959
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.6995726823806763,
      "learning_rate": 0.0001377261865605227,
      "loss": 0.8855,
      "step": 3960
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8985742330551147,
      "learning_rate": 0.00013769834855963708,
      "loss": 0.9417,
      "step": 3961
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7605254650115967,
      "learning_rate": 0.00013767050715291914,
      "loss": 0.9318,
      "step": 3962
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7218833565711975,
      "learning_rate": 0.00013764266234288412,
      "loss": 1.017,
      "step": 3963
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8716464638710022,
      "learning_rate": 0.00013761481413204773,
      "loss": 0.9225,
      "step": 3964
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8877992033958435,
      "learning_rate": 0.00013758696252292584,
      "loss": 1.04,
      "step": 3965
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.6978724598884583,
      "learning_rate": 0.0001375591075180347,
      "loss": 0.8764,
      "step": 3966
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8825734853744507,
      "learning_rate": 0.00013753124911989088,
      "loss": 1.0465,
      "step": 3967
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8109865188598633,
      "learning_rate": 0.00013750338733101115,
      "loss": 1.1471,
      "step": 3968
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7337487936019897,
      "learning_rate": 0.0001374755221539128,
      "loss": 0.942,
      "step": 3969
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7397893071174622,
      "learning_rate": 0.00013744765359111313,
      "loss": 1.1048,
      "step": 3970
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.773961067199707,
      "learning_rate": 0.00013741978164513,
      "loss": 1.0013,
      "step": 3971
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7440094947814941,
      "learning_rate": 0.0001373919063184815,
      "loss": 1.0297,
      "step": 3972
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8709926009178162,
      "learning_rate": 0.00013736402761368598,
      "loss": 1.0916,
      "step": 3973
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8232678174972534,
      "learning_rate": 0.00013733614553326212,
      "loss": 0.9848,
      "step": 3974
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7813726663589478,
      "learning_rate": 0.0001373082600797289,
      "loss": 1.0767,
      "step": 3975
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7393433451652527,
      "learning_rate": 0.0001372803712556056,
      "loss": 1.0539,
      "step": 3976
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7628100514411926,
      "learning_rate": 0.00013725247906341186,
      "loss": 0.9012,
      "step": 3977
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7220436334609985,
      "learning_rate": 0.00013722458350566753,
      "loss": 0.8679,
      "step": 3978
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.6969638466835022,
      "learning_rate": 0.0001371966845848929,
      "loss": 0.7855,
      "step": 3979
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8976461887359619,
      "learning_rate": 0.00013716878230360838,
      "loss": 1.0124,
      "step": 3980
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8752234578132629,
      "learning_rate": 0.00013714087666433483,
      "loss": 1.0404,
      "step": 3981
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7346514463424683,
      "learning_rate": 0.00013711296766959335,
      "loss": 1.0326,
      "step": 3982
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8979107737541199,
      "learning_rate": 0.0001370850553219054,
      "loss": 0.9406,
      "step": 3983
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8028544783592224,
      "learning_rate": 0.00013705713962379267,
      "loss": 1.1347,
      "step": 3984
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9713441133499146,
      "learning_rate": 0.00013702922057777723,
      "loss": 0.9485,
      "step": 3985
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.6593590974807739,
      "learning_rate": 0.00013700129818638133,
      "loss": 0.9953,
      "step": 3986
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.6654981374740601,
      "learning_rate": 0.00013697337245212763,
      "loss": 0.9089,
      "step": 3987
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9414458870887756,
      "learning_rate": 0.0001369454433775391,
      "loss": 1.244,
      "step": 3988
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.7171873450279236,
      "learning_rate": 0.00013691751096513892,
      "loss": 0.8293,
      "step": 3989
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8503627181053162,
      "learning_rate": 0.00013688957521745067,
      "loss": 1.1873,
      "step": 3990
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8794218897819519,
      "learning_rate": 0.00013686163613699814,
      "loss": 1.0791,
      "step": 3991
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7562376856803894,
      "learning_rate": 0.00013683369372630554,
      "loss": 0.9468,
      "step": 3992
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.747326135635376,
      "learning_rate": 0.00013680574798789722,
      "loss": 1.0067,
      "step": 3993
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.6801347732543945,
      "learning_rate": 0.00013677779892429798,
      "loss": 0.8768,
      "step": 3994
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9982435703277588,
      "learning_rate": 0.00013674984653803284,
      "loss": 0.9128,
      "step": 3995
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7373642921447754,
      "learning_rate": 0.00013672189083162715,
      "loss": 0.9162,
      "step": 3996
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.6660494804382324,
      "learning_rate": 0.0001366939318076065,
      "loss": 0.9756,
      "step": 3997
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7656930685043335,
      "learning_rate": 0.00013666596946849694,
      "loss": 0.9965,
      "step": 3998
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8281304836273193,
      "learning_rate": 0.00013663800381682464,
      "loss": 0.9022,
      "step": 3999
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7213367223739624,
      "learning_rate": 0.0001366100348551161,
      "loss": 1.0319,
      "step": 4000
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8114079833030701,
      "learning_rate": 0.0001365820625858982,
      "loss": 1.0008,
      "step": 4001
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7172978520393372,
      "learning_rate": 0.0001365540870116981,
      "loss": 0.9841,
      "step": 4002
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7489575743675232,
      "learning_rate": 0.00013652610813504317,
      "loss": 1.0965,
      "step": 4003
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8985177874565125,
      "learning_rate": 0.0001364981259584612,
      "loss": 1.0878,
      "step": 4004
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7306704521179199,
      "learning_rate": 0.0001364701404844802,
      "loss": 1.0132,
      "step": 4005
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8897285461425781,
      "learning_rate": 0.0001364421517156285,
      "loss": 0.8861,
      "step": 4006
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7293336987495422,
      "learning_rate": 0.0001364141596544347,
      "loss": 1.024,
      "step": 4007
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.6985815167427063,
      "learning_rate": 0.00013638616430342777,
      "loss": 0.902,
      "step": 4008
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7957209348678589,
      "learning_rate": 0.00013635816566513696,
      "loss": 1.1111,
      "step": 4009
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.756161630153656,
      "learning_rate": 0.00013633016374209168,
      "loss": 0.9724,
      "step": 4010
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8828256130218506,
      "learning_rate": 0.0001363021585368218,
      "loss": 0.8165,
      "step": 4011
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7583345174789429,
      "learning_rate": 0.0001362741500518575,
      "loss": 1.1074,
      "step": 4012
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8206169605255127,
      "learning_rate": 0.00013624613828972908,
      "loss": 0.9496,
      "step": 4013
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.6681649088859558,
      "learning_rate": 0.0001362181232529673,
      "loss": 0.9365,
      "step": 4014
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7639154195785522,
      "learning_rate": 0.0001361901049441032,
      "loss": 0.8674,
      "step": 4015
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8410853147506714,
      "learning_rate": 0.00013616208336566794,
      "loss": 1.0192,
      "step": 4016
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7662779092788696,
      "learning_rate": 0.00013613405852019324,
      "loss": 0.8263,
      "step": 4017
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7114266753196716,
      "learning_rate": 0.00013610603041021096,
      "loss": 0.8868,
      "step": 4018
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7133528590202332,
      "learning_rate": 0.00013607799903825324,
      "loss": 1.0178,
      "step": 4019
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7552312612533569,
      "learning_rate": 0.0001360499644068526,
      "loss": 0.9497,
      "step": 4020
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8263150453567505,
      "learning_rate": 0.00013602192651854174,
      "loss": 1.154,
      "step": 4021
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7287723422050476,
      "learning_rate": 0.00013599388537585382,
      "loss": 0.957,
      "step": 4022
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8263999819755554,
      "learning_rate": 0.00013596584098132217,
      "loss": 1.1154,
      "step": 4023
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7667343020439148,
      "learning_rate": 0.00013593779333748038,
      "loss": 0.9197,
      "step": 4024
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7314568758010864,
      "learning_rate": 0.00013590974244686248,
      "loss": 0.9844,
      "step": 4025
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7331525683403015,
      "learning_rate": 0.00013588168831200263,
      "loss": 0.999,
      "step": 4026
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7766103148460388,
      "learning_rate": 0.00013585363093543542,
      "loss": 0.99,
      "step": 4027
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7825350165367126,
      "learning_rate": 0.00013582557031969567,
      "loss": 0.9469,
      "step": 4028
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7960454821586609,
      "learning_rate": 0.00013579750646731847,
      "loss": 0.965,
      "step": 4029
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7917805314064026,
      "learning_rate": 0.00013576943938083927,
      "loss": 0.982,
      "step": 4030
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7845766544342041,
      "learning_rate": 0.0001357413690627937,
      "loss": 0.9755,
      "step": 4031
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7146531343460083,
      "learning_rate": 0.00013571329551571783,
      "loss": 0.8617,
      "step": 4032
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7622401118278503,
      "learning_rate": 0.00013568521874214796,
      "loss": 0.7809,
      "step": 4033
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7676883339881897,
      "learning_rate": 0.00013565713874462058,
      "loss": 1.0348,
      "step": 4034
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9790372252464294,
      "learning_rate": 0.0001356290555256726,
      "loss": 1.0473,
      "step": 4035
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9019134640693665,
      "learning_rate": 0.00013560096908784119,
      "loss": 1.0282,
      "step": 4036
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8120095133781433,
      "learning_rate": 0.0001355728794336638,
      "loss": 1.0385,
      "step": 4037
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8038890957832336,
      "learning_rate": 0.00013554478656567818,
      "loss": 1.0467,
      "step": 4038
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.6963586807250977,
      "learning_rate": 0.00013551669048642232,
      "loss": 0.993,
      "step": 4039
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.797791063785553,
      "learning_rate": 0.00013548859119843457,
      "loss": 1.0253,
      "step": 4040
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7721526026725769,
      "learning_rate": 0.00013546048870425356,
      "loss": 1.0582,
      "step": 4041
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8010839223861694,
      "learning_rate": 0.0001354323830064182,
      "loss": 1.0061,
      "step": 4042
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.723960816860199,
      "learning_rate": 0.00013540427410746762,
      "loss": 0.8673,
      "step": 4043
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7292559742927551,
      "learning_rate": 0.0001353761620099413,
      "loss": 1.277,
      "step": 4044
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7340478897094727,
      "learning_rate": 0.0001353480467163791,
      "loss": 0.9326,
      "step": 4045
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.6826416850090027,
      "learning_rate": 0.00013531992822932097,
      "loss": 0.96,
      "step": 4046
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9213247299194336,
      "learning_rate": 0.00013529180655130735,
      "loss": 0.8988,
      "step": 4047
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8114209175109863,
      "learning_rate": 0.0001352636816848788,
      "loss": 1.0367,
      "step": 4048
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7611256241798401,
      "learning_rate": 0.0001352355536325763,
      "loss": 0.9916,
      "step": 4049
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7742871046066284,
      "learning_rate": 0.00013520742239694103,
      "loss": 0.9578,
      "step": 4050
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7500746846199036,
      "learning_rate": 0.00013517928798051444,
      "loss": 1.0355,
      "step": 4051
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9206206202507019,
      "learning_rate": 0.0001351511503858384,
      "loss": 1.1098,
      "step": 4052
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7646648287773132,
      "learning_rate": 0.00013512300961545496,
      "loss": 1.1684,
      "step": 4053
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7009553909301758,
      "learning_rate": 0.00013509486567190646,
      "loss": 0.8637,
      "step": 4054
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7195073366165161,
      "learning_rate": 0.00013506671855773554,
      "loss": 0.8753,
      "step": 4055
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7679314613342285,
      "learning_rate": 0.00013503856827548513,
      "loss": 1.0202,
      "step": 4056
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9173232316970825,
      "learning_rate": 0.0001350104148276985,
      "loss": 0.911,
      "step": 4057
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8022065758705139,
      "learning_rate": 0.0001349822582169191,
      "loss": 1.0949,
      "step": 4058
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7112104296684265,
      "learning_rate": 0.00013495409844569076,
      "loss": 1.1556,
      "step": 4059
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7301331162452698,
      "learning_rate": 0.00013492593551655752,
      "loss": 0.8969,
      "step": 4060
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8814001083374023,
      "learning_rate": 0.00013489776943206375,
      "loss": 1.1629,
      "step": 4061
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.6501629948616028,
      "learning_rate": 0.0001348696001947541,
      "loss": 0.8705,
      "step": 4062
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.78933185338974,
      "learning_rate": 0.00013484142780717355,
      "loss": 1.1239,
      "step": 4063
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8425937294960022,
      "learning_rate": 0.0001348132522718672,
      "loss": 1.0504,
      "step": 4064
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.729695737361908,
      "learning_rate": 0.00013478507359138066,
      "loss": 0.8882,
      "step": 4065
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7847967147827148,
      "learning_rate": 0.00013475689176825967,
      "loss": 0.9521,
      "step": 4066
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8310196995735168,
      "learning_rate": 0.00013472870680505028,
      "loss": 0.8892,
      "step": 4067
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7683903574943542,
      "learning_rate": 0.00013470051870429888,
      "loss": 1.0349,
      "step": 4068
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8151059746742249,
      "learning_rate": 0.00013467232746855213,
      "loss": 0.861,
      "step": 4069
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9114918112754822,
      "learning_rate": 0.00013464413310035686,
      "loss": 0.9156,
      "step": 4070
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7894443869590759,
      "learning_rate": 0.0001346159356022603,
      "loss": 1.0753,
      "step": 4071
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7362051010131836,
      "learning_rate": 0.00013458773497680997,
      "loss": 0.8435,
      "step": 4072
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7690173387527466,
      "learning_rate": 0.00013455953122655366,
      "loss": 1.1125,
      "step": 4073
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7989892959594727,
      "learning_rate": 0.00013453132435403936,
      "loss": 0.8798,
      "step": 4074
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.709227979183197,
      "learning_rate": 0.00013450311436181538,
      "loss": 0.8941,
      "step": 4075
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8580532073974609,
      "learning_rate": 0.00013447490125243042,
      "loss": 0.9188,
      "step": 4076
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7872312068939209,
      "learning_rate": 0.0001344466850284333,
      "loss": 0.8544,
      "step": 4077
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7518987655639648,
      "learning_rate": 0.00013441846569237323,
      "loss": 0.8517,
      "step": 4078
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7955414056777954,
      "learning_rate": 0.00013439024324679968,
      "loss": 0.9751,
      "step": 4079
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8094600439071655,
      "learning_rate": 0.00013436201769426235,
      "loss": 1.0051,
      "step": 4080
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8085797429084778,
      "learning_rate": 0.00013433378903731125,
      "loss": 1.0915,
      "step": 4081
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8019757270812988,
      "learning_rate": 0.0001343055572784968,
      "loss": 0.9156,
      "step": 4082
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7250171899795532,
      "learning_rate": 0.0001342773224203694,
      "loss": 0.883,
      "step": 4083
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8753831386566162,
      "learning_rate": 0.00013424908446548004,
      "loss": 0.983,
      "step": 4084
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7171947956085205,
      "learning_rate": 0.00013422084341637979,
      "loss": 0.9899,
      "step": 4085
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7305735349655151,
      "learning_rate": 0.00013419259927562011,
      "loss": 0.8964,
      "step": 4086
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8032180070877075,
      "learning_rate": 0.0001341643520457527,
      "loss": 0.9373,
      "step": 4087
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7876158356666565,
      "learning_rate": 0.0001341361017293295,
      "loss": 0.967,
      "step": 4088
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.679061770439148,
      "learning_rate": 0.00013410784832890277,
      "loss": 0.7776,
      "step": 4089
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8126422166824341,
      "learning_rate": 0.0001340795918470251,
      "loss": 0.9163,
      "step": 4090
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8398410677909851,
      "learning_rate": 0.0001340513322862492,
      "loss": 0.9656,
      "step": 4091
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8889121413230896,
      "learning_rate": 0.0001340230696491283,
      "loss": 0.9531,
      "step": 4092
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.7200760245323181,
      "learning_rate": 0.00013399480393821569,
      "loss": 0.9501,
      "step": 4093
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.826648473739624,
      "learning_rate": 0.00013396653515606498,
      "loss": 1.053,
      "step": 4094
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7926040887832642,
      "learning_rate": 0.0001339382633052302,
      "loss": 1.0978,
      "step": 4095
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.788530170917511,
      "learning_rate": 0.00013390998838826542,
      "loss": 1.0361,
      "step": 4096
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7253285050392151,
      "learning_rate": 0.0001338817104077252,
      "loss": 0.6804,
      "step": 4097
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8385868668556213,
      "learning_rate": 0.0001338534293661643,
      "loss": 0.8149,
      "step": 4098
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.6631655693054199,
      "learning_rate": 0.00013382514526613773,
      "loss": 0.9524,
      "step": 4099
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7763095498085022,
      "learning_rate": 0.0001337968581102008,
      "loss": 0.9904,
      "step": 4100
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.6638621687889099,
      "learning_rate": 0.0001337685679009091,
      "loss": 0.9542,
      "step": 4101
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7378938794136047,
      "learning_rate": 0.00013374027464081847,
      "loss": 0.9329,
      "step": 4102
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7555968761444092,
      "learning_rate": 0.00013371197833248507,
      "loss": 0.8773,
      "step": 4103
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8379522562026978,
      "learning_rate": 0.00013368367897846533,
      "loss": 1.0052,
      "step": 4104
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.990310788154602,
      "learning_rate": 0.0001336553765813159,
      "loss": 1.2306,
      "step": 4105
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.6684477925300598,
      "learning_rate": 0.00013362707114359374,
      "loss": 0.8162,
      "step": 4106
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7488822340965271,
      "learning_rate": 0.0001335987626678561,
      "loss": 1.0563,
      "step": 4107
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8354619145393372,
      "learning_rate": 0.0001335704511566605,
      "loss": 1.0533,
      "step": 4108
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.829838752746582,
      "learning_rate": 0.00013354213661256472,
      "loss": 0.9526,
      "step": 4109
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7552241683006287,
      "learning_rate": 0.0001335138190381268,
      "loss": 0.8768,
      "step": 4110
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.6395108103752136,
      "learning_rate": 0.00013348549843590504,
      "loss": 0.9264,
      "step": 4111
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7724140882492065,
      "learning_rate": 0.00013345717480845817,
      "loss": 1.0384,
      "step": 4112
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7144059538841248,
      "learning_rate": 0.00013342884815834493,
      "loss": 1.069,
      "step": 4113
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7421502470970154,
      "learning_rate": 0.00013340051848812458,
      "loss": 1.0402,
      "step": 4114
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7333446741104126,
      "learning_rate": 0.00013337218580035647,
      "loss": 1.0038,
      "step": 4115
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8908722400665283,
      "learning_rate": 0.00013334385009760032,
      "loss": 1.1055,
      "step": 4116
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.6802178621292114,
      "learning_rate": 0.0001333155113824161,
      "loss": 0.8414,
      "step": 4117
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8234414458274841,
      "learning_rate": 0.00013328716965736408,
      "loss": 1.1654,
      "step": 4118
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.749435305595398,
      "learning_rate": 0.00013325882492500477,
      "loss": 1.0812,
      "step": 4119
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7330403923988342,
      "learning_rate": 0.0001332304771878989,
      "loss": 0.9709,
      "step": 4120
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.794902503490448,
      "learning_rate": 0.00013320212644860757,
      "loss": 0.9516,
      "step": 4121
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7441633939743042,
      "learning_rate": 0.00013317377270969212,
      "loss": 1.0171,
      "step": 4122
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8326237201690674,
      "learning_rate": 0.00013314541597371412,
      "loss": 0.9255,
      "step": 4123
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8376749753952026,
      "learning_rate": 0.0001331170562432355,
      "loss": 1.0207,
      "step": 4124
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7114330530166626,
      "learning_rate": 0.00013308869352081834,
      "loss": 0.79,
      "step": 4125
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7856300473213196,
      "learning_rate": 0.00013306032780902502,
      "loss": 0.9169,
      "step": 4126
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7197442650794983,
      "learning_rate": 0.00013303195911041832,
      "loss": 0.9439,
      "step": 4127
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7740024924278259,
      "learning_rate": 0.0001330035874275611,
      "loss": 1.0669,
      "step": 4128
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8333525061607361,
      "learning_rate": 0.00013297521276301664,
      "loss": 1.1763,
      "step": 4129
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8492178320884705,
      "learning_rate": 0.00013294683511934845,
      "loss": 0.9247,
      "step": 4130
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8128172755241394,
      "learning_rate": 0.0001329184544991202,
      "loss": 1.0977,
      "step": 4131
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7545269131660461,
      "learning_rate": 0.000132890070904896,
      "loss": 0.8532,
      "step": 4132
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7314281463623047,
      "learning_rate": 0.0001328616843392401,
      "loss": 0.8558,
      "step": 4133
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.6871530413627625,
      "learning_rate": 0.0001328332948047171,
      "loss": 0.6922,
      "step": 4134
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7285220623016357,
      "learning_rate": 0.0001328049023038918,
      "loss": 0.9591,
      "step": 4135
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8840282559394836,
      "learning_rate": 0.00013277650683932933,
      "loss": 0.9727,
      "step": 4136
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7239083647727966,
      "learning_rate": 0.00013274810841359502,
      "loss": 0.9342,
      "step": 4137
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9012027978897095,
      "learning_rate": 0.00013271970702925456,
      "loss": 1.0467,
      "step": 4138
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7457629442214966,
      "learning_rate": 0.00013269130268887382,
      "loss": 0.9247,
      "step": 4139
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7632551193237305,
      "learning_rate": 0.000132662895395019,
      "loss": 1.1305,
      "step": 4140
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8433321118354797,
      "learning_rate": 0.00013263448515025647,
      "loss": 1.1017,
      "step": 4141
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7884944081306458,
      "learning_rate": 0.00013260607195715303,
      "loss": 0.9197,
      "step": 4142
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8757911920547485,
      "learning_rate": 0.0001325776558182756,
      "loss": 1.1378,
      "step": 4143
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.985319197177887,
      "learning_rate": 0.0001325492367361914,
      "loss": 1.2242,
      "step": 4144
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.6358724236488342,
      "learning_rate": 0.00013252081471346795,
      "loss": 0.9055,
      "step": 4145
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7014467716217041,
      "learning_rate": 0.00013249238975267304,
      "loss": 1.0029,
      "step": 4146
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8960198760032654,
      "learning_rate": 0.00013246396185637467,
      "loss": 1.0953,
      "step": 4147
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7448418140411377,
      "learning_rate": 0.00013243553102714118,
      "loss": 0.8569,
      "step": 4148
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8383636474609375,
      "learning_rate": 0.0001324070972675411,
      "loss": 1.1298,
      "step": 4149
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.6508424282073975,
      "learning_rate": 0.00013237866058014326,
      "loss": 0.7927,
      "step": 4150
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7935487627983093,
      "learning_rate": 0.00013235022096751676,
      "loss": 0.9801,
      "step": 4151
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7764027118682861,
      "learning_rate": 0.000132321778432231,
      "loss": 1.1115,
      "step": 4152
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8405836820602417,
      "learning_rate": 0.00013229333297685552,
      "loss": 0.9208,
      "step": 4153
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7989316582679749,
      "learning_rate": 0.00013226488460396028,
      "loss": 1.0406,
      "step": 4154
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7648982405662537,
      "learning_rate": 0.00013223643331611537,
      "loss": 0.96,
      "step": 4155
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7140777111053467,
      "learning_rate": 0.00013220797911589127,
      "loss": 0.8974,
      "step": 4156
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7468279004096985,
      "learning_rate": 0.00013217952200585858,
      "loss": 0.8749,
      "step": 4157
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7843912839889526,
      "learning_rate": 0.0001321510619885883,
      "loss": 0.9856,
      "step": 4158
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7949358820915222,
      "learning_rate": 0.0001321225990666516,
      "loss": 1.0912,
      "step": 4159
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.824992835521698,
      "learning_rate": 0.00013209413324261998,
      "loss": 0.9953,
      "step": 4160
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7797181606292725,
      "learning_rate": 0.00013206566451906508,
      "loss": 0.9361,
      "step": 4161
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7983461022377014,
      "learning_rate": 0.000132037192898559,
      "loss": 0.9637,
      "step": 4162
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7876800894737244,
      "learning_rate": 0.0001320087183836739,
      "loss": 0.9159,
      "step": 4163
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7618224024772644,
      "learning_rate": 0.00013198024097698237,
      "loss": 0.7168,
      "step": 4164
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.6901776790618896,
      "learning_rate": 0.00013195176068105712,
      "loss": 0.8032,
      "step": 4165
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8126145601272583,
      "learning_rate": 0.00013192327749847116,
      "loss": 0.9431,
      "step": 4166
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7916374802589417,
      "learning_rate": 0.00013189479143179789,
      "loss": 0.9993,
      "step": 4167
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8331419825553894,
      "learning_rate": 0.0001318663024836108,
      "loss": 0.9779,
      "step": 4168
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8191587328910828,
      "learning_rate": 0.00013183781065648364,
      "loss": 1.0442,
      "step": 4169
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7728171944618225,
      "learning_rate": 0.0001318093159529906,
      "loss": 0.9568,
      "step": 4170
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7042739987373352,
      "learning_rate": 0.00013178081837570598,
      "loss": 0.7985,
      "step": 4171
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7546452879905701,
      "learning_rate": 0.0001317523179272043,
      "loss": 0.9792,
      "step": 4172
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9695934057235718,
      "learning_rate": 0.00013172381461006057,
      "loss": 0.9767,
      "step": 4173
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9317048788070679,
      "learning_rate": 0.00013169530842684974,
      "loss": 1.0918,
      "step": 4174
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.6831107139587402,
      "learning_rate": 0.00013166679938014726,
      "loss": 0.7801,
      "step": 4175
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8045737147331238,
      "learning_rate": 0.00013163828747252878,
      "loss": 0.9696,
      "step": 4176
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7625781297683716,
      "learning_rate": 0.00013160977270657012,
      "loss": 0.9746,
      "step": 4177
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7749671936035156,
      "learning_rate": 0.0001315812550848475,
      "loss": 0.8871,
      "step": 4178
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8254308104515076,
      "learning_rate": 0.00013155273460993728,
      "loss": 0.9818,
      "step": 4179
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8079273700714111,
      "learning_rate": 0.00013152421128441617,
      "loss": 1.0846,
      "step": 4180
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8321089744567871,
      "learning_rate": 0.00013149568511086103,
      "loss": 1.1869,
      "step": 4181
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8857249617576599,
      "learning_rate": 0.00013146715609184907,
      "loss": 1.1094,
      "step": 4182
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8076190948486328,
      "learning_rate": 0.00013143862422995776,
      "loss": 1.0152,
      "step": 4183
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.800880491733551,
      "learning_rate": 0.00013141008952776475,
      "loss": 0.9568,
      "step": 4184
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7508581876754761,
      "learning_rate": 0.00013138155198784795,
      "loss": 0.8768,
      "step": 4185
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8762493133544922,
      "learning_rate": 0.00013135301161278564,
      "loss": 1.1144,
      "step": 4186
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7529940605163574,
      "learning_rate": 0.00013132446840515623,
      "loss": 0.8161,
      "step": 4187
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.6504919528961182,
      "learning_rate": 0.00013129592236753846,
      "loss": 0.8698,
      "step": 4188
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.801974892616272,
      "learning_rate": 0.00013126737350251135,
      "loss": 0.9568,
      "step": 4189
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.737905740737915,
      "learning_rate": 0.00013123882181265403,
      "loss": 0.9602,
      "step": 4190
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8118032813072205,
      "learning_rate": 0.00013121026730054605,
      "loss": 0.9315,
      "step": 4191
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8885903358459473,
      "learning_rate": 0.00013118170996876714,
      "loss": 0.9292,
      "step": 4192
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7553448677062988,
      "learning_rate": 0.00013115314981989725,
      "loss": 1.0179,
      "step": 4193
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8398468494415283,
      "learning_rate": 0.00013112458685651668,
      "loss": 0.9042,
      "step": 4194
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.761009693145752,
      "learning_rate": 0.00013109602108120593,
      "loss": 0.9415,
      "step": 4195
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7048000693321228,
      "learning_rate": 0.0001310674524965457,
      "loss": 0.8545,
      "step": 4196
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7960367202758789,
      "learning_rate": 0.0001310388811051171,
      "loss": 1.0841,
      "step": 4197
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8713595867156982,
      "learning_rate": 0.0001310103069095013,
      "loss": 1.1152,
      "step": 4198
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.7115532159805298,
      "learning_rate": 0.00013098172991227985,
      "loss": 1.083,
      "step": 4199
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7147960662841797,
      "learning_rate": 0.00013095315011603454,
      "loss": 0.9406,
      "step": 4200
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8112908005714417,
      "learning_rate": 0.00013092456752334734,
      "loss": 1.0151,
      "step": 4201
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8009018898010254,
      "learning_rate": 0.00013089598213680058,
      "loss": 1.0362,
      "step": 4202
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8003485798835754,
      "learning_rate": 0.00013086739395897677,
      "loss": 0.7262,
      "step": 4203
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7411230206489563,
      "learning_rate": 0.00013083880299245867,
      "loss": 0.9487,
      "step": 4204
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7706307768821716,
      "learning_rate": 0.00013081020923982935,
      "loss": 0.8444,
      "step": 4205
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8083503842353821,
      "learning_rate": 0.00013078161270367207,
      "loss": 0.9118,
      "step": 4206
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.750612199306488,
      "learning_rate": 0.00013075301338657036,
      "loss": 1.0934,
      "step": 4207
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8122290968894958,
      "learning_rate": 0.00013072441129110803,
      "loss": 1.0062,
      "step": 4208
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.6777164340019226,
      "learning_rate": 0.00013069580641986907,
      "loss": 0.9177,
      "step": 4209
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9212057590484619,
      "learning_rate": 0.00013066719877543782,
      "loss": 1.1174,
      "step": 4210
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.75261390209198,
      "learning_rate": 0.00013063858836039876,
      "loss": 0.9974,
      "step": 4211
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8800492286682129,
      "learning_rate": 0.00013060997517733676,
      "loss": 0.9127,
      "step": 4212
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7661362290382385,
      "learning_rate": 0.0001305813592288368,
      "loss": 0.9915,
      "step": 4213
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8467798829078674,
      "learning_rate": 0.00013055274051748417,
      "loss": 1.2046,
      "step": 4214
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7771646976470947,
      "learning_rate": 0.00013052411904586445,
      "loss": 0.9882,
      "step": 4215
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8530020117759705,
      "learning_rate": 0.0001304954948165634,
      "loss": 1.0174,
      "step": 4216
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7605085968971252,
      "learning_rate": 0.00013046686783216703,
      "loss": 1.0307,
      "step": 4217
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.6596463322639465,
      "learning_rate": 0.0001304382380952617,
      "loss": 0.7798,
      "step": 4218
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.74962317943573,
      "learning_rate": 0.0001304096056084339,
      "loss": 0.8796,
      "step": 4219
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7627604603767395,
      "learning_rate": 0.00013038097037427037,
      "loss": 0.9478,
      "step": 4220
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7824596762657166,
      "learning_rate": 0.00013035233239535817,
      "loss": 1.0287,
      "step": 4221
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8189258575439453,
      "learning_rate": 0.0001303236916742846,
      "loss": 0.9816,
      "step": 4222
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7686831951141357,
      "learning_rate": 0.00013029504821363724,
      "loss": 0.9055,
      "step": 4223
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8042637705802917,
      "learning_rate": 0.00013026640201600377,
      "loss": 0.9434,
      "step": 4224
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8106065988540649,
      "learning_rate": 0.00013023775308397222,
      "loss": 1.1275,
      "step": 4225
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8299038410186768,
      "learning_rate": 0.00013020910142013092,
      "loss": 1.0951,
      "step": 4226
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8075307607650757,
      "learning_rate": 0.00013018044702706836,
      "loss": 0.931,
      "step": 4227
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7008748650550842,
      "learning_rate": 0.00013015178990737324,
      "loss": 1.0001,
      "step": 4228
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8621682524681091,
      "learning_rate": 0.00013012313006363467,
      "loss": 0.9809,
      "step": 4229
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7828312516212463,
      "learning_rate": 0.00013009446749844182,
      "loss": 0.8595,
      "step": 4230
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7661519646644592,
      "learning_rate": 0.00013006580221438423,
      "loss": 0.7974,
      "step": 4231
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7845397591590881,
      "learning_rate": 0.00013003713421405166,
      "loss": 0.9607,
      "step": 4232
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7443758249282837,
      "learning_rate": 0.0001300084635000341,
      "loss": 1.0433,
      "step": 4233
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7668888568878174,
      "learning_rate": 0.00012997979007492173,
      "loss": 1.0791,
      "step": 4234
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7785104513168335,
      "learning_rate": 0.0001299511139413051,
      "loss": 0.881,
      "step": 4235
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8156974911689758,
      "learning_rate": 0.00012992243510177487,
      "loss": 1.1101,
      "step": 4236
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7256063222885132,
      "learning_rate": 0.0001298937535589221,
      "loss": 0.9809,
      "step": 4237
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7718733549118042,
      "learning_rate": 0.0001298650693153379,
      "loss": 0.9407,
      "step": 4238
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.72328120470047,
      "learning_rate": 0.00012983638237361382,
      "loss": 0.9191,
      "step": 4239
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.978679895401001,
      "learning_rate": 0.00012980769273634156,
      "loss": 1.1024,
      "step": 4240
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.767125129699707,
      "learning_rate": 0.00012977900040611295,
      "loss": 1.1594,
      "step": 4241
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7211596965789795,
      "learning_rate": 0.00012975030538552032,
      "loss": 0.8163,
      "step": 4242
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8155632019042969,
      "learning_rate": 0.00012972160767715603,
      "loss": 1.1148,
      "step": 4243
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7436785101890564,
      "learning_rate": 0.00012969290728361273,
      "loss": 1.105,
      "step": 4244
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7024710178375244,
      "learning_rate": 0.00012966420420748345,
      "loss": 0.6854,
      "step": 4245
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7749850153923035,
      "learning_rate": 0.00012963549845136123,
      "loss": 0.9937,
      "step": 4246
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.6667187213897705,
      "learning_rate": 0.00012960679001783952,
      "loss": 0.7582,
      "step": 4247
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7609478831291199,
      "learning_rate": 0.000129578078909512,
      "loss": 0.8994,
      "step": 4248
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.6955171823501587,
      "learning_rate": 0.00012954936512897248,
      "loss": 1.1101,
      "step": 4249
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.6980825066566467,
      "learning_rate": 0.00012952064867881514,
      "loss": 1.1454,
      "step": 4250
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.7012463212013245,
      "learning_rate": 0.00012949192956163437,
      "loss": 0.8899,
      "step": 4251
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8292000889778137,
      "learning_rate": 0.0001294632077800247,
      "loss": 0.9554,
      "step": 4252
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7803111672401428,
      "learning_rate": 0.00012943448333658107,
      "loss": 0.8801,
      "step": 4253
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9052737355232239,
      "learning_rate": 0.00012940575623389847,
      "loss": 0.9664,
      "step": 4254
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7098575234413147,
      "learning_rate": 0.00012937702647457234,
      "loss": 1.0257,
      "step": 4255
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.682171106338501,
      "learning_rate": 0.00012934829406119818,
      "loss": 0.897,
      "step": 4256
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7598903775215149,
      "learning_rate": 0.00012931955899637181,
      "loss": 0.9512,
      "step": 4257
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7627248167991638,
      "learning_rate": 0.00012929082128268932,
      "loss": 0.9523,
      "step": 4258
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7255799174308777,
      "learning_rate": 0.000129262080922747,
      "loss": 1.0527,
      "step": 4259
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7689884305000305,
      "learning_rate": 0.0001292333379191413,
      "loss": 1.1442,
      "step": 4260
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8066941499710083,
      "learning_rate": 0.00012920459227446903,
      "loss": 0.9867,
      "step": 4261
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.775963544845581,
      "learning_rate": 0.00012917584399132722,
      "loss": 1.027,
      "step": 4262
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7826582789421082,
      "learning_rate": 0.0001291470930723131,
      "loss": 0.978,
      "step": 4263
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7322339415550232,
      "learning_rate": 0.00012911833952002416,
      "loss": 1.0189,
      "step": 4264
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7850020527839661,
      "learning_rate": 0.00012908958333705808,
      "loss": 1.1508,
      "step": 4265
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7325538992881775,
      "learning_rate": 0.00012906082452601288,
      "loss": 0.8523,
      "step": 4266
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8257113099098206,
      "learning_rate": 0.00012903206308948674,
      "loss": 0.9526,
      "step": 4267
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7576756477355957,
      "learning_rate": 0.00012900329903007805,
      "loss": 0.9564,
      "step": 4268
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7235482931137085,
      "learning_rate": 0.00012897453235038552,
      "loss": 1.044,
      "step": 4269
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7639951705932617,
      "learning_rate": 0.00012894576305300806,
      "loss": 1.1088,
      "step": 4270
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8137654066085815,
      "learning_rate": 0.00012891699114054477,
      "loss": 1.0389,
      "step": 4271
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7565323710441589,
      "learning_rate": 0.00012888821661559508,
      "loss": 0.9217,
      "step": 4272
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.6858435869216919,
      "learning_rate": 0.00012885943948075856,
      "loss": 0.9392,
      "step": 4273
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.820617139339447,
      "learning_rate": 0.0001288306597386351,
      "loss": 1.0266,
      "step": 4274
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8873162269592285,
      "learning_rate": 0.00012880187739182478,
      "loss": 0.8399,
      "step": 4275
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.7443687915802002,
      "learning_rate": 0.00012877309244292787,
      "loss": 0.8388,
      "step": 4276
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.858851969242096,
      "learning_rate": 0.00012874430489454502,
      "loss": 0.9356,
      "step": 4277
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.850151002407074,
      "learning_rate": 0.00012871551474927698,
      "loss": 1.021,
      "step": 4278
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7548375725746155,
      "learning_rate": 0.00012868672200972475,
      "loss": 0.7736,
      "step": 4279
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.6958070397377014,
      "learning_rate": 0.0001286579266784896,
      "loss": 0.9859,
      "step": 4280
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7235026359558105,
      "learning_rate": 0.00012862912875817305,
      "loss": 0.9848,
      "step": 4281
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7634601593017578,
      "learning_rate": 0.00012860032825137677,
      "loss": 0.9571,
      "step": 4282
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7626282572746277,
      "learning_rate": 0.00012857152516070283,
      "loss": 1.0399,
      "step": 4283
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7731013298034668,
      "learning_rate": 0.00012854271948875332,
      "loss": 0.9295,
      "step": 4284
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8193756341934204,
      "learning_rate": 0.00012851391123813074,
      "loss": 1.1184,
      "step": 4285
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7630685567855835,
      "learning_rate": 0.0001284851004114377,
      "loss": 0.9509,
      "step": 4286
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7839036583900452,
      "learning_rate": 0.00012845628701127712,
      "loss": 0.8557,
      "step": 4287
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7505083084106445,
      "learning_rate": 0.00012842747104025215,
      "loss": 0.8715,
      "step": 4288
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8192002773284912,
      "learning_rate": 0.00012839865250096613,
      "loss": 1.1744,
      "step": 4289
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7832188606262207,
      "learning_rate": 0.00012836983139602264,
      "loss": 1.0416,
      "step": 4290
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8175313472747803,
      "learning_rate": 0.00012834100772802552,
      "loss": 1.043,
      "step": 4291
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.6733172535896301,
      "learning_rate": 0.00012831218149957877,
      "loss": 1.066,
      "step": 4292
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.775862991809845,
      "learning_rate": 0.00012828335271328678,
      "loss": 0.949,
      "step": 4293
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7226769328117371,
      "learning_rate": 0.000128254521371754,
      "loss": 0.9923,
      "step": 4294
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.762204647064209,
      "learning_rate": 0.00012822568747758518,
      "loss": 0.9009,
      "step": 4295
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8002398014068604,
      "learning_rate": 0.00012819685103338533,
      "loss": 1.013,
      "step": 4296
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7634932398796082,
      "learning_rate": 0.00012816801204175964,
      "loss": 0.9535,
      "step": 4297
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7286591529846191,
      "learning_rate": 0.00012813917050531352,
      "loss": 0.8235,
      "step": 4298
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7977073788642883,
      "learning_rate": 0.00012811032642665268,
      "loss": 0.86,
      "step": 4299
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.737250030040741,
      "learning_rate": 0.00012808147980838302,
      "loss": 1.0641,
      "step": 4300
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7920666337013245,
      "learning_rate": 0.00012805263065311063,
      "loss": 0.8847,
      "step": 4301
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7638289332389832,
      "learning_rate": 0.00012802377896344193,
      "loss": 1.1462,
      "step": 4302
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.7117271423339844,
      "learning_rate": 0.00012799492474198344,
      "loss": 0.9799,
      "step": 4303
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8018771409988403,
      "learning_rate": 0.00012796606799134202,
      "loss": 1.0583,
      "step": 4304
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8100774884223938,
      "learning_rate": 0.00012793720871412468,
      "loss": 0.9696,
      "step": 4305
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7184478044509888,
      "learning_rate": 0.00012790834691293873,
      "loss": 0.9018,
      "step": 4306
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8160077333450317,
      "learning_rate": 0.00012787948259039165,
      "loss": 1.1151,
      "step": 4307
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.670409083366394,
      "learning_rate": 0.0001278506157490912,
      "loss": 0.9515,
      "step": 4308
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.730998694896698,
      "learning_rate": 0.0001278217463916453,
      "loss": 1.0486,
      "step": 4309
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7451099753379822,
      "learning_rate": 0.0001277928745206621,
      "loss": 0.8497,
      "step": 4310
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7183091640472412,
      "learning_rate": 0.00012776400013875006,
      "loss": 0.9432,
      "step": 4311
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7617359161376953,
      "learning_rate": 0.0001277351232485178,
      "loss": 0.8215,
      "step": 4312
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7619341015815735,
      "learning_rate": 0.00012770624385257423,
      "loss": 1.071,
      "step": 4313
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9530133008956909,
      "learning_rate": 0.0001276773619535284,
      "loss": 0.9284,
      "step": 4314
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8487468957901001,
      "learning_rate": 0.0001276484775539896,
      "loss": 1.1118,
      "step": 4315
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8442698121070862,
      "learning_rate": 0.00012761959065656737,
      "loss": 0.8761,
      "step": 4316
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7341495156288147,
      "learning_rate": 0.00012759070126387156,
      "loss": 0.9816,
      "step": 4317
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.900442361831665,
      "learning_rate": 0.0001275618093785121,
      "loss": 0.9645,
      "step": 4318
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8728200793266296,
      "learning_rate": 0.00012753291500309922,
      "loss": 0.9267,
      "step": 4319
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7861907482147217,
      "learning_rate": 0.00012750401814024334,
      "loss": 1.1425,
      "step": 4320
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8056981563568115,
      "learning_rate": 0.00012747511879255513,
      "loss": 0.959,
      "step": 4321
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8388335704803467,
      "learning_rate": 0.00012744621696264555,
      "loss": 1.0679,
      "step": 4322
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7195658087730408,
      "learning_rate": 0.00012741731265312564,
      "loss": 0.8373,
      "step": 4323
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7100113034248352,
      "learning_rate": 0.00012738840586660677,
      "loss": 0.9086,
      "step": 4324
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7771344780921936,
      "learning_rate": 0.0001273594966057005,
      "loss": 1.0846,
      "step": 4325
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.748335599899292,
      "learning_rate": 0.00012733058487301863,
      "loss": 0.9785,
      "step": 4326
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8741627335548401,
      "learning_rate": 0.00012730167067117313,
      "loss": 0.9909,
      "step": 4327
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8887857794761658,
      "learning_rate": 0.0001272727540027763,
      "loss": 0.951,
      "step": 4328
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7612056136131287,
      "learning_rate": 0.00012724383487044055,
      "loss": 1.0076,
      "step": 4329
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7381542921066284,
      "learning_rate": 0.00012721491327677862,
      "loss": 0.9325,
      "step": 4330
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.77471524477005,
      "learning_rate": 0.0001271859892244033,
      "loss": 0.9169,
      "step": 4331
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8204985857009888,
      "learning_rate": 0.0001271570627159278,
      "loss": 1.063,
      "step": 4332
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.745324432849884,
      "learning_rate": 0.00012712813375396546,
      "loss": 0.9889,
      "step": 4333
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7239276170730591,
      "learning_rate": 0.00012709920234112984,
      "loss": 0.973,
      "step": 4334
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.6909604668617249,
      "learning_rate": 0.00012707026848003471,
      "loss": 0.8158,
      "step": 4335
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9055747389793396,
      "learning_rate": 0.00012704133217329407,
      "loss": 0.9749,
      "step": 4336
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8274331092834473,
      "learning_rate": 0.00012701239342352222,
      "loss": 0.939,
      "step": 4337
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8642345070838928,
      "learning_rate": 0.00012698345223333355,
      "loss": 0.9019,
      "step": 4338
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8090344667434692,
      "learning_rate": 0.00012695450860534277,
      "loss": 0.9395,
      "step": 4339
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7602019906044006,
      "learning_rate": 0.00012692556254216471,
      "loss": 0.9288,
      "step": 4340
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7767679691314697,
      "learning_rate": 0.00012689661404641456,
      "loss": 1.0532,
      "step": 4341
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.6911072134971619,
      "learning_rate": 0.0001268676631207076,
      "loss": 0.8661,
      "step": 4342
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7428793907165527,
      "learning_rate": 0.0001268387097676594,
      "loss": 0.953,
      "step": 4343
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7641132473945618,
      "learning_rate": 0.00012680975398988577,
      "loss": 0.8437,
      "step": 4344
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7458558082580566,
      "learning_rate": 0.00012678079579000264,
      "loss": 0.8762,
      "step": 4345
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8327491283416748,
      "learning_rate": 0.0001267518351706262,
      "loss": 1.0132,
      "step": 4346
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7638497948646545,
      "learning_rate": 0.000126722872134373,
      "loss": 0.803,
      "step": 4347
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.720806896686554,
      "learning_rate": 0.00012669390668385954,
      "loss": 1.0511,
      "step": 4348
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7823967933654785,
      "learning_rate": 0.0001266649388217028,
      "loss": 0.9631,
      "step": 4349
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7093958258628845,
      "learning_rate": 0.0001266359685505198,
      "loss": 0.9645,
      "step": 4350
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9044625759124756,
      "learning_rate": 0.00012660699587292782,
      "loss": 1.0589,
      "step": 4351
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8032203316688538,
      "learning_rate": 0.00012657802079154445,
      "loss": 1.1116,
      "step": 4352
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8076255917549133,
      "learning_rate": 0.00012654904330898742,
      "loss": 1.0103,
      "step": 4353
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7746655344963074,
      "learning_rate": 0.00012652006342787458,
      "loss": 0.8966,
      "step": 4354
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.854537308216095,
      "learning_rate": 0.00012649108115082418,
      "loss": 0.9969,
      "step": 4355
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.6939969062805176,
      "learning_rate": 0.0001264620964804546,
      "loss": 0.8869,
      "step": 4356
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7948812246322632,
      "learning_rate": 0.00012643310941938443,
      "loss": 0.9075,
      "step": 4357
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7065130472183228,
      "learning_rate": 0.00012640411997023247,
      "loss": 1.0169,
      "step": 4358
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7679563760757446,
      "learning_rate": 0.00012637512813561778,
      "loss": 0.9954,
      "step": 4359
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7309630513191223,
      "learning_rate": 0.0001263461339181596,
      "loss": 0.924,
      "step": 4360
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7593986988067627,
      "learning_rate": 0.00012631713732047742,
      "loss": 1.1259,
      "step": 4361
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7429969906806946,
      "learning_rate": 0.00012628813834519083,
      "loss": 1.0467,
      "step": 4362
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8035417199134827,
      "learning_rate": 0.00012625913699491984,
      "loss": 1.0063,
      "step": 4363
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7101812958717346,
      "learning_rate": 0.00012623013327228447,
      "loss": 1.0125,
      "step": 4364
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7777599692344666,
      "learning_rate": 0.0001262011271799051,
      "loss": 0.9239,
      "step": 4365
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.680126428604126,
      "learning_rate": 0.0001261721187204022,
      "loss": 0.8978,
      "step": 4366
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7517068982124329,
      "learning_rate": 0.0001261431078963966,
      "loss": 1.0133,
      "step": 4367
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7588438987731934,
      "learning_rate": 0.00012611409471050924,
      "loss": 1.0754,
      "step": 4368
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7387205362319946,
      "learning_rate": 0.00012608507916536126,
      "loss": 1.0153,
      "step": 4369
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7710445523262024,
      "learning_rate": 0.00012605606126357403,
      "loss": 0.8623,
      "step": 4370
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7151059508323669,
      "learning_rate": 0.00012602704100776922,
      "loss": 1.0148,
      "step": 4371
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7390320897102356,
      "learning_rate": 0.00012599801840056863,
      "loss": 1.0822,
      "step": 4372
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7536359429359436,
      "learning_rate": 0.00012596899344459429,
      "loss": 0.969,
      "step": 4373
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8541299700737,
      "learning_rate": 0.00012593996614246843,
      "loss": 1.0275,
      "step": 4374
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8438847661018372,
      "learning_rate": 0.00012591093649681344,
      "loss": 0.9905,
      "step": 4375
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7539075613021851,
      "learning_rate": 0.00012588190451025207,
      "loss": 1.0476,
      "step": 4376
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8668161630630493,
      "learning_rate": 0.0001258528701854072,
      "loss": 1.0088,
      "step": 4377
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.6120255589485168,
      "learning_rate": 0.00012582383352490188,
      "loss": 0.7642,
      "step": 4378
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8503530621528625,
      "learning_rate": 0.00012579479453135941,
      "loss": 1.1022,
      "step": 4379
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.7852493524551392,
      "learning_rate": 0.00012576575320740328,
      "loss": 1.0352,
      "step": 4380
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8241454362869263,
      "learning_rate": 0.00012573670955565725,
      "loss": 0.8224,
      "step": 4381
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8677917718887329,
      "learning_rate": 0.00012570766357874525,
      "loss": 0.7866,
      "step": 4382
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7602413892745972,
      "learning_rate": 0.00012567861527929138,
      "loss": 0.9296,
      "step": 4383
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7921785712242126,
      "learning_rate": 0.00012564956465992003,
      "loss": 0.9769,
      "step": 4384
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7256081104278564,
      "learning_rate": 0.00012562051172325576,
      "loss": 1.0039,
      "step": 4385
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7994236350059509,
      "learning_rate": 0.00012559145647192326,
      "loss": 1.1076,
      "step": 4386
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7854109406471252,
      "learning_rate": 0.00012556239890854762,
      "loss": 1.0892,
      "step": 4387
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.6456579566001892,
      "learning_rate": 0.00012553333903575394,
      "loss": 0.8171,
      "step": 4388
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7927770018577576,
      "learning_rate": 0.00012550427685616765,
      "loss": 0.9639,
      "step": 4389
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7582933902740479,
      "learning_rate": 0.0001254752123724144,
      "loss": 0.9513,
      "step": 4390
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7228808999061584,
      "learning_rate": 0.00012544614558711987,
      "loss": 0.8841,
      "step": 4391
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7322767376899719,
      "learning_rate": 0.0001254170765029102,
      "loss": 0.8727,
      "step": 4392
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8463506102561951,
      "learning_rate": 0.00012538800512241157,
      "loss": 0.9998,
      "step": 4393
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8110516667366028,
      "learning_rate": 0.00012535893144825043,
      "loss": 1.0143,
      "step": 4394
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.6583903431892395,
      "learning_rate": 0.00012532985548305346,
      "loss": 0.7218,
      "step": 4395
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7003604173660278,
      "learning_rate": 0.0001253007772294474,
      "loss": 0.8849,
      "step": 4396
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7542617321014404,
      "learning_rate": 0.0001252716966900594,
      "loss": 0.9391,
      "step": 4397
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.807867705821991,
      "learning_rate": 0.00012524261386751668,
      "loss": 0.89,
      "step": 4398
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7230579257011414,
      "learning_rate": 0.00012521352876444672,
      "loss": 0.9678,
      "step": 4399
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7677543759346008,
      "learning_rate": 0.0001251844413834772,
      "loss": 1.042,
      "step": 4400
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7520354390144348,
      "learning_rate": 0.000125155351727236,
      "loss": 1.1399,
      "step": 4401
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7441850304603577,
      "learning_rate": 0.0001251262597983512,
      "loss": 0.999,
      "step": 4402
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7426802515983582,
      "learning_rate": 0.0001250971655994511,
      "loss": 1.1166,
      "step": 4403
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8040526509284973,
      "learning_rate": 0.00012506806913316416,
      "loss": 0.9493,
      "step": 4404
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8495841026306152,
      "learning_rate": 0.00012503897040211913,
      "loss": 1.1029,
      "step": 4405
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7909674644470215,
      "learning_rate": 0.0001250098694089449,
      "loss": 1.1178,
      "step": 4406
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7505643367767334,
      "learning_rate": 0.00012498076615627055,
      "loss": 0.9376,
      "step": 4407
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7308511734008789,
      "learning_rate": 0.00012495166064672547,
      "loss": 1.059,
      "step": 4408
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7313070297241211,
      "learning_rate": 0.00012492255288293908,
      "loss": 0.9639,
      "step": 4409
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8354009985923767,
      "learning_rate": 0.00012489344286754114,
      "loss": 1.1099,
      "step": 4410
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8166077733039856,
      "learning_rate": 0.0001248643306031616,
      "loss": 0.9699,
      "step": 4411
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7767633199691772,
      "learning_rate": 0.00012483521609243058,
      "loss": 1.0304,
      "step": 4412
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.750893235206604,
      "learning_rate": 0.00012480609933797837,
      "loss": 1.075,
      "step": 4413
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7821363210678101,
      "learning_rate": 0.0001247769803424356,
      "loss": 0.9453,
      "step": 4414
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7858138680458069,
      "learning_rate": 0.00012474785910843286,
      "loss": 0.921,
      "step": 4415
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7058123350143433,
      "learning_rate": 0.00012471873563860123,
      "loss": 0.9727,
      "step": 4416
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8942750692367554,
      "learning_rate": 0.0001246896099355718,
      "loss": 1.08,
      "step": 4417
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7779878973960876,
      "learning_rate": 0.00012466048200197586,
      "loss": 1.0273,
      "step": 4418
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8175771236419678,
      "learning_rate": 0.00012463135184044504,
      "loss": 0.9516,
      "step": 4419
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7624543905258179,
      "learning_rate": 0.00012460221945361102,
      "loss": 1.0565,
      "step": 4420
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7882060408592224,
      "learning_rate": 0.00012457308484410576,
      "loss": 1.0501,
      "step": 4421
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7546730637550354,
      "learning_rate": 0.00012454394801456143,
      "loss": 1.0056,
      "step": 4422
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0927858352661133,
      "learning_rate": 0.00012451480896761037,
      "loss": 1.0178,
      "step": 4423
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7927306890487671,
      "learning_rate": 0.00012448566770588513,
      "loss": 1.0552,
      "step": 4424
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8408909440040588,
      "learning_rate": 0.00012445652423201845,
      "loss": 1.1028,
      "step": 4425
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7818527221679688,
      "learning_rate": 0.00012442737854864323,
      "loss": 0.9129,
      "step": 4426
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8871914744377136,
      "learning_rate": 0.0001243982306583927,
      "loss": 1.0881,
      "step": 4427
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7353706955909729,
      "learning_rate": 0.0001243690805639002,
      "loss": 0.9453,
      "step": 4428
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7749044895172119,
      "learning_rate": 0.00012433992826779922,
      "loss": 1.053,
      "step": 4429
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.6286710500717163,
      "learning_rate": 0.00012431077377272356,
      "loss": 0.8711,
      "step": 4430
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8130338788032532,
      "learning_rate": 0.00012428161708130707,
      "loss": 0.9351,
      "step": 4431
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.791189432144165,
      "learning_rate": 0.000124252458196184,
      "loss": 1.1517,
      "step": 4432
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8531875014305115,
      "learning_rate": 0.00012422329711998868,
      "loss": 1.031,
      "step": 4433
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9011879563331604,
      "learning_rate": 0.00012419413385535556,
      "loss": 0.9193,
      "step": 4434
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7443050742149353,
      "learning_rate": 0.0001241649684049195,
      "loss": 1.0149,
      "step": 4435
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.885461151599884,
      "learning_rate": 0.0001241358007713153,
      "loss": 0.9496,
      "step": 4436
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.707545816898346,
      "learning_rate": 0.00012410663095717818,
      "loss": 1.14,
      "step": 4437
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7754905223846436,
      "learning_rate": 0.00012407745896514348,
      "loss": 1.0194,
      "step": 4438
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7244762182235718,
      "learning_rate": 0.00012404828479784663,
      "loss": 0.9338,
      "step": 4439
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.837539792060852,
      "learning_rate": 0.00012401910845792345,
      "loss": 0.9798,
      "step": 4440
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8418575525283813,
      "learning_rate": 0.0001239899299480098,
      "loss": 1.0648,
      "step": 4441
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7444628477096558,
      "learning_rate": 0.0001239607492707418,
      "loss": 1.0499,
      "step": 4442
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.6809917092323303,
      "learning_rate": 0.0001239315664287558,
      "loss": 0.8453,
      "step": 4443
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7794191837310791,
      "learning_rate": 0.00012390238142468826,
      "loss": 0.9172,
      "step": 4444
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7316984534263611,
      "learning_rate": 0.00012387319426117587,
      "loss": 0.9582,
      "step": 4445
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7308700084686279,
      "learning_rate": 0.00012384400494085557,
      "loss": 0.9009,
      "step": 4446
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7057145237922668,
      "learning_rate": 0.00012381481346636442,
      "loss": 1.0447,
      "step": 4447
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9074232578277588,
      "learning_rate": 0.0001237856198403397,
      "loss": 0.8815,
      "step": 4448
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8144044876098633,
      "learning_rate": 0.00012375642406541894,
      "loss": 1.1888,
      "step": 4449
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.901542603969574,
      "learning_rate": 0.00012372722614423974,
      "loss": 0.9081,
      "step": 4450
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7050389051437378,
      "learning_rate": 0.00012369802607944,
      "loss": 0.9428,
      "step": 4451
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9148438572883606,
      "learning_rate": 0.0001236688238736578,
      "loss": 1.0899,
      "step": 4452
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.6836315393447876,
      "learning_rate": 0.00012363961952953137,
      "loss": 0.8702,
      "step": 4453
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8242979049682617,
      "learning_rate": 0.0001236104130496992,
      "loss": 1.0303,
      "step": 4454
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7668551206588745,
      "learning_rate": 0.00012358120443679984,
      "loss": 1.0039,
      "step": 4455
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.6840111613273621,
      "learning_rate": 0.0001235519936934722,
      "loss": 0.9588,
      "step": 4456
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7306951284408569,
      "learning_rate": 0.00012352278082235527,
      "loss": 0.9726,
      "step": 4457
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7979377508163452,
      "learning_rate": 0.00012349356582608826,
      "loss": 0.9883,
      "step": 4458
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.6972531676292419,
      "learning_rate": 0.00012346434870731066,
      "loss": 0.9739,
      "step": 4459
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8911920785903931,
      "learning_rate": 0.00012343512946866195,
      "loss": 0.9426,
      "step": 4460
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7012267112731934,
      "learning_rate": 0.00012340590811278198,
      "loss": 0.8533,
      "step": 4461
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7518066167831421,
      "learning_rate": 0.00012337668464231075,
      "loss": 1.028,
      "step": 4462
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7549425959587097,
      "learning_rate": 0.00012334745905988846,
      "loss": 1.0232,
      "step": 4463
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9345147609710693,
      "learning_rate": 0.0001233182313681554,
      "loss": 0.876,
      "step": 4464
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8345490097999573,
      "learning_rate": 0.00012328900156975217,
      "loss": 0.9704,
      "step": 4465
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7318769693374634,
      "learning_rate": 0.00012325976966731948,
      "loss": 0.9204,
      "step": 4466
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7904888987541199,
      "learning_rate": 0.00012323053566349834,
      "loss": 1.0055,
      "step": 4467
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7308286428451538,
      "learning_rate": 0.00012320129956092983,
      "loss": 1.0266,
      "step": 4468
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8149561285972595,
      "learning_rate": 0.00012317206136225524,
      "loss": 1.0419,
      "step": 4469
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8579559922218323,
      "learning_rate": 0.00012314282107011613,
      "loss": 1.0645,
      "step": 4470
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7712220549583435,
      "learning_rate": 0.00012311357868715416,
      "loss": 0.8687,
      "step": 4471
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7389052510261536,
      "learning_rate": 0.00012308433421601123,
      "loss": 0.9033,
      "step": 4472
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7898889780044556,
      "learning_rate": 0.0001230550876593294,
      "loss": 1.0619,
      "step": 4473
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7135383486747742,
      "learning_rate": 0.00012302583901975094,
      "loss": 0.9028,
      "step": 4474
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8847342729568481,
      "learning_rate": 0.0001229965882999183,
      "loss": 1.1626,
      "step": 4475
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.731102466583252,
      "learning_rate": 0.0001229673355024741,
      "loss": 0.9203,
      "step": 4476
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.6908726692199707,
      "learning_rate": 0.0001229380806300612,
      "loss": 0.9012,
      "step": 4477
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7428862452507019,
      "learning_rate": 0.00012290882368532258,
      "loss": 1.1055,
      "step": 4478
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8211758136749268,
      "learning_rate": 0.00012287956467090148,
      "loss": 1.1169,
      "step": 4479
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7881626486778259,
      "learning_rate": 0.00012285030358944122,
      "loss": 0.9172,
      "step": 4480
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8638648986816406,
      "learning_rate": 0.0001228210404435854,
      "loss": 1.0965,
      "step": 4481
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.798534095287323,
      "learning_rate": 0.00012279177523597779,
      "loss": 0.9125,
      "step": 4482
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.6998365521430969,
      "learning_rate": 0.00012276250796926236,
      "loss": 0.8585,
      "step": 4483
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8411529064178467,
      "learning_rate": 0.0001227332386460832,
      "loss": 0.9901,
      "step": 4484
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.812244176864624,
      "learning_rate": 0.00012270396726908467,
      "loss": 0.9175,
      "step": 4485
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7803433537483215,
      "learning_rate": 0.00012267469384091118,
      "loss": 0.9656,
      "step": 4486
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7464903593063354,
      "learning_rate": 0.00012264541836420756,
      "loss": 0.8366,
      "step": 4487
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7667901515960693,
      "learning_rate": 0.00012261614084161857,
      "loss": 0.9332,
      "step": 4488
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7276467084884644,
      "learning_rate": 0.00012258686127578934,
      "loss": 0.7042,
      "step": 4489
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.6728271842002869,
      "learning_rate": 0.00012255757966936502,
      "loss": 0.9618,
      "step": 4490
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8029244542121887,
      "learning_rate": 0.00012252829602499114,
      "loss": 1.1388,
      "step": 4491
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8557717204093933,
      "learning_rate": 0.0001224990103453133,
      "loss": 0.9178,
      "step": 4492
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7355091571807861,
      "learning_rate": 0.0001224697226329772,
      "loss": 1.0869,
      "step": 4493
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7244634628295898,
      "learning_rate": 0.00012244043289062891,
      "loss": 1.0627,
      "step": 4494
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8612757325172424,
      "learning_rate": 0.00012241114112091457,
      "loss": 1.1135,
      "step": 4495
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8004133105278015,
      "learning_rate": 0.00012238184732648048,
      "loss": 1.0435,
      "step": 4496
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7044484615325928,
      "learning_rate": 0.00012235255150997327,
      "loss": 0.904,
      "step": 4497
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8220788240432739,
      "learning_rate": 0.00012232325367403957,
      "loss": 0.9572,
      "step": 4498
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7570661306381226,
      "learning_rate": 0.0001222939538213263,
      "loss": 1.1492,
      "step": 4499
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7427449822425842,
      "learning_rate": 0.00012226465195448055,
      "loss": 0.8829,
      "step": 4500
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7731136679649353,
      "learning_rate": 0.00012223534807614947,
      "loss": 0.889,
      "step": 4501
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8267369270324707,
      "learning_rate": 0.00012220604218898066,
      "loss": 1.0526,
      "step": 4502
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8772794008255005,
      "learning_rate": 0.00012217673429562167,
      "loss": 1.0476,
      "step": 4503
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2616798877716064,
      "learning_rate": 0.0001221474243987203,
      "loss": 0.9864,
      "step": 4504
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7818945050239563,
      "learning_rate": 0.0001221181125009245,
      "loss": 1.0259,
      "step": 4505
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7399562001228333,
      "learning_rate": 0.0001220887986048825,
      "loss": 0.9789,
      "step": 4506
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7645268440246582,
      "learning_rate": 0.00012205948271324259,
      "loss": 1.1027,
      "step": 4507
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8672826290130615,
      "learning_rate": 0.00012203016482865335,
      "loss": 0.8492,
      "step": 4508
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7715070247650146,
      "learning_rate": 0.00012200084495376341,
      "loss": 1.1246,
      "step": 4509
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7514753937721252,
      "learning_rate": 0.00012197152309122173,
      "loss": 1.0028,
      "step": 4510
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.721250057220459,
      "learning_rate": 0.00012194219924367731,
      "loss": 1.0471,
      "step": 4511
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7699748873710632,
      "learning_rate": 0.00012191287341377943,
      "loss": 0.9206,
      "step": 4512
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7520911693572998,
      "learning_rate": 0.00012188354560417755,
      "loss": 1.0336,
      "step": 4513
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.839390754699707,
      "learning_rate": 0.00012185421581752119,
      "loss": 1.0853,
      "step": 4514
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8181332349777222,
      "learning_rate": 0.0001218248840564602,
      "loss": 1.0199,
      "step": 4515
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8653576374053955,
      "learning_rate": 0.00012179555032364447,
      "loss": 0.9699,
      "step": 4516
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7609573602676392,
      "learning_rate": 0.00012176621462172417,
      "loss": 0.8121,
      "step": 4517
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.80059814453125,
      "learning_rate": 0.00012173687695334966,
      "loss": 0.9658,
      "step": 4518
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9648892879486084,
      "learning_rate": 0.00012170753732117137,
      "loss": 1.0759,
      "step": 4519
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7956419587135315,
      "learning_rate": 0.00012167819572783998,
      "loss": 0.9202,
      "step": 4520
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.800743818283081,
      "learning_rate": 0.00012164885217600637,
      "loss": 1.0032,
      "step": 4521
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8416756391525269,
      "learning_rate": 0.00012161950666832153,
      "loss": 0.9712,
      "step": 4522
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8566890358924866,
      "learning_rate": 0.00012159015920743666,
      "loss": 1.1431,
      "step": 4523
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7481673955917358,
      "learning_rate": 0.00012156080979600317,
      "loss": 0.8806,
      "step": 4524
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.707356870174408,
      "learning_rate": 0.0001215314584366726,
      "loss": 0.9942,
      "step": 4525
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.739875078201294,
      "learning_rate": 0.00012150210513209664,
      "loss": 1.0685,
      "step": 4526
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.770443320274353,
      "learning_rate": 0.00012147274988492726,
      "loss": 0.9136,
      "step": 4527
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9301662445068359,
      "learning_rate": 0.00012144339269781647,
      "loss": 0.9608,
      "step": 4528
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7839982509613037,
      "learning_rate": 0.00012141403357341659,
      "loss": 1.0218,
      "step": 4529
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7178850173950195,
      "learning_rate": 0.00012138467251437999,
      "loss": 0.9524,
      "step": 4530
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8009347915649414,
      "learning_rate": 0.00012135530952335931,
      "loss": 1.0965,
      "step": 4531
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8162659406661987,
      "learning_rate": 0.00012132594460300737,
      "loss": 1.0292,
      "step": 4532
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.685593843460083,
      "learning_rate": 0.00012129657775597705,
      "loss": 1.0476,
      "step": 4533
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9040690064430237,
      "learning_rate": 0.00012126720898492155,
      "loss": 1.034,
      "step": 4534
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.797566831111908,
      "learning_rate": 0.00012123783829249411,
      "loss": 1.0243,
      "step": 4535
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7130178213119507,
      "learning_rate": 0.00012120846568134819,
      "loss": 0.8894,
      "step": 4536
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7870303392410278,
      "learning_rate": 0.0001211790911541375,
      "loss": 1.0783,
      "step": 4537
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8182770013809204,
      "learning_rate": 0.00012114971471351588,
      "loss": 0.9987,
      "step": 4538
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8288918733596802,
      "learning_rate": 0.00012112033636213725,
      "loss": 1.0125,
      "step": 4539
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.674735963344574,
      "learning_rate": 0.00012109095610265582,
      "loss": 0.9747,
      "step": 4540
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.752810001373291,
      "learning_rate": 0.00012106157393772588,
      "loss": 1.0043,
      "step": 4541
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.737766683101654,
      "learning_rate": 0.00012103218987000205,
      "loss": 0.9728,
      "step": 4542
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7705424427986145,
      "learning_rate": 0.00012100280390213894,
      "loss": 0.9582,
      "step": 4543
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.6810630559921265,
      "learning_rate": 0.00012097341603679137,
      "loss": 0.891,
      "step": 4544
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.6789401173591614,
      "learning_rate": 0.00012094402627661447,
      "loss": 0.926,
      "step": 4545
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7720015048980713,
      "learning_rate": 0.00012091463462426334,
      "loss": 1.0054,
      "step": 4546
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7323508262634277,
      "learning_rate": 0.00012088524108239338,
      "loss": 1.0275,
      "step": 4547
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7956165671348572,
      "learning_rate": 0.0001208558456536602,
      "loss": 1.1275,
      "step": 4548
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8188318014144897,
      "learning_rate": 0.00012082644834071939,
      "loss": 1.0057,
      "step": 4549
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8520888090133667,
      "learning_rate": 0.00012079704914622691,
      "loss": 0.986,
      "step": 4550
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7133857011795044,
      "learning_rate": 0.0001207676480728388,
      "loss": 0.9289,
      "step": 4551
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7838119268417358,
      "learning_rate": 0.00012073824512321126,
      "loss": 0.8795,
      "step": 4552
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8384571075439453,
      "learning_rate": 0.00012070884030000071,
      "loss": 1.0691,
      "step": 4553
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7315096855163574,
      "learning_rate": 0.00012067943360586369,
      "loss": 0.9933,
      "step": 4554
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9339639544487,
      "learning_rate": 0.00012065002504345692,
      "loss": 0.97,
      "step": 4555
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.869044840335846,
      "learning_rate": 0.0001206206146154373,
      "loss": 1.0211,
      "step": 4556
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7361880540847778,
      "learning_rate": 0.00012059120232446191,
      "loss": 0.7521,
      "step": 4557
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.767719030380249,
      "learning_rate": 0.000120561788173188,
      "loss": 1.1313,
      "step": 4558
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7440416812896729,
      "learning_rate": 0.00012053237216427293,
      "loss": 0.9252,
      "step": 4559
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8230945467948914,
      "learning_rate": 0.00012050295430037428,
      "loss": 1.0488,
      "step": 4560
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7725455164909363,
      "learning_rate": 0.00012047353458414981,
      "loss": 0.8682,
      "step": 4561
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7603073120117188,
      "learning_rate": 0.00012044411301825742,
      "loss": 0.9021,
      "step": 4562
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9435483813285828,
      "learning_rate": 0.00012041468960535516,
      "loss": 0.9668,
      "step": 4563
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7270300388336182,
      "learning_rate": 0.00012038526434810129,
      "loss": 0.9416,
      "step": 4564
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7929261326789856,
      "learning_rate": 0.00012035583724915418,
      "loss": 1.0635,
      "step": 4565
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.6248114109039307,
      "learning_rate": 0.00012032640831117243,
      "loss": 0.7921,
      "step": 4566
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8861623406410217,
      "learning_rate": 0.00012029697753681482,
      "loss": 1.1924,
      "step": 4567
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.6975063681602478,
      "learning_rate": 0.00012026754492874017,
      "loss": 0.9724,
      "step": 4568
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7411131262779236,
      "learning_rate": 0.00012023811048960763,
      "loss": 0.902,
      "step": 4569
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7723414301872253,
      "learning_rate": 0.00012020867422207637,
      "loss": 0.96,
      "step": 4570
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8154706358909607,
      "learning_rate": 0.00012017923612880579,
      "loss": 0.975,
      "step": 4571
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7440472841262817,
      "learning_rate": 0.00012014979621245552,
      "loss": 1.1156,
      "step": 4572
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.014880895614624,
      "learning_rate": 0.00012012035447568524,
      "loss": 0.9663,
      "step": 4573
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7933245301246643,
      "learning_rate": 0.00012009091092115486,
      "loss": 0.9387,
      "step": 4574
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8018094301223755,
      "learning_rate": 0.00012006146555152442,
      "loss": 0.9435,
      "step": 4575
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7780526280403137,
      "learning_rate": 0.00012003201836945414,
      "loss": 0.8845,
      "step": 4576
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8302222490310669,
      "learning_rate": 0.00012000256937760445,
      "loss": 1.0314,
      "step": 4577
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7264339923858643,
      "learning_rate": 0.00011997311857863587,
      "loss": 1.0435,
      "step": 4578
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8215857744216919,
      "learning_rate": 0.0001199436659752091,
      "loss": 1.1421,
      "step": 4579
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7523236274719238,
      "learning_rate": 0.00011991421156998506,
      "loss": 0.8928,
      "step": 4580
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.6750524044036865,
      "learning_rate": 0.00011988475536562471,
      "loss": 0.8169,
      "step": 4581
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7918744683265686,
      "learning_rate": 0.00011985529736478931,
      "loss": 1.1446,
      "step": 4582
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7331019639968872,
      "learning_rate": 0.00011982583757014025,
      "loss": 0.8463,
      "step": 4583
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8649581670761108,
      "learning_rate": 0.00011979637598433899,
      "loss": 0.8682,
      "step": 4584
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8099588751792908,
      "learning_rate": 0.00011976691261004727,
      "loss": 1.1053,
      "step": 4585
    },
    {
      "epoch": 1.73,
      "eval_loss": 0.9523382782936096,
      "eval_runtime": 17.757,
      "eval_samples_per_second": 153.686,
      "eval_steps_per_second": 38.464,
      "step": 4585
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8093003630638123,
      "learning_rate": 0.00011973744744992688,
      "loss": 0.8328,
      "step": 4586
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7210092544555664,
      "learning_rate": 0.00011970798050663987,
      "loss": 0.8347,
      "step": 4587
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8135833144187927,
      "learning_rate": 0.00011967851178284843,
      "loss": 1.0053,
      "step": 4588
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7191389203071594,
      "learning_rate": 0.00011964904128121486,
      "loss": 0.9883,
      "step": 4589
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7279300689697266,
      "learning_rate": 0.00011961956900440168,
      "loss": 0.9336,
      "step": 4590
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.6938570141792297,
      "learning_rate": 0.0001195900949550715,
      "loss": 0.9598,
      "step": 4591
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7962003350257874,
      "learning_rate": 0.00011956061913588716,
      "loss": 0.9078,
      "step": 4592
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7468202114105225,
      "learning_rate": 0.00011953114154951166,
      "loss": 0.7303,
      "step": 4593
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7891976237297058,
      "learning_rate": 0.0001195016621986081,
      "loss": 0.9259,
      "step": 4594
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0634568929672241,
      "learning_rate": 0.00011947218108583973,
      "loss": 0.5085,
      "step": 4595
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8258597254753113,
      "learning_rate": 0.0001194426982138701,
      "loss": 0.8869,
      "step": 4596
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7219350337982178,
      "learning_rate": 0.00011941321358536276,
      "loss": 0.8407,
      "step": 4597
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9141265749931335,
      "learning_rate": 0.0001193837272029815,
      "loss": 1.0333,
      "step": 4598
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7562088966369629,
      "learning_rate": 0.00011935423906939023,
      "loss": 0.9877,
      "step": 4599
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8154380321502686,
      "learning_rate": 0.00011932474918725303,
      "loss": 0.8455,
      "step": 4600
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8079681396484375,
      "learning_rate": 0.00011929525755923417,
      "loss": 1.0385,
      "step": 4601
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7577683925628662,
      "learning_rate": 0.00011926576418799804,
      "loss": 1.1408,
      "step": 4602
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8447030186653137,
      "learning_rate": 0.0001192362690762092,
      "loss": 0.9535,
      "step": 4603
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.6743953227996826,
      "learning_rate": 0.00011920677222653236,
      "loss": 0.7593,
      "step": 4604
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9185356497764587,
      "learning_rate": 0.00011917727364163238,
      "loss": 0.9552,
      "step": 4605
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8450148701667786,
      "learning_rate": 0.00011914777332417433,
      "loss": 0.9634,
      "step": 4606
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8802364468574524,
      "learning_rate": 0.00011911827127682337,
      "loss": 0.7896,
      "step": 4607
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8088971972465515,
      "learning_rate": 0.00011908876750224485,
      "loss": 0.8993,
      "step": 4608
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7739360332489014,
      "learning_rate": 0.00011905926200310427,
      "loss": 0.9611,
      "step": 4609
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9451324343681335,
      "learning_rate": 0.00011902975478206727,
      "loss": 1.0925,
      "step": 4610
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8480505347251892,
      "learning_rate": 0.00011900024584179966,
      "loss": 1.1357,
      "step": 4611
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7975031733512878,
      "learning_rate": 0.00011897073518496744,
      "loss": 1.024,
      "step": 4612
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7346519231796265,
      "learning_rate": 0.00011894122281423672,
      "loss": 0.7701,
      "step": 4613
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7548093795776367,
      "learning_rate": 0.00011891170873227373,
      "loss": 0.9828,
      "step": 4614
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7536081671714783,
      "learning_rate": 0.00011888219294174494,
      "loss": 0.9656,
      "step": 4615
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.837548553943634,
      "learning_rate": 0.00011885267544531695,
      "loss": 1.024,
      "step": 4616
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.865839421749115,
      "learning_rate": 0.00011882315624565645,
      "loss": 0.9148,
      "step": 4617
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7667220830917358,
      "learning_rate": 0.0001187936353454304,
      "loss": 0.9866,
      "step": 4618
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.6959596872329712,
      "learning_rate": 0.00011876411274730577,
      "loss": 0.9945,
      "step": 4619
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7026944160461426,
      "learning_rate": 0.00011873458845394979,
      "loss": 0.8613,
      "step": 4620
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.736354649066925,
      "learning_rate": 0.00011870506246802984,
      "loss": 0.8945,
      "step": 4621
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9902938604354858,
      "learning_rate": 0.00011867553479221339,
      "loss": 0.9826,
      "step": 4622
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7676527500152588,
      "learning_rate": 0.00011864600542916813,
      "loss": 0.9257,
      "step": 4623
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.6731281280517578,
      "learning_rate": 0.00011861647438156183,
      "loss": 0.8324,
      "step": 4624
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8030645251274109,
      "learning_rate": 0.00011858694165206248,
      "loss": 1.0004,
      "step": 4625
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7600473761558533,
      "learning_rate": 0.00011855740724333821,
      "loss": 0.9411,
      "step": 4626
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.6938033699989319,
      "learning_rate": 0.00011852787115805726,
      "loss": 0.8006,
      "step": 4627
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7645683884620667,
      "learning_rate": 0.00011849833339888805,
      "loss": 0.9532,
      "step": 4628
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7587887644767761,
      "learning_rate": 0.00011846879396849917,
      "loss": 0.9675,
      "step": 4629
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9182590842247009,
      "learning_rate": 0.00011843925286955929,
      "loss": 0.9558,
      "step": 4630
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7563308477401733,
      "learning_rate": 0.00011840971010473734,
      "loss": 0.9988,
      "step": 4631
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.6797767281532288,
      "learning_rate": 0.00011838016567670231,
      "loss": 0.8188,
      "step": 4632
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7414708733558655,
      "learning_rate": 0.00011835061958812339,
      "loss": 0.9559,
      "step": 4633
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7425746321678162,
      "learning_rate": 0.0001183210718416699,
      "loss": 1.1321,
      "step": 4634
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7789604067802429,
      "learning_rate": 0.00011829152244001127,
      "loss": 0.8989,
      "step": 4635
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8125724196434021,
      "learning_rate": 0.00011826197138581719,
      "loss": 0.9971,
      "step": 4636
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7495322227478027,
      "learning_rate": 0.00011823241868175738,
      "loss": 0.9785,
      "step": 4637
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7192978858947754,
      "learning_rate": 0.00011820286433050176,
      "loss": 0.974,
      "step": 4638
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7956582307815552,
      "learning_rate": 0.00011817330833472042,
      "loss": 0.8802,
      "step": 4639
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8172240257263184,
      "learning_rate": 0.00011814375069708357,
      "loss": 1.0455,
      "step": 4640
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.7698330283164978,
      "learning_rate": 0.00011811419142026156,
      "loss": 0.9873,
      "step": 4641
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8238794207572937,
      "learning_rate": 0.00011808463050692495,
      "loss": 1.0997,
      "step": 4642
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8221603035926819,
      "learning_rate": 0.00011805506795974435,
      "loss": 1.038,
      "step": 4643
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8607699871063232,
      "learning_rate": 0.0001180255037813906,
      "loss": 0.9641,
      "step": 4644
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7579261064529419,
      "learning_rate": 0.00011799593797453462,
      "loss": 0.942,
      "step": 4645
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8301036357879639,
      "learning_rate": 0.00011796637054184753,
      "loss": 0.9381,
      "step": 4646
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7816101908683777,
      "learning_rate": 0.00011793680148600062,
      "loss": 0.9926,
      "step": 4647
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8143907189369202,
      "learning_rate": 0.00011790723080966524,
      "loss": 1.1058,
      "step": 4648
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7717945575714111,
      "learning_rate": 0.00011787765851551296,
      "loss": 1.13,
      "step": 4649
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8154736757278442,
      "learning_rate": 0.00011784808460621541,
      "loss": 1.0087,
      "step": 4650
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7709436416625977,
      "learning_rate": 0.0001178185090844445,
      "loss": 0.869,
      "step": 4651
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7582793831825256,
      "learning_rate": 0.00011778893195287222,
      "loss": 1.0573,
      "step": 4652
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8012235760688782,
      "learning_rate": 0.00011775935321417062,
      "loss": 0.8933,
      "step": 4653
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7932279109954834,
      "learning_rate": 0.000117729772871012,
      "loss": 1.0786,
      "step": 4654
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9135048985481262,
      "learning_rate": 0.00011770019092606879,
      "loss": 1.0244,
      "step": 4655
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8108317852020264,
      "learning_rate": 0.00011767060738201358,
      "loss": 1.0483,
      "step": 4656
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.808539867401123,
      "learning_rate": 0.000117641022241519,
      "loss": 0.8809,
      "step": 4657
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8646968603134155,
      "learning_rate": 0.00011761143550725799,
      "loss": 0.9518,
      "step": 4658
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7806581258773804,
      "learning_rate": 0.00011758184718190346,
      "loss": 1.0036,
      "step": 4659
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.6390308737754822,
      "learning_rate": 0.0001175522572681286,
      "loss": 0.7685,
      "step": 4660
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7053999304771423,
      "learning_rate": 0.00011752266576860668,
      "loss": 0.9902,
      "step": 4661
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7349623441696167,
      "learning_rate": 0.00011749307268601111,
      "loss": 0.9056,
      "step": 4662
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7815699577331543,
      "learning_rate": 0.0001174634780230155,
      "loss": 0.9977,
      "step": 4663
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7745949029922485,
      "learning_rate": 0.00011743388178229349,
      "loss": 1.0299,
      "step": 4664
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8292572498321533,
      "learning_rate": 0.00011740428396651896,
      "loss": 1.0306,
      "step": 4665
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8445864319801331,
      "learning_rate": 0.00011737468457836598,
      "loss": 1.0174,
      "step": 4666
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7877253890037537,
      "learning_rate": 0.00011734508362050856,
      "loss": 0.904,
      "step": 4667
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8767524361610413,
      "learning_rate": 0.00011731548109562109,
      "loss": 1.0872,
      "step": 4668
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7433600425720215,
      "learning_rate": 0.00011728587700637792,
      "loss": 0.8279,
      "step": 4669
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7911831736564636,
      "learning_rate": 0.0001172562713554536,
      "loss": 1.0507,
      "step": 4670
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7265031933784485,
      "learning_rate": 0.00011722666414552293,
      "loss": 0.9152,
      "step": 4671
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.757936954498291,
      "learning_rate": 0.00011719705537926069,
      "loss": 0.995,
      "step": 4672
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7522801160812378,
      "learning_rate": 0.00011716744505934182,
      "loss": 0.7812,
      "step": 4673
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.6782301664352417,
      "learning_rate": 0.00011713783318844154,
      "loss": 0.9455,
      "step": 4674
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7656162977218628,
      "learning_rate": 0.000117108219769235,
      "loss": 0.9878,
      "step": 4675
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8663710355758667,
      "learning_rate": 0.00011707860480439772,
      "loss": 1.1014,
      "step": 4676
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7111193537712097,
      "learning_rate": 0.00011704898829660518,
      "loss": 0.9375,
      "step": 4677
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7736416459083557,
      "learning_rate": 0.00011701937024853308,
      "loss": 0.8987,
      "step": 4678
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7856674194335938,
      "learning_rate": 0.00011698975066285727,
      "loss": 1.0611,
      "step": 4679
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7942124009132385,
      "learning_rate": 0.00011696012954225364,
      "loss": 1.0369,
      "step": 4680
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7673232555389404,
      "learning_rate": 0.00011693050688939833,
      "loss": 0.8537,
      "step": 4681
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8419005870819092,
      "learning_rate": 0.00011690088270696761,
      "loss": 0.9412,
      "step": 4682
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8815070390701294,
      "learning_rate": 0.00011687125699763781,
      "loss": 1.0334,
      "step": 4683
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7238216400146484,
      "learning_rate": 0.0001168416297640855,
      "loss": 0.9714,
      "step": 4684
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8091062307357788,
      "learning_rate": 0.00011681200100898727,
      "loss": 0.9395,
      "step": 4685
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8089106678962708,
      "learning_rate": 0.00011678237073501992,
      "loss": 1.0512,
      "step": 4686
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7475075721740723,
      "learning_rate": 0.00011675273894486046,
      "loss": 0.8836,
      "step": 4687
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.825634241104126,
      "learning_rate": 0.00011672310564118585,
      "loss": 0.9447,
      "step": 4688
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8098149299621582,
      "learning_rate": 0.00011669347082667332,
      "loss": 0.9268,
      "step": 4689
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7696269154548645,
      "learning_rate": 0.00011666383450400022,
      "loss": 0.9186,
      "step": 4690
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7242420315742493,
      "learning_rate": 0.00011663419667584404,
      "loss": 0.992,
      "step": 4691
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.816675066947937,
      "learning_rate": 0.00011660455734488236,
      "loss": 1.1345,
      "step": 4692
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7482806444168091,
      "learning_rate": 0.00011657491651379296,
      "loss": 0.9011,
      "step": 4693
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.709613561630249,
      "learning_rate": 0.00011654527418525367,
      "loss": 0.9484,
      "step": 4694
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7983031272888184,
      "learning_rate": 0.00011651563036194253,
      "loss": 1.0922,
      "step": 4695
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.7807701826095581,
      "learning_rate": 0.00011648598504653774,
      "loss": 0.9652,
      "step": 4696
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.721126914024353,
      "learning_rate": 0.00011645633824171751,
      "loss": 0.9591,
      "step": 4697
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7826765179634094,
      "learning_rate": 0.00011642668995016033,
      "loss": 0.9512,
      "step": 4698
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7089790105819702,
      "learning_rate": 0.0001163970401745447,
      "loss": 0.7106,
      "step": 4699
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.77752685546875,
      "learning_rate": 0.00011636738891754931,
      "loss": 0.9311,
      "step": 4700
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7967430949211121,
      "learning_rate": 0.00011633773618185302,
      "loss": 1.1004,
      "step": 4701
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7402793765068054,
      "learning_rate": 0.00011630808197013477,
      "loss": 0.8836,
      "step": 4702
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.6968073844909668,
      "learning_rate": 0.00011627842628507366,
      "loss": 1.0618,
      "step": 4703
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.743408739566803,
      "learning_rate": 0.00011624876912934892,
      "loss": 0.9196,
      "step": 4704
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9421427249908447,
      "learning_rate": 0.00011621911050563983,
      "loss": 1.0486,
      "step": 4705
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8492775559425354,
      "learning_rate": 0.00011618945041662599,
      "loss": 1.31,
      "step": 4706
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8934208750724792,
      "learning_rate": 0.00011615978886498695,
      "loss": 0.9059,
      "step": 4707
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8813003301620483,
      "learning_rate": 0.00011613012585340249,
      "loss": 0.9019,
      "step": 4708
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.924180269241333,
      "learning_rate": 0.00011610046138455249,
      "loss": 0.9498,
      "step": 4709
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7572709321975708,
      "learning_rate": 0.00011607079546111692,
      "loss": 0.8413,
      "step": 4710
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7872298359870911,
      "learning_rate": 0.00011604112808577603,
      "loss": 0.81,
      "step": 4711
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8559049963951111,
      "learning_rate": 0.00011601145926121005,
      "loss": 0.745,
      "step": 4712
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7605762481689453,
      "learning_rate": 0.00011598178899009933,
      "loss": 0.8235,
      "step": 4713
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.768850564956665,
      "learning_rate": 0.0001159521172751245,
      "loss": 0.8782,
      "step": 4714
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9820607304573059,
      "learning_rate": 0.00011592244411896618,
      "loss": 1.1446,
      "step": 4715
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7328512668609619,
      "learning_rate": 0.00011589276952430517,
      "loss": 0.9038,
      "step": 4716
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8597776889801025,
      "learning_rate": 0.00011586309349382245,
      "loss": 0.9516,
      "step": 4717
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8076006770133972,
      "learning_rate": 0.00011583341603019903,
      "loss": 0.85,
      "step": 4718
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9194772243499756,
      "learning_rate": 0.00011580373713611615,
      "loss": 1.0622,
      "step": 4719
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7424063682556152,
      "learning_rate": 0.00011577405681425506,
      "loss": 0.8937,
      "step": 4720
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.737322986125946,
      "learning_rate": 0.00011574437506729726,
      "loss": 1.0043,
      "step": 4721
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9296445250511169,
      "learning_rate": 0.00011571469189792433,
      "loss": 0.8489,
      "step": 4722
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7422108054161072,
      "learning_rate": 0.00011568500730881795,
      "loss": 0.9638,
      "step": 4723
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7797747850418091,
      "learning_rate": 0.00011565532130265999,
      "loss": 0.9662,
      "step": 4724
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8456657528877258,
      "learning_rate": 0.00011562563388213236,
      "loss": 0.8727,
      "step": 4725
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7890819907188416,
      "learning_rate": 0.0001155959450499172,
      "loss": 0.9409,
      "step": 4726
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9545003175735474,
      "learning_rate": 0.0001155662548086967,
      "loss": 0.9497,
      "step": 4727
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9436349868774414,
      "learning_rate": 0.00011553656316115325,
      "loss": 1.0286,
      "step": 4728
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7221186757087708,
      "learning_rate": 0.00011550687010996924,
      "loss": 0.9814,
      "step": 4729
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7962702512741089,
      "learning_rate": 0.00011547717565782732,
      "loss": 0.9702,
      "step": 4730
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8563812971115112,
      "learning_rate": 0.00011544747980741021,
      "loss": 0.9316,
      "step": 4731
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.6784048080444336,
      "learning_rate": 0.0001154177825614008,
      "loss": 0.8695,
      "step": 4732
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.6901435256004333,
      "learning_rate": 0.00011538808392248199,
      "loss": 0.8863,
      "step": 4733
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7063350677490234,
      "learning_rate": 0.00011535838389333694,
      "loss": 0.9459,
      "step": 4734
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8247732520103455,
      "learning_rate": 0.00011532868247664885,
      "loss": 0.9109,
      "step": 4735
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8078949451446533,
      "learning_rate": 0.00011529897967510113,
      "loss": 1.0646,
      "step": 4736
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8562504053115845,
      "learning_rate": 0.00011526927549137716,
      "loss": 1.0633,
      "step": 4737
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7846590280532837,
      "learning_rate": 0.00011523956992816065,
      "loss": 1.0485,
      "step": 4738
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8158285617828369,
      "learning_rate": 0.00011520986298813527,
      "loss": 0.9678,
      "step": 4739
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.6645245552062988,
      "learning_rate": 0.00011518015467398489,
      "loss": 0.9141,
      "step": 4740
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7792890667915344,
      "learning_rate": 0.00011515044498839347,
      "loss": 0.8894,
      "step": 4741
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7953566312789917,
      "learning_rate": 0.00011512073393404514,
      "loss": 1.0201,
      "step": 4742
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.759485125541687,
      "learning_rate": 0.00011509102151362412,
      "loss": 1.0014,
      "step": 4743
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7176987528800964,
      "learning_rate": 0.00011506130772981476,
      "loss": 0.9569,
      "step": 4744
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7796375751495361,
      "learning_rate": 0.00011503159258530147,
      "loss": 0.9428,
      "step": 4745
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8124610185623169,
      "learning_rate": 0.00011500187608276893,
      "loss": 0.9826,
      "step": 4746
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7761412858963013,
      "learning_rate": 0.00011497215822490186,
      "loss": 0.7457,
      "step": 4747
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.6975809931755066,
      "learning_rate": 0.00011494243901438501,
      "loss": 0.8538,
      "step": 4748
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8496014475822449,
      "learning_rate": 0.00011491271845390345,
      "loss": 1.0841,
      "step": 4749
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7893025875091553,
      "learning_rate": 0.00011488299654614219,
      "loss": 1.0616,
      "step": 4750
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.6957244277000427,
      "learning_rate": 0.00011485327329378644,
      "loss": 0.8699,
      "step": 4751
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7174831032752991,
      "learning_rate": 0.00011482354869952156,
      "loss": 0.851,
      "step": 4752
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7984290719032288,
      "learning_rate": 0.000114793822766033,
      "loss": 1.0114,
      "step": 4753
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7365847826004028,
      "learning_rate": 0.00011476409549600631,
      "loss": 1.0527,
      "step": 4754
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7391736507415771,
      "learning_rate": 0.00011473436689212718,
      "loss": 1.0829,
      "step": 4755
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7169732451438904,
      "learning_rate": 0.00011470463695708142,
      "loss": 1.0403,
      "step": 4756
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8890628218650818,
      "learning_rate": 0.00011467490569355499,
      "loss": 0.8923,
      "step": 4757
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7755351066589355,
      "learning_rate": 0.00011464517310423391,
      "loss": 0.9588,
      "step": 4758
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7561038136482239,
      "learning_rate": 0.00011461543919180436,
      "loss": 0.8168,
      "step": 4759
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.858525276184082,
      "learning_rate": 0.0001145857039589526,
      "loss": 1.0978,
      "step": 4760
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7823034524917603,
      "learning_rate": 0.00011455596740836512,
      "loss": 0.9898,
      "step": 4761
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7411243915557861,
      "learning_rate": 0.00011452622954272838,
      "loss": 0.8592,
      "step": 4762
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8100503087043762,
      "learning_rate": 0.00011449649036472907,
      "loss": 0.9123,
      "step": 4763
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8389270305633545,
      "learning_rate": 0.0001144667498770539,
      "loss": 0.9103,
      "step": 4764
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8715300559997559,
      "learning_rate": 0.0001144370080823898,
      "loss": 1.0515,
      "step": 4765
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7054761052131653,
      "learning_rate": 0.00011440726498342376,
      "loss": 0.9034,
      "step": 4766
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7008001804351807,
      "learning_rate": 0.0001143775205828429,
      "loss": 0.9998,
      "step": 4767
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7793424725532532,
      "learning_rate": 0.00011434777488333448,
      "loss": 1.0721,
      "step": 4768
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7706813216209412,
      "learning_rate": 0.00011431802788758581,
      "loss": 0.9,
      "step": 4769
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7926739454269409,
      "learning_rate": 0.00011428827959828441,
      "loss": 1.1088,
      "step": 4770
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7428482174873352,
      "learning_rate": 0.00011425853001811784,
      "loss": 0.999,
      "step": 4771
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8474048972129822,
      "learning_rate": 0.00011422877914977381,
      "loss": 0.9764,
      "step": 4772
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9430420994758606,
      "learning_rate": 0.00011419902699594016,
      "loss": 1.2097,
      "step": 4773
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.902398943901062,
      "learning_rate": 0.0001141692735593048,
      "loss": 0.9944,
      "step": 4774
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.7506820559501648,
      "learning_rate": 0.0001141395188425558,
      "loss": 1.1252,
      "step": 4775
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7798844575881958,
      "learning_rate": 0.00011410976284838135,
      "loss": 1.0435,
      "step": 4776
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7502029538154602,
      "learning_rate": 0.0001140800055794697,
      "loss": 1.0445,
      "step": 4777
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7140774726867676,
      "learning_rate": 0.00011405024703850929,
      "loss": 1.0022,
      "step": 4778
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.6743187308311462,
      "learning_rate": 0.0001140204872281886,
      "loss": 0.9194,
      "step": 4779
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8572053909301758,
      "learning_rate": 0.00011399072615119623,
      "loss": 0.9496,
      "step": 4780
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7410863041877747,
      "learning_rate": 0.00011396096381022103,
      "loss": 0.9508,
      "step": 4781
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7025617957115173,
      "learning_rate": 0.00011393120020795178,
      "loss": 0.7898,
      "step": 4782
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7673794627189636,
      "learning_rate": 0.00011390143534707744,
      "loss": 0.9508,
      "step": 4783
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8125484585762024,
      "learning_rate": 0.00011387166923028716,
      "loss": 1.0228,
      "step": 4784
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7624559998512268,
      "learning_rate": 0.00011384190186027007,
      "loss": 0.8023,
      "step": 4785
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7645179629325867,
      "learning_rate": 0.00011381213323971555,
      "loss": 0.9577,
      "step": 4786
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8080947995185852,
      "learning_rate": 0.00011378236337131301,
      "loss": 0.9995,
      "step": 4787
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.883863627910614,
      "learning_rate": 0.00011375259225775192,
      "loss": 0.9384,
      "step": 4788
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7815911769866943,
      "learning_rate": 0.00011372281990172204,
      "loss": 0.9968,
      "step": 4789
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7550731897354126,
      "learning_rate": 0.00011369304630591304,
      "loss": 0.8006,
      "step": 4790
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2814509868621826,
      "learning_rate": 0.00011366327147301486,
      "loss": 1.0837,
      "step": 4791
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7658206224441528,
      "learning_rate": 0.00011363349540571747,
      "loss": 0.9169,
      "step": 4792
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7140726447105408,
      "learning_rate": 0.00011360371810671094,
      "loss": 0.8128,
      "step": 4793
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.785155177116394,
      "learning_rate": 0.00011357393957868554,
      "loss": 1.0296,
      "step": 4794
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8359742164611816,
      "learning_rate": 0.00011354415982433151,
      "loss": 0.9683,
      "step": 4795
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7292312383651733,
      "learning_rate": 0.00011351437884633933,
      "loss": 0.8378,
      "step": 4796
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7945870757102966,
      "learning_rate": 0.00011348459664739956,
      "loss": 1.2087,
      "step": 4797
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8345226645469666,
      "learning_rate": 0.00011345481323020287,
      "loss": 0.9874,
      "step": 4798
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7681534886360168,
      "learning_rate": 0.0001134250285974399,
      "loss": 0.9763,
      "step": 4799
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8912960886955261,
      "learning_rate": 0.00011339524275180167,
      "loss": 1.0071,
      "step": 4800
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8802440762519836,
      "learning_rate": 0.00011336545569597908,
      "loss": 1.0812,
      "step": 4801
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.6936193108558655,
      "learning_rate": 0.00011333566743266324,
      "loss": 0.8666,
      "step": 4802
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7325948476791382,
      "learning_rate": 0.00011330587796454537,
      "loss": 0.9684,
      "step": 4803
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7145864963531494,
      "learning_rate": 0.00011327608729431673,
      "loss": 0.948,
      "step": 4804
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7360267043113708,
      "learning_rate": 0.0001132462954246688,
      "loss": 0.8956,
      "step": 4805
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7906669974327087,
      "learning_rate": 0.00011321650235829309,
      "loss": 0.9409,
      "step": 4806
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7279953956604004,
      "learning_rate": 0.00011318670809788119,
      "loss": 0.9309,
      "step": 4807
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7352005839347839,
      "learning_rate": 0.00011315691264612492,
      "loss": 0.9411,
      "step": 4808
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7934585809707642,
      "learning_rate": 0.00011312711600571604,
      "loss": 0.9128,
      "step": 4809
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8576295375823975,
      "learning_rate": 0.00011309731817934656,
      "loss": 0.9998,
      "step": 4810
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7300700545310974,
      "learning_rate": 0.00011306751916970857,
      "loss": 0.9852,
      "step": 4811
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8238819241523743,
      "learning_rate": 0.0001130377189794942,
      "loss": 0.9775,
      "step": 4812
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7973847985267639,
      "learning_rate": 0.00011300791761139575,
      "loss": 0.9101,
      "step": 4813
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8104704022407532,
      "learning_rate": 0.00011297811506810556,
      "loss": 1.1568,
      "step": 4814
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7424970269203186,
      "learning_rate": 0.00011294831135231617,
      "loss": 0.9855,
      "step": 4815
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7911181449890137,
      "learning_rate": 0.00011291850646672017,
      "loss": 0.9468,
      "step": 4816
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7143746018409729,
      "learning_rate": 0.00011288870041401025,
      "loss": 0.9719,
      "step": 4817
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8238941431045532,
      "learning_rate": 0.00011285889319687923,
      "loss": 0.9534,
      "step": 4818
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7150064706802368,
      "learning_rate": 0.00011282908481802004,
      "loss": 1.0732,
      "step": 4819
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.6949624419212341,
      "learning_rate": 0.00011279927528012561,
      "loss": 0.8743,
      "step": 4820
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.858025074005127,
      "learning_rate": 0.00011276946458588917,
      "loss": 0.9061,
      "step": 4821
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8336944580078125,
      "learning_rate": 0.00011273965273800391,
      "loss": 0.8504,
      "step": 4822
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8705890774726868,
      "learning_rate": 0.00011270983973916314,
      "loss": 1.0703,
      "step": 4823
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9273001551628113,
      "learning_rate": 0.00011268002559206034,
      "loss": 1.037,
      "step": 4824
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8158978223800659,
      "learning_rate": 0.00011265021029938897,
      "loss": 0.9633,
      "step": 4825
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7523478269577026,
      "learning_rate": 0.00011262039386384277,
      "loss": 0.9112,
      "step": 4826
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.7214547395706177,
      "learning_rate": 0.00011259057628811542,
      "loss": 1.0415,
      "step": 4827
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7658652663230896,
      "learning_rate": 0.00011256075757490081,
      "loss": 0.8409,
      "step": 4828
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7982742190361023,
      "learning_rate": 0.00011253093772689288,
      "loss": 1.036,
      "step": 4829
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7945194244384766,
      "learning_rate": 0.00011250111674678563,
      "loss": 0.8592,
      "step": 4830
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7518361806869507,
      "learning_rate": 0.00011247129463727328,
      "loss": 0.9347,
      "step": 4831
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8912938833236694,
      "learning_rate": 0.00011244147140105007,
      "loss": 0.8304,
      "step": 4832
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.731968104839325,
      "learning_rate": 0.00011241164704081038,
      "loss": 1.009,
      "step": 4833
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8283110857009888,
      "learning_rate": 0.00011238182155924866,
      "loss": 0.9884,
      "step": 4834
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7054871320724487,
      "learning_rate": 0.00011235199495905943,
      "loss": 1.0458,
      "step": 4835
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.6782692670822144,
      "learning_rate": 0.00011232216724293742,
      "loss": 0.9012,
      "step": 4836
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7592952251434326,
      "learning_rate": 0.00011229233841357738,
      "loss": 0.9308,
      "step": 4837
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7900505065917969,
      "learning_rate": 0.00011226250847367417,
      "loss": 0.9745,
      "step": 4838
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.729471743106842,
      "learning_rate": 0.00011223267742592272,
      "loss": 0.906,
      "step": 4839
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7428840398788452,
      "learning_rate": 0.00011220284527301812,
      "loss": 1.0044,
      "step": 4840
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7849833369255066,
      "learning_rate": 0.00011217301201765556,
      "loss": 1.0778,
      "step": 4841
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7510427832603455,
      "learning_rate": 0.00011214317766253026,
      "loss": 1.0539,
      "step": 4842
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7398820519447327,
      "learning_rate": 0.00011211334221033767,
      "loss": 0.9486,
      "step": 4843
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7769248485565186,
      "learning_rate": 0.00011208350566377315,
      "loss": 0.9762,
      "step": 4844
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8509570360183716,
      "learning_rate": 0.0001120536680255323,
      "loss": 1.0259,
      "step": 4845
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7902756929397583,
      "learning_rate": 0.00011202382929831083,
      "loss": 0.9836,
      "step": 4846
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7448802590370178,
      "learning_rate": 0.00011199398948480445,
      "loss": 0.9311,
      "step": 4847
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7944951057434082,
      "learning_rate": 0.00011196414858770903,
      "loss": 0.9934,
      "step": 4848
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7836403846740723,
      "learning_rate": 0.00011193430660972051,
      "loss": 0.9239,
      "step": 4849
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8021852970123291,
      "learning_rate": 0.00011190446355353497,
      "loss": 0.9101,
      "step": 4850
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8140667080879211,
      "learning_rate": 0.00011187461942184856,
      "loss": 1.0416,
      "step": 4851
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7513343691825867,
      "learning_rate": 0.00011184477421735748,
      "loss": 0.9801,
      "step": 4852
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7877258658409119,
      "learning_rate": 0.00011181492794275817,
      "loss": 1.0065,
      "step": 4853
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7535899877548218,
      "learning_rate": 0.000111785080600747,
      "loss": 0.8739,
      "step": 4854
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7377232909202576,
      "learning_rate": 0.00011175523219402048,
      "loss": 0.7785,
      "step": 4855
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7618876099586487,
      "learning_rate": 0.00011172538272527535,
      "loss": 0.9044,
      "step": 4856
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8464749455451965,
      "learning_rate": 0.00011169553219720828,
      "loss": 0.9384,
      "step": 4857
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8161110877990723,
      "learning_rate": 0.00011166568061251605,
      "loss": 1.0834,
      "step": 4858
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.772560715675354,
      "learning_rate": 0.00011163582797389567,
      "loss": 0.894,
      "step": 4859
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8467875123023987,
      "learning_rate": 0.00011160597428404407,
      "loss": 0.9931,
      "step": 4860
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7472082376480103,
      "learning_rate": 0.00011157611954565844,
      "loss": 0.8825,
      "step": 4861
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.790768563747406,
      "learning_rate": 0.00011154626376143593,
      "loss": 0.933,
      "step": 4862
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7249228358268738,
      "learning_rate": 0.00011151640693407387,
      "loss": 0.9692,
      "step": 4863
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7793048620223999,
      "learning_rate": 0.00011148654906626967,
      "loss": 0.9523,
      "step": 4864
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8146613240242004,
      "learning_rate": 0.00011145669016072078,
      "loss": 0.9364,
      "step": 4865
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.6911619305610657,
      "learning_rate": 0.00011142683022012478,
      "loss": 0.932,
      "step": 4866
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9315606355667114,
      "learning_rate": 0.00011139696924717941,
      "loss": 0.8802,
      "step": 4867
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.786227822303772,
      "learning_rate": 0.00011136710724458238,
      "loss": 1.0699,
      "step": 4868
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7897815704345703,
      "learning_rate": 0.00011133724421503157,
      "loss": 0.9314,
      "step": 4869
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8168982267379761,
      "learning_rate": 0.00011130738016122495,
      "loss": 0.8716,
      "step": 4870
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.6992068886756897,
      "learning_rate": 0.00011127751508586053,
      "loss": 0.8776,
      "step": 4871
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7909550666809082,
      "learning_rate": 0.00011124764899163649,
      "loss": 1.0275,
      "step": 4872
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.6491546630859375,
      "learning_rate": 0.00011121778188125107,
      "loss": 0.8812,
      "step": 4873
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8003860116004944,
      "learning_rate": 0.00011118791375740254,
      "loss": 1.0509,
      "step": 4874
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7199599146842957,
      "learning_rate": 0.00011115804462278933,
      "loss": 0.9646,
      "step": 4875
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.824131965637207,
      "learning_rate": 0.00011112817448010999,
      "loss": 0.8921,
      "step": 4876
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7687273025512695,
      "learning_rate": 0.00011109830333206311,
      "loss": 0.883,
      "step": 4877
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7793501019477844,
      "learning_rate": 0.00011106843118134736,
      "loss": 0.964,
      "step": 4878
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7844024896621704,
      "learning_rate": 0.0001110385580306615,
      "loss": 1.0223,
      "step": 4879
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7543127536773682,
      "learning_rate": 0.00011100868388270442,
      "loss": 1.0382,
      "step": 4880
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8293694853782654,
      "learning_rate": 0.0001109788087401751,
      "loss": 1.0587,
      "step": 4881
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.6826415061950684,
      "learning_rate": 0.00011094893260577256,
      "loss": 0.8231,
      "step": 4882
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.746153712272644,
      "learning_rate": 0.00011091905548219597,
      "loss": 0.9327,
      "step": 4883
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8294541239738464,
      "learning_rate": 0.00011088917737214452,
      "loss": 0.9046,
      "step": 4884
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7316751480102539,
      "learning_rate": 0.00011085929827831754,
      "loss": 0.9777,
      "step": 4885
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8139753937721252,
      "learning_rate": 0.00011082941820341448,
      "loss": 0.9985,
      "step": 4886
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7943364381790161,
      "learning_rate": 0.00011079953715013478,
      "loss": 1.0446,
      "step": 4887
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7968236207962036,
      "learning_rate": 0.00011076965512117805,
      "loss": 1.1082,
      "step": 4888
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8494092226028442,
      "learning_rate": 0.00011073977211924397,
      "loss": 0.8863,
      "step": 4889
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.90047287940979,
      "learning_rate": 0.00011070988814703225,
      "loss": 1.104,
      "step": 4890
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7975635528564453,
      "learning_rate": 0.00011068000320724281,
      "loss": 1.0129,
      "step": 4891
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7787724733352661,
      "learning_rate": 0.00011065011730257552,
      "loss": 0.9088,
      "step": 4892
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.6692328453063965,
      "learning_rate": 0.00011062023043573047,
      "loss": 0.9661,
      "step": 4893
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.82075434923172,
      "learning_rate": 0.00011059034260940773,
      "loss": 0.941,
      "step": 4894
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7842402458190918,
      "learning_rate": 0.00011056045382630744,
      "loss": 1.0,
      "step": 4895
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7709221839904785,
      "learning_rate": 0.00011053056408913001,
      "loss": 0.876,
      "step": 4896
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.6689760684967041,
      "learning_rate": 0.0001105006734005757,
      "loss": 0.8053,
      "step": 4897
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7427790760993958,
      "learning_rate": 0.00011047078176334502,
      "loss": 0.9647,
      "step": 4898
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8376590013504028,
      "learning_rate": 0.00011044088918013849,
      "loss": 0.9913,
      "step": 4899
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7888361215591431,
      "learning_rate": 0.00011041099565365674,
      "loss": 0.8347,
      "step": 4900
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8089134097099304,
      "learning_rate": 0.00011038110118660045,
      "loss": 1.1087,
      "step": 4901
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9311762452125549,
      "learning_rate": 0.00011035120578167049,
      "loss": 1.0656,
      "step": 4902
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7809112071990967,
      "learning_rate": 0.00011032130944156767,
      "loss": 1.0239,
      "step": 4903
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.6649876832962036,
      "learning_rate": 0.00011029141216899299,
      "loss": 0.8859,
      "step": 4904
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.776587724685669,
      "learning_rate": 0.00011026151396664747,
      "loss": 0.9574,
      "step": 4905
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9212097525596619,
      "learning_rate": 0.00011023161483723228,
      "loss": 0.9127,
      "step": 4906
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9740622043609619,
      "learning_rate": 0.00011020171478344863,
      "loss": 1.1391,
      "step": 4907
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8400608897209167,
      "learning_rate": 0.0001101718138079978,
      "loss": 0.9239,
      "step": 4908
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7605606317520142,
      "learning_rate": 0.00011014191191358119,
      "loss": 1.0455,
      "step": 4909
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7030631899833679,
      "learning_rate": 0.00011011200910290024,
      "loss": 1.0018,
      "step": 4910
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8395674228668213,
      "learning_rate": 0.00011008210537865653,
      "loss": 0.9589,
      "step": 4911
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8159804940223694,
      "learning_rate": 0.00011005220074355171,
      "loss": 0.9439,
      "step": 4912
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7161035537719727,
      "learning_rate": 0.00011002229520028746,
      "loss": 0.7207,
      "step": 4913
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7870673537254333,
      "learning_rate": 0.00010999238875156559,
      "loss": 1.1633,
      "step": 4914
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.6282675266265869,
      "learning_rate": 0.00010996248140008794,
      "loss": 0.8594,
      "step": 4915
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8430740833282471,
      "learning_rate": 0.00010993257314855653,
      "loss": 0.9314,
      "step": 4916
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.6887407302856445,
      "learning_rate": 0.00010990266399967337,
      "loss": 0.8555,
      "step": 4917
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7878185510635376,
      "learning_rate": 0.00010987275395614058,
      "loss": 1.0859,
      "step": 4918
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7717889547348022,
      "learning_rate": 0.00010984284302066037,
      "loss": 1.0218,
      "step": 4919
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.771571934223175,
      "learning_rate": 0.00010981293119593503,
      "loss": 1.0549,
      "step": 4920
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7822331786155701,
      "learning_rate": 0.00010978301848466692,
      "loss": 1.0124,
      "step": 4921
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9212383031845093,
      "learning_rate": 0.00010975310488955846,
      "loss": 1.082,
      "step": 4922
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7388486862182617,
      "learning_rate": 0.00010972319041331223,
      "loss": 0.9977,
      "step": 4923
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7861277461051941,
      "learning_rate": 0.00010969327505863076,
      "loss": 0.9302,
      "step": 4924
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8004422187805176,
      "learning_rate": 0.00010966335882821678,
      "loss": 0.9582,
      "step": 4925
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7476869821548462,
      "learning_rate": 0.00010963344172477308,
      "loss": 1.1044,
      "step": 4926
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.813045084476471,
      "learning_rate": 0.00010960352375100242,
      "loss": 1.0339,
      "step": 4927
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9535208344459534,
      "learning_rate": 0.0001095736049096078,
      "loss": 0.9714,
      "step": 4928
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8231042623519897,
      "learning_rate": 0.00010954368520329217,
      "loss": 0.9928,
      "step": 4929
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7152572870254517,
      "learning_rate": 0.00010951376463475858,
      "loss": 0.8452,
      "step": 4930
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.711341142654419,
      "learning_rate": 0.00010948384320671027,
      "loss": 0.9219,
      "step": 4931
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7866460680961609,
      "learning_rate": 0.00010945392092185043,
      "loss": 1.0716,
      "step": 4932
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7420953512191772,
      "learning_rate": 0.00010942399778288233,
      "loss": 0.8871,
      "step": 4933
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8441842794418335,
      "learning_rate": 0.0001093940737925094,
      "loss": 1.0337,
      "step": 4934
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7497798800468445,
      "learning_rate": 0.0001093641489534351,
      "loss": 0.9642,
      "step": 4935
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9009296894073486,
      "learning_rate": 0.00010933422326836294,
      "loss": 0.8992,
      "step": 4936
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8999432325363159,
      "learning_rate": 0.00010930429673999658,
      "loss": 1.135,
      "step": 4937
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9044147729873657,
      "learning_rate": 0.00010927436937103968,
      "loss": 0.9029,
      "step": 4938
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7883457541465759,
      "learning_rate": 0.00010924444116419604,
      "loss": 0.9975,
      "step": 4939
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7764599919319153,
      "learning_rate": 0.00010921451212216945,
      "loss": 0.8132,
      "step": 4940
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7747498750686646,
      "learning_rate": 0.00010918458224766386,
      "loss": 0.859,
      "step": 4941
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7943763136863708,
      "learning_rate": 0.0001091546515433833,
      "loss": 1.011,
      "step": 4942
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7181071043014526,
      "learning_rate": 0.00010912472001203177,
      "loss": 0.8989,
      "step": 4943
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7549866437911987,
      "learning_rate": 0.00010909478765631349,
      "loss": 1.0726,
      "step": 4944
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.798565149307251,
      "learning_rate": 0.0001090648544789326,
      "loss": 1.0303,
      "step": 4945
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7307342290878296,
      "learning_rate": 0.00010903492048259345,
      "loss": 0.9789,
      "step": 4946
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7824006080627441,
      "learning_rate": 0.00010900498567000039,
      "loss": 0.9985,
      "step": 4947
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8141332268714905,
      "learning_rate": 0.00010897505004385788,
      "loss": 0.9185,
      "step": 4948
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8538034558296204,
      "learning_rate": 0.00010894511360687039,
      "loss": 0.9765,
      "step": 4949
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7719243764877319,
      "learning_rate": 0.00010891517636174253,
      "loss": 1.0315,
      "step": 4950
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8093717098236084,
      "learning_rate": 0.00010888523831117899,
      "loss": 1.0684,
      "step": 4951
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8469390869140625,
      "learning_rate": 0.00010885529945788446,
      "loss": 1.1159,
      "step": 4952
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8024957776069641,
      "learning_rate": 0.00010882535980456376,
      "loss": 1.1199,
      "step": 4953
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.6433467268943787,
      "learning_rate": 0.00010879541935392176,
      "loss": 0.8979,
      "step": 4954
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7785211205482483,
      "learning_rate": 0.00010876547810866344,
      "loss": 1.0434,
      "step": 4955
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8724049925804138,
      "learning_rate": 0.0001087355360714938,
      "loss": 0.8796,
      "step": 4956
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7402239441871643,
      "learning_rate": 0.00010870559324511793,
      "loss": 0.9601,
      "step": 4957
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8949133157730103,
      "learning_rate": 0.00010867564963224101,
      "loss": 1.026,
      "step": 4958
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.6975107192993164,
      "learning_rate": 0.00010864570523556827,
      "loss": 0.9029,
      "step": 4959
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9479396939277649,
      "learning_rate": 0.000108615760057805,
      "loss": 1.0211,
      "step": 4960
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8513799905776978,
      "learning_rate": 0.0001085858141016566,
      "loss": 1.045,
      "step": 4961
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8274711966514587,
      "learning_rate": 0.0001085558673698285,
      "loss": 0.9805,
      "step": 4962
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.720975399017334,
      "learning_rate": 0.00010852591986502625,
      "loss": 0.9955,
      "step": 4963
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8408166766166687,
      "learning_rate": 0.00010849597158995542,
      "loss": 0.8946,
      "step": 4964
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7662861943244934,
      "learning_rate": 0.00010846602254732158,
      "loss": 1.1358,
      "step": 4965
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8497337102890015,
      "learning_rate": 0.00010843607273983063,
      "loss": 0.9136,
      "step": 4966
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0238080024719238,
      "learning_rate": 0.00010840612217018822,
      "loss": 0.963,
      "step": 4967
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.6822003722190857,
      "learning_rate": 0.0001083761708411003,
      "loss": 0.6186,
      "step": 4968
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7723561525344849,
      "learning_rate": 0.00010834621875527278,
      "loss": 1.0276,
      "step": 4969
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.792113721370697,
      "learning_rate": 0.00010831626591541157,
      "loss": 1.037,
      "step": 4970
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7487592101097107,
      "learning_rate": 0.00010828631232422291,
      "loss": 0.9075,
      "step": 4971
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8601886630058289,
      "learning_rate": 0.00010825635798441283,
      "loss": 0.8613,
      "step": 4972
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8530653119087219,
      "learning_rate": 0.00010822640289868753,
      "loss": 1.0797,
      "step": 4973
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7469228506088257,
      "learning_rate": 0.00010819644706975332,
      "loss": 0.8785,
      "step": 4974
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7503280639648438,
      "learning_rate": 0.0001081664905003165,
      "loss": 1.0749,
      "step": 4975
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7468873262405396,
      "learning_rate": 0.00010813653319308353,
      "loss": 1.1124,
      "step": 4976
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7453671097755432,
      "learning_rate": 0.00010810657515076086,
      "loss": 0.916,
      "step": 4977
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7897058725357056,
      "learning_rate": 0.000108076616376055,
      "loss": 0.947,
      "step": 4978
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.804443895816803,
      "learning_rate": 0.00010804665687167262,
      "loss": 1.016,
      "step": 4979
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.6845722794532776,
      "learning_rate": 0.0001080166966403203,
      "loss": 0.8888,
      "step": 4980
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7716487646102905,
      "learning_rate": 0.00010798673568470485,
      "loss": 1.0317,
      "step": 4981
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7068995833396912,
      "learning_rate": 0.00010795677400753306,
      "loss": 0.9932,
      "step": 4982
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8245829343795776,
      "learning_rate": 0.00010792681161151177,
      "loss": 0.9657,
      "step": 4983
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9497877359390259,
      "learning_rate": 0.00010789684849934796,
      "loss": 1.0754,
      "step": 4984
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7938148975372314,
      "learning_rate": 0.00010786688467374859,
      "loss": 1.02,
      "step": 4985
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7475448846817017,
      "learning_rate": 0.0001078369201374207,
      "loss": 1.0186,
      "step": 4986
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7335013747215271,
      "learning_rate": 0.0001078069548930715,
      "loss": 0.9782,
      "step": 4987
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8388778567314148,
      "learning_rate": 0.00010777698894340813,
      "loss": 1.0908,
      "step": 4988
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8602745532989502,
      "learning_rate": 0.0001077470222911378,
      "loss": 0.8835,
      "step": 4989
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7140038013458252,
      "learning_rate": 0.00010771705493896788,
      "loss": 0.8857,
      "step": 4990
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8438007831573486,
      "learning_rate": 0.00010768708688960575,
      "loss": 1.0315,
      "step": 4991
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7744883298873901,
      "learning_rate": 0.00010765711814575883,
      "loss": 1.0601,
      "step": 4992
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8040426969528198,
      "learning_rate": 0.00010762714871013467,
      "loss": 0.9017,
      "step": 4993
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8766114711761475,
      "learning_rate": 0.00010759717858544078,
      "loss": 0.8371,
      "step": 4994
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7012424468994141,
      "learning_rate": 0.00010756720777438482,
      "loss": 0.9728,
      "step": 4995
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7083522081375122,
      "learning_rate": 0.00010753723627967451,
      "loss": 0.966,
      "step": 4996
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8759257197380066,
      "learning_rate": 0.00010750726410401754,
      "loss": 1.1248,
      "step": 4997
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9319406747817993,
      "learning_rate": 0.0001074772912501218,
      "loss": 1.1054,
      "step": 4998
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8455122709274292,
      "learning_rate": 0.00010744731772069512,
      "loss": 1.0103,
      "step": 4999
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7683991193771362,
      "learning_rate": 0.00010741734351844544,
      "loss": 0.9358,
      "step": 5000
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7314528226852417,
      "learning_rate": 0.00010738736864608079,
      "loss": 0.8393,
      "step": 5001
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7942683696746826,
      "learning_rate": 0.00010735739310630916,
      "loss": 0.9442,
      "step": 5002
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0038470029830933,
      "learning_rate": 0.00010732741690183879,
      "loss": 1.0515,
      "step": 5003
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7769905924797058,
      "learning_rate": 0.00010729744003537776,
      "loss": 0.973,
      "step": 5004
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7850512862205505,
      "learning_rate": 0.00010726746250963431,
      "loss": 0.9856,
      "step": 5005
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8909277319908142,
      "learning_rate": 0.00010723748432731682,
      "loss": 0.9844,
      "step": 5006
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7409754991531372,
      "learning_rate": 0.00010720750549113359,
      "loss": 1.0171,
      "step": 5007
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7498346567153931,
      "learning_rate": 0.00010717752600379303,
      "loss": 1.1193,
      "step": 5008
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.7579041123390198,
      "learning_rate": 0.00010714754586800364,
      "loss": 0.9046,
      "step": 5009
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8222535252571106,
      "learning_rate": 0.00010711756508647395,
      "loss": 0.9735,
      "step": 5010
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.6744101643562317,
      "learning_rate": 0.00010708758366191254,
      "loss": 0.9995,
      "step": 5011
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7639139294624329,
      "learning_rate": 0.00010705760159702811,
      "loss": 0.9265,
      "step": 5012
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7082836627960205,
      "learning_rate": 0.0001070276188945293,
      "loss": 0.831,
      "step": 5013
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7807806730270386,
      "learning_rate": 0.00010699763555712495,
      "loss": 0.8165,
      "step": 5014
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8013209104537964,
      "learning_rate": 0.00010696765158752383,
      "loss": 0.9743,
      "step": 5015
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7248284816741943,
      "learning_rate": 0.00010693766698843482,
      "loss": 0.9452,
      "step": 5016
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7307727336883545,
      "learning_rate": 0.00010690768176256689,
      "loss": 0.9154,
      "step": 5017
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8232489824295044,
      "learning_rate": 0.00010687769591262904,
      "loss": 1.0497,
      "step": 5018
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.945981502532959,
      "learning_rate": 0.00010684770944133031,
      "loss": 0.92,
      "step": 5019
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.828311562538147,
      "learning_rate": 0.00010681772235137977,
      "loss": 0.9855,
      "step": 5020
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7558428645133972,
      "learning_rate": 0.00010678773464548663,
      "loss": 0.8314,
      "step": 5021
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9145514369010925,
      "learning_rate": 0.00010675774632636011,
      "loss": 1.0534,
      "step": 5022
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.711399257183075,
      "learning_rate": 0.00010672775739670949,
      "loss": 0.9327,
      "step": 5023
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.89848792552948,
      "learning_rate": 0.00010669776785924404,
      "loss": 1.046,
      "step": 5024
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8986443877220154,
      "learning_rate": 0.00010666777771667321,
      "loss": 1.1313,
      "step": 5025
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7316709756851196,
      "learning_rate": 0.0001066377869717064,
      "loss": 0.9042,
      "step": 5026
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8240397572517395,
      "learning_rate": 0.00010660779562705312,
      "loss": 1.1324,
      "step": 5027
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8348107933998108,
      "learning_rate": 0.00010657780368542293,
      "loss": 1.0284,
      "step": 5028
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7479124665260315,
      "learning_rate": 0.00010654781114952543,
      "loss": 1.0645,
      "step": 5029
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7343984246253967,
      "learning_rate": 0.00010651781802207023,
      "loss": 0.8719,
      "step": 5030
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2171791791915894,
      "learning_rate": 0.00010648782430576713,
      "loss": 0.9959,
      "step": 5031
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8316954374313354,
      "learning_rate": 0.00010645783000332578,
      "loss": 1.0267,
      "step": 5032
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8453816771507263,
      "learning_rate": 0.00010642783511745611,
      "loss": 1.0098,
      "step": 5033
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.722510814666748,
      "learning_rate": 0.0001063978396508679,
      "loss": 0.7125,
      "step": 5034
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8132488131523132,
      "learning_rate": 0.00010636784360627108,
      "loss": 1.1609,
      "step": 5035
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.7693787217140198,
      "learning_rate": 0.0001063378469863757,
      "loss": 1.0599,
      "step": 5036
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.6765629053115845,
      "learning_rate": 0.00010630784979389168,
      "loss": 0.9047,
      "step": 5037
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8674545288085938,
      "learning_rate": 0.00010627785203152918,
      "loss": 1.0258,
      "step": 5038
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.982782244682312,
      "learning_rate": 0.0001062478537019983,
      "loss": 0.962,
      "step": 5039
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7085098624229431,
      "learning_rate": 0.00010621785480800914,
      "loss": 0.9935,
      "step": 5040
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9388718605041504,
      "learning_rate": 0.00010618785535227209,
      "loss": 1.1226,
      "step": 5041
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8830339312553406,
      "learning_rate": 0.00010615785533749729,
      "loss": 1.2911,
      "step": 5042
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7700386643409729,
      "learning_rate": 0.00010612785476639518,
      "loss": 0.8633,
      "step": 5043
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7826847434043884,
      "learning_rate": 0.00010609785364167606,
      "loss": 0.9842,
      "step": 5044
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9161370396614075,
      "learning_rate": 0.00010606785196605033,
      "loss": 1.0458,
      "step": 5045
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8588224053382874,
      "learning_rate": 0.00010603784974222861,
      "loss": 0.9512,
      "step": 5046
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7238483428955078,
      "learning_rate": 0.00010600784697292134,
      "loss": 0.838,
      "step": 5047
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7885895371437073,
      "learning_rate": 0.0001059778436608391,
      "loss": 1.0034,
      "step": 5048
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7344606518745422,
      "learning_rate": 0.00010594783980869254,
      "loss": 0.9331,
      "step": 5049
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7655723690986633,
      "learning_rate": 0.00010591783541919232,
      "loss": 0.8249,
      "step": 5050
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7848523259162903,
      "learning_rate": 0.00010588783049504915,
      "loss": 0.9683,
      "step": 5051
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7566481828689575,
      "learning_rate": 0.00010585782503897388,
      "loss": 0.9803,
      "step": 5052
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9054071307182312,
      "learning_rate": 0.00010582781905367725,
      "loss": 0.9562,
      "step": 5053
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7615681290626526,
      "learning_rate": 0.0001057978125418702,
      "loss": 0.9088,
      "step": 5054
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7379032373428345,
      "learning_rate": 0.00010576780550626357,
      "loss": 0.9506,
      "step": 5055
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8043782711029053,
      "learning_rate": 0.0001057377979495684,
      "loss": 0.9858,
      "step": 5056
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8147229552268982,
      "learning_rate": 0.00010570778987449568,
      "loss": 1.0159,
      "step": 5057
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8456796407699585,
      "learning_rate": 0.00010567778128375642,
      "loss": 1.0781,
      "step": 5058
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7417508363723755,
      "learning_rate": 0.00010564777218006181,
      "loss": 1.003,
      "step": 5059
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7917277812957764,
      "learning_rate": 0.00010561776256612295,
      "loss": 0.9502,
      "step": 5060
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8203314542770386,
      "learning_rate": 0.00010558775244465102,
      "loss": 0.8545,
      "step": 5061
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.869928777217865,
      "learning_rate": 0.00010555774181835733,
      "loss": 1.0155,
      "step": 5062
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8022637367248535,
      "learning_rate": 0.00010552773068995314,
      "loss": 1.0879,
      "step": 5063
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.672001838684082,
      "learning_rate": 0.00010549771906214979,
      "loss": 0.9135,
      "step": 5064
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7225290536880493,
      "learning_rate": 0.0001054677069376586,
      "loss": 0.7734,
      "step": 5065
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8035221099853516,
      "learning_rate": 0.00010543769431919108,
      "loss": 0.9138,
      "step": 5066
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8685542941093445,
      "learning_rate": 0.00010540768120945865,
      "loss": 1.0208,
      "step": 5067
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7406560778617859,
      "learning_rate": 0.00010537766761117286,
      "loss": 0.9031,
      "step": 5068
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7587171196937561,
      "learning_rate": 0.00010534765352704523,
      "loss": 0.8852,
      "step": 5069
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.6719124913215637,
      "learning_rate": 0.00010531763895978739,
      "loss": 0.8829,
      "step": 5070
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8159770965576172,
      "learning_rate": 0.000105287623912111,
      "loss": 0.9413,
      "step": 5071
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8835862874984741,
      "learning_rate": 0.00010525760838672769,
      "loss": 1.1545,
      "step": 5072
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8177557587623596,
      "learning_rate": 0.00010522759238634927,
      "loss": 0.8752,
      "step": 5073
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7685697674751282,
      "learning_rate": 0.00010519757591368744,
      "loss": 0.9306,
      "step": 5074
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8134411573410034,
      "learning_rate": 0.00010516755897145408,
      "loss": 0.7861,
      "step": 5075
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8483935594558716,
      "learning_rate": 0.00010513754156236102,
      "loss": 0.9184,
      "step": 5076
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9010506868362427,
      "learning_rate": 0.00010510752368912014,
      "loss": 0.9618,
      "step": 5077
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7965332269668579,
      "learning_rate": 0.00010507750535444346,
      "loss": 1.0317,
      "step": 5078
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.6694085597991943,
      "learning_rate": 0.0001050474865610429,
      "loss": 0.8317,
      "step": 5079
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7621757388114929,
      "learning_rate": 0.00010501746731163049,
      "loss": 0.9535,
      "step": 5080
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7820752859115601,
      "learning_rate": 0.00010498744760891834,
      "loss": 0.9037,
      "step": 5081
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7151625156402588,
      "learning_rate": 0.00010495742745561854,
      "loss": 0.8744,
      "step": 5082
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.742006778717041,
      "learning_rate": 0.00010492740685444322,
      "loss": 1.0673,
      "step": 5083
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8351051807403564,
      "learning_rate": 0.00010489738580810462,
      "loss": 1.0883,
      "step": 5084
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.6983124613761902,
      "learning_rate": 0.0001048673643193149,
      "loss": 0.9804,
      "step": 5085
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8488489985466003,
      "learning_rate": 0.0001048373423907864,
      "loss": 1.0387,
      "step": 5086
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8636870980262756,
      "learning_rate": 0.00010480732002523139,
      "loss": 1.0967,
      "step": 5087
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7339608073234558,
      "learning_rate": 0.00010477729722536225,
      "loss": 0.9628,
      "step": 5088
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7165808081626892,
      "learning_rate": 0.00010474727399389134,
      "loss": 0.8708,
      "step": 5089
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.6733713746070862,
      "learning_rate": 0.0001047172503335311,
      "loss": 0.9271,
      "step": 5090
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.6857487559318542,
      "learning_rate": 0.00010468722624699401,
      "loss": 0.9018,
      "step": 5091
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7812450528144836,
      "learning_rate": 0.00010465720173699259,
      "loss": 0.9974,
      "step": 5092
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8059720396995544,
      "learning_rate": 0.00010462717680623933,
      "loss": 0.8986,
      "step": 5093
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7781034708023071,
      "learning_rate": 0.00010459715145744686,
      "loss": 1.031,
      "step": 5094
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7936933040618896,
      "learning_rate": 0.00010456712569332777,
      "loss": 0.9529,
      "step": 5095
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7723500728607178,
      "learning_rate": 0.00010453709951659475,
      "loss": 1.0107,
      "step": 5096
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.6995753049850464,
      "learning_rate": 0.00010450707292996047,
      "loss": 0.9326,
      "step": 5097
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8494675159454346,
      "learning_rate": 0.00010447704593613768,
      "loss": 1.0187,
      "step": 5098
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.855739176273346,
      "learning_rate": 0.00010444701853783911,
      "loss": 1.005,
      "step": 5099
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7245144248008728,
      "learning_rate": 0.00010441699073777762,
      "loss": 0.8173,
      "step": 5100
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8450302481651306,
      "learning_rate": 0.00010438696253866603,
      "loss": 0.8375,
      "step": 5101
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.767345666885376,
      "learning_rate": 0.00010435693394321717,
      "loss": 0.9019,
      "step": 5102
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8089317083358765,
      "learning_rate": 0.00010432690495414405,
      "loss": 0.9923,
      "step": 5103
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7338062524795532,
      "learning_rate": 0.00010429687557415954,
      "loss": 0.8316,
      "step": 5104
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.6969124674797058,
      "learning_rate": 0.00010426684580597666,
      "loss": 0.8215,
      "step": 5105
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8065308332443237,
      "learning_rate": 0.0001042368156523084,
      "loss": 1.0661,
      "step": 5106
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7004128098487854,
      "learning_rate": 0.00010420678511586785,
      "loss": 0.7736,
      "step": 5107
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7430830597877502,
      "learning_rate": 0.00010417675419936809,
      "loss": 0.9779,
      "step": 5108
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9451195597648621,
      "learning_rate": 0.00010414672290552223,
      "loss": 0.8763,
      "step": 5109
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8106129169464111,
      "learning_rate": 0.00010411669123704344,
      "loss": 1.0782,
      "step": 5110
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.6853575110435486,
      "learning_rate": 0.00010408665919664491,
      "loss": 0.8011,
      "step": 5111
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8419955968856812,
      "learning_rate": 0.00010405662678703984,
      "loss": 1.0292,
      "step": 5112
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7453970313072205,
      "learning_rate": 0.00010402659401094152,
      "loss": 1.0889,
      "step": 5113
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.819788932800293,
      "learning_rate": 0.00010399656087106324,
      "loss": 1.0317,
      "step": 5114
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7040184140205383,
      "learning_rate": 0.00010396652737011828,
      "loss": 1.0628,
      "step": 5115
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.740817666053772,
      "learning_rate": 0.00010393649351082006,
      "loss": 0.9928,
      "step": 5116
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8002640604972839,
      "learning_rate": 0.00010390645929588196,
      "loss": 1.0403,
      "step": 5117
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7260842323303223,
      "learning_rate": 0.00010387642472801733,
      "loss": 0.9452,
      "step": 5118
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7765595316886902,
      "learning_rate": 0.00010384638980993972,
      "loss": 1.0988,
      "step": 5119
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8337041139602661,
      "learning_rate": 0.00010381635454436252,
      "loss": 0.9593,
      "step": 5120
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7491742968559265,
      "learning_rate": 0.00010378631893399933,
      "loss": 0.9047,
      "step": 5121
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7119678258895874,
      "learning_rate": 0.00010375628298156365,
      "loss": 0.9697,
      "step": 5122
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8019363880157471,
      "learning_rate": 0.00010372624668976909,
      "loss": 0.8877,
      "step": 5123
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.6626631021499634,
      "learning_rate": 0.00010369621006132923,
      "loss": 0.651,
      "step": 5124
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8198673129081726,
      "learning_rate": 0.0001036661730989577,
      "loss": 0.9331,
      "step": 5125
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7067912817001343,
      "learning_rate": 0.0001036361358053682,
      "loss": 1.0908,
      "step": 5126
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7765764594078064,
      "learning_rate": 0.00010360609818327443,
      "loss": 0.8615,
      "step": 5127
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8399764895439148,
      "learning_rate": 0.00010357606023539008,
      "loss": 1.1321,
      "step": 5128
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8030591011047363,
      "learning_rate": 0.00010354602196442897,
      "loss": 0.7552,
      "step": 5129
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7561306953430176,
      "learning_rate": 0.00010351598337310482,
      "loss": 0.9402,
      "step": 5130
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7164328098297119,
      "learning_rate": 0.00010348594446413148,
      "loss": 0.9956,
      "step": 5131
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7632815837860107,
      "learning_rate": 0.00010345590524022283,
      "loss": 0.9489,
      "step": 5132
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7721334099769592,
      "learning_rate": 0.00010342586570409267,
      "loss": 0.7268,
      "step": 5133
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8274480700492859,
      "learning_rate": 0.00010339582585845498,
      "loss": 0.9895,
      "step": 5134
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7295353412628174,
      "learning_rate": 0.00010336578570602362,
      "loss": 0.9588,
      "step": 5135
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.811366617679596,
      "learning_rate": 0.00010333574524951257,
      "loss": 0.9832,
      "step": 5136
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7662849426269531,
      "learning_rate": 0.00010330570449163586,
      "loss": 1.0832,
      "step": 5137
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7018322944641113,
      "learning_rate": 0.00010327566343510746,
      "loss": 0.9579,
      "step": 5138
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8456294536590576,
      "learning_rate": 0.00010324562208264139,
      "loss": 0.9006,
      "step": 5139
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7415403127670288,
      "learning_rate": 0.00010321558043695174,
      "loss": 0.8829,
      "step": 5140
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8214705586433411,
      "learning_rate": 0.00010318553850075265,
      "loss": 0.9886,
      "step": 5141
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7233492732048035,
      "learning_rate": 0.00010315549627675816,
      "loss": 0.919,
      "step": 5142
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9265508055686951,
      "learning_rate": 0.00010312545376768246,
      "loss": 0.8932,
      "step": 5143
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8299009799957275,
      "learning_rate": 0.00010309541097623967,
      "loss": 0.8363,
      "step": 5144
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.6598271727561951,
      "learning_rate": 0.00010306536790514406,
      "loss": 0.6518,
      "step": 5145
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8086226582527161,
      "learning_rate": 0.0001030353245571098,
      "loss": 0.9596,
      "step": 5146
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7078489661216736,
      "learning_rate": 0.00010300528093485115,
      "loss": 0.8122,
      "step": 5147
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7355614900588989,
      "learning_rate": 0.00010297523704108242,
      "loss": 1.0471,
      "step": 5148
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7539297938346863,
      "learning_rate": 0.00010294519287851783,
      "loss": 0.9966,
      "step": 5149
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7892729043960571,
      "learning_rate": 0.00010291514844987174,
      "loss": 0.966,
      "step": 5150
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8298482298851013,
      "learning_rate": 0.00010288510375785851,
      "loss": 0.9127,
      "step": 5151
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7779788374900818,
      "learning_rate": 0.00010285505880519249,
      "loss": 1.0337,
      "step": 5152
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8932836651802063,
      "learning_rate": 0.00010282501359458807,
      "loss": 1.1316,
      "step": 5153
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7916525602340698,
      "learning_rate": 0.00010279496812875969,
      "loss": 0.9774,
      "step": 5154
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7860032916069031,
      "learning_rate": 0.0001027649224104217,
      "loss": 0.9521,
      "step": 5155
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7322259545326233,
      "learning_rate": 0.00010273487644228868,
      "loss": 0.9672,
      "step": 5156
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.733378529548645,
      "learning_rate": 0.00010270483022707506,
      "loss": 0.9651,
      "step": 5157
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7881492972373962,
      "learning_rate": 0.00010267478376749531,
      "loss": 1.0659,
      "step": 5158
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8630042672157288,
      "learning_rate": 0.00010264473706626404,
      "loss": 0.9151,
      "step": 5159
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7404335737228394,
      "learning_rate": 0.0001026146901260957,
      "loss": 0.8849,
      "step": 5160
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7069669365882874,
      "learning_rate": 0.00010258464294970492,
      "loss": 0.8823,
      "step": 5161
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.755982518196106,
      "learning_rate": 0.0001025545955398063,
      "loss": 1.0988,
      "step": 5162
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7744724750518799,
      "learning_rate": 0.00010252454789911443,
      "loss": 0.9114,
      "step": 5163
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.6935697793960571,
      "learning_rate": 0.00010249450003034395,
      "loss": 1.1287,
      "step": 5164
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7566258907318115,
      "learning_rate": 0.00010246445193620953,
      "loss": 1.0492,
      "step": 5165
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7999916672706604,
      "learning_rate": 0.00010243440361942581,
      "loss": 1.0886,
      "step": 5166
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7703303098678589,
      "learning_rate": 0.00010240435508270752,
      "loss": 0.9508,
      "step": 5167
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8494589328765869,
      "learning_rate": 0.00010237430632876934,
      "loss": 0.9582,
      "step": 5168
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.777059018611908,
      "learning_rate": 0.00010234425736032607,
      "loss": 1.1731,
      "step": 5169
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.6870027780532837,
      "learning_rate": 0.00010231420818009238,
      "loss": 0.956,
      "step": 5170
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8410665988922119,
      "learning_rate": 0.00010228415879078308,
      "loss": 1.1189,
      "step": 5171
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8162976503372192,
      "learning_rate": 0.00010225410919511301,
      "loss": 1.0337,
      "step": 5172
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7400126457214355,
      "learning_rate": 0.00010222405939579692,
      "loss": 1.0547,
      "step": 5173
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7200222015380859,
      "learning_rate": 0.00010219400939554964,
      "loss": 0.9803,
      "step": 5174
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.6938434839248657,
      "learning_rate": 0.00010216395919708603,
      "loss": 0.8514,
      "step": 5175
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7400026321411133,
      "learning_rate": 0.000102133908803121,
      "loss": 1.0465,
      "step": 5176
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8585032224655151,
      "learning_rate": 0.00010210385821636934,
      "loss": 0.8917,
      "step": 5177
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7792884111404419,
      "learning_rate": 0.00010207380743954602,
      "loss": 1.0266,
      "step": 5178
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8200801610946655,
      "learning_rate": 0.00010204375647536595,
      "loss": 0.9524,
      "step": 5179
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8211904764175415,
      "learning_rate": 0.00010201370532654404,
      "loss": 1.0111,
      "step": 5180
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8270883560180664,
      "learning_rate": 0.00010198365399579528,
      "loss": 1.0467,
      "step": 5181
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.825343132019043,
      "learning_rate": 0.00010195360248583458,
      "loss": 0.9845,
      "step": 5182
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9144495725631714,
      "learning_rate": 0.000101923550799377,
      "loss": 0.9904,
      "step": 5183
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.6933146119117737,
      "learning_rate": 0.00010189349893913745,
      "loss": 0.9956,
      "step": 5184
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7854787707328796,
      "learning_rate": 0.00010186344690783101,
      "loss": 0.9385,
      "step": 5185
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8274226188659668,
      "learning_rate": 0.00010183339470817271,
      "loss": 0.9576,
      "step": 5186
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7524675726890564,
      "learning_rate": 0.00010180334234287754,
      "loss": 0.9829,
      "step": 5187
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7909670472145081,
      "learning_rate": 0.00010177328981466064,
      "loss": 1.0907,
      "step": 5188
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7706515192985535,
      "learning_rate": 0.00010174323712623704,
      "loss": 0.8728,
      "step": 5189
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.837437629699707,
      "learning_rate": 0.00010171318428032179,
      "loss": 1.1172,
      "step": 5190
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.70237797498703,
      "learning_rate": 0.00010168313127963008,
      "loss": 0.8799,
      "step": 5191
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7743717432022095,
      "learning_rate": 0.000101653078126877,
      "loss": 0.9472,
      "step": 5192
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8090443015098572,
      "learning_rate": 0.00010162302482477764,
      "loss": 1.1221,
      "step": 5193
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8558246493339539,
      "learning_rate": 0.0001015929713760472,
      "loss": 0.9798,
      "step": 5194
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.6947144269943237,
      "learning_rate": 0.00010156291778340079,
      "loss": 0.7357,
      "step": 5195
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9039539694786072,
      "learning_rate": 0.00010153286404955362,
      "loss": 1.0181,
      "step": 5196
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7144058346748352,
      "learning_rate": 0.00010150281017722088,
      "loss": 0.7697,
      "step": 5197
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8410497307777405,
      "learning_rate": 0.00010147275616911774,
      "loss": 0.9568,
      "step": 5198
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8819612860679626,
      "learning_rate": 0.00010144270202795943,
      "loss": 1.121,
      "step": 5199
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9397442936897278,
      "learning_rate": 0.00010141264775646114,
      "loss": 0.9298,
      "step": 5200
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8475532531738281,
      "learning_rate": 0.0001013825933573381,
      "loss": 0.9965,
      "step": 5201
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9101167917251587,
      "learning_rate": 0.00010135253883330563,
      "loss": 0.8776,
      "step": 5202
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7638474702835083,
      "learning_rate": 0.00010132248418707888,
      "loss": 0.9963,
      "step": 5203
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9219881892204285,
      "learning_rate": 0.00010129242942137322,
      "loss": 0.9861,
      "step": 5204
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7822911739349365,
      "learning_rate": 0.00010126237453890386,
      "loss": 1.101,
      "step": 5205
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8252258896827698,
      "learning_rate": 0.0001012323195423861,
      "loss": 0.9579,
      "step": 5206
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9163365960121155,
      "learning_rate": 0.00010120226443453526,
      "loss": 1.0257,
      "step": 5207
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7574987411499023,
      "learning_rate": 0.00010117220921806664,
      "loss": 0.87,
      "step": 5208
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8726235628128052,
      "learning_rate": 0.00010114215389569555,
      "loss": 1.0485,
      "step": 5209
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8032419681549072,
      "learning_rate": 0.00010111209847013732,
      "loss": 1.1435,
      "step": 5210
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8374848961830139,
      "learning_rate": 0.00010108204294410726,
      "loss": 1.0055,
      "step": 5211
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.6941425800323486,
      "learning_rate": 0.0001010519873203208,
      "loss": 0.8382,
      "step": 5212
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.87954181432724,
      "learning_rate": 0.00010102193160149322,
      "loss": 0.9534,
      "step": 5213
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7242146730422974,
      "learning_rate": 0.0001009918757903399,
      "loss": 0.9834,
      "step": 5214
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8842020630836487,
      "learning_rate": 0.0001009618198895762,
      "loss": 1.0604,
      "step": 5215
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.72428959608078,
      "learning_rate": 0.00010093176390191757,
      "loss": 0.8531,
      "step": 5216
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8273959755897522,
      "learning_rate": 0.0001009017078300793,
      "loss": 1.0667,
      "step": 5217
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7096107602119446,
      "learning_rate": 0.00010087165167677686,
      "loss": 0.9236,
      "step": 5218
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8173868656158447,
      "learning_rate": 0.00010084159544472562,
      "loss": 0.9115,
      "step": 5219
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8997098207473755,
      "learning_rate": 0.00010081153913664099,
      "loss": 0.9685,
      "step": 5220
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7394409775733948,
      "learning_rate": 0.0001007814827552384,
      "loss": 0.8171,
      "step": 5221
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8413199782371521,
      "learning_rate": 0.00010075142630323324,
      "loss": 0.8588,
      "step": 5222
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7438793182373047,
      "learning_rate": 0.00010072136978334102,
      "loss": 0.8727,
      "step": 5223
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8331936001777649,
      "learning_rate": 0.00010069131319827709,
      "loss": 1.1788,
      "step": 5224
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7141227722167969,
      "learning_rate": 0.0001006612565507569,
      "loss": 0.6969,
      "step": 5225
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7840338349342346,
      "learning_rate": 0.00010063119984349598,
      "loss": 0.9761,
      "step": 5226
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9646100997924805,
      "learning_rate": 0.00010060114307920972,
      "loss": 1.2251,
      "step": 5227
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8005926012992859,
      "learning_rate": 0.00010057108626061358,
      "loss": 0.9434,
      "step": 5228
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8847595453262329,
      "learning_rate": 0.00010054102939042302,
      "loss": 1.1027,
      "step": 5229
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.792830228805542,
      "learning_rate": 0.00010051097247135351,
      "loss": 0.9165,
      "step": 5230
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.663872241973877,
      "learning_rate": 0.00010048091550612055,
      "loss": 0.8339,
      "step": 5231
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7960761785507202,
      "learning_rate": 0.0001004508584974396,
      "loss": 0.9844,
      "step": 5232
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7719110250473022,
      "learning_rate": 0.0001004208014480261,
      "loss": 1.0782,
      "step": 5233
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.839648425579071,
      "learning_rate": 0.00010039074436059561,
      "loss": 1.0546,
      "step": 5234
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8012295365333557,
      "learning_rate": 0.00010036068723786356,
      "loss": 0.9247,
      "step": 5235
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7242897152900696,
      "learning_rate": 0.00010033063008254546,
      "loss": 1.0221,
      "step": 5236
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7097167372703552,
      "learning_rate": 0.00010030057289735683,
      "loss": 0.8956,
      "step": 5237
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8258913159370422,
      "learning_rate": 0.00010027051568501311,
      "loss": 0.988,
      "step": 5238
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8673880696296692,
      "learning_rate": 0.00010024045844822987,
      "loss": 0.9511,
      "step": 5239
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7785250544548035,
      "learning_rate": 0.00010021040118972252,
      "loss": 0.8667,
      "step": 5240
    },
    {
      "epoch": 1.98,
      "eval_loss": 0.9470303058624268,
      "eval_runtime": 17.7619,
      "eval_samples_per_second": 153.644,
      "eval_steps_per_second": 38.453,
      "step": 5240
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8872907757759094,
      "learning_rate": 0.00010018034391220663,
      "loss": 1.199,
      "step": 5241
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.6809889078140259,
      "learning_rate": 0.0001001502866183977,
      "loss": 1.0009,
      "step": 5242
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7920276522636414,
      "learning_rate": 0.00010012022931101122,
      "loss": 1.1129,
      "step": 5243
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.764680027961731,
      "learning_rate": 0.00010009017199276271,
      "loss": 0.9712,
      "step": 5244
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7828908562660217,
      "learning_rate": 0.00010006011466636764,
      "loss": 0.9613,
      "step": 5245
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8445849418640137,
      "learning_rate": 0.00010003005733454158,
      "loss": 1.1052,
      "step": 5246
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7842500805854797,
      "learning_rate": 0.0001,
      "loss": 0.9977,
      "step": 5247
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9056081771850586,
      "learning_rate": 9.996994266545843e-05,
      "loss": 0.8206,
      "step": 5248
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8096466064453125,
      "learning_rate": 9.993988533363237e-05,
      "loss": 0.99,
      "step": 5249
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8498285412788391,
      "learning_rate": 9.990982800723731e-05,
      "loss": 0.945,
      "step": 5250
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8508979082107544,
      "learning_rate": 9.98797706889888e-05,
      "loss": 1.0747,
      "step": 5251
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7915741801261902,
      "learning_rate": 9.984971338160231e-05,
      "loss": 0.9748,
      "step": 5252
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7941058874130249,
      "learning_rate": 9.981965608779337e-05,
      "loss": 0.9875,
      "step": 5253
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8532423973083496,
      "learning_rate": 9.97895988102775e-05,
      "loss": 1.0914,
      "step": 5254
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.7412028312683105,
      "learning_rate": 9.975954155177016e-05,
      "loss": 0.909,
      "step": 5255
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.6522200703620911,
      "learning_rate": 9.972948431498691e-05,
      "loss": 0.7474,
      "step": 5256
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.869448184967041,
      "learning_rate": 9.969942710264319e-05,
      "loss": 0.9348,
      "step": 5257
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.636637806892395,
      "learning_rate": 9.966936991745453e-05,
      "loss": 0.6835,
      "step": 5258
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8568200469017029,
      "learning_rate": 9.963931276213645e-05,
      "loss": 0.9184,
      "step": 5259
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.885409951210022,
      "learning_rate": 9.96092556394044e-05,
      "loss": 1.0004,
      "step": 5260
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8801878690719604,
      "learning_rate": 9.957919855197392e-05,
      "loss": 0.9671,
      "step": 5261
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.790540337562561,
      "learning_rate": 9.954914150256043e-05,
      "loss": 0.8733,
      "step": 5262
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8582541942596436,
      "learning_rate": 9.951908449387946e-05,
      "loss": 1.0068,
      "step": 5263
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8530950546264648,
      "learning_rate": 9.948902752864651e-05,
      "loss": 0.9571,
      "step": 5264
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.770168662071228,
      "learning_rate": 9.9458970609577e-05,
      "loss": 0.985,
      "step": 5265
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8078840374946594,
      "learning_rate": 9.942891373938647e-05,
      "loss": 0.9477,
      "step": 5266
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8191209435462952,
      "learning_rate": 9.939885692079032e-05,
      "loss": 1.0934,
      "step": 5267
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8750408291816711,
      "learning_rate": 9.9368800156504e-05,
      "loss": 1.1182,
      "step": 5268
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0841648578643799,
      "learning_rate": 9.93387434492431e-05,
      "loss": 1.0621,
      "step": 5269
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.836670458316803,
      "learning_rate": 9.930868680172293e-05,
      "loss": 0.9893,
      "step": 5270
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.787344217300415,
      "learning_rate": 9.927863021665903e-05,
      "loss": 0.8796,
      "step": 5271
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8018674254417419,
      "learning_rate": 9.924857369676676e-05,
      "loss": 1.1202,
      "step": 5272
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7358179688453674,
      "learning_rate": 9.92185172447616e-05,
      "loss": 0.8717,
      "step": 5273
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7181487679481506,
      "learning_rate": 9.918846086335904e-05,
      "loss": 0.8874,
      "step": 5274
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7962504029273987,
      "learning_rate": 9.915840455527441e-05,
      "loss": 0.9029,
      "step": 5275
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8399698138237,
      "learning_rate": 9.912834832322317e-05,
      "loss": 1.0941,
      "step": 5276
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7942226529121399,
      "learning_rate": 9.909829216992071e-05,
      "loss": 1.0475,
      "step": 5277
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.844666600227356,
      "learning_rate": 9.906823609808246e-05,
      "loss": 0.9996,
      "step": 5278
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7771795392036438,
      "learning_rate": 9.90381801104238e-05,
      "loss": 1.1003,
      "step": 5279
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.833823025226593,
      "learning_rate": 9.900812420966012e-05,
      "loss": 0.9391,
      "step": 5280
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7674879431724548,
      "learning_rate": 9.897806839850682e-05,
      "loss": 0.9218,
      "step": 5281
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7346460223197937,
      "learning_rate": 9.894801267967922e-05,
      "loss": 0.9084,
      "step": 5282
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8743252158164978,
      "learning_rate": 9.891795705589274e-05,
      "loss": 1.0336,
      "step": 5283
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7466950416564941,
      "learning_rate": 9.888790152986272e-05,
      "loss": 1.1382,
      "step": 5284
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7701571583747864,
      "learning_rate": 9.885784610430448e-05,
      "loss": 1.006,
      "step": 5285
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7315901517868042,
      "learning_rate": 9.882779078193339e-05,
      "loss": 0.8858,
      "step": 5286
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7886285781860352,
      "learning_rate": 9.879773556546475e-05,
      "loss": 0.9147,
      "step": 5287
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7413890361785889,
      "learning_rate": 9.876768045761389e-05,
      "loss": 0.8985,
      "step": 5288
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9307523369789124,
      "learning_rate": 9.873762546109616e-05,
      "loss": 0.9567,
      "step": 5289
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7683214545249939,
      "learning_rate": 9.87075705786268e-05,
      "loss": 0.9975,
      "step": 5290
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7447076439857483,
      "learning_rate": 9.867751581292113e-05,
      "loss": 0.8063,
      "step": 5291
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7607795000076294,
      "learning_rate": 9.86474611666944e-05,
      "loss": 0.8658,
      "step": 5292
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8467500805854797,
      "learning_rate": 9.861740664266191e-05,
      "loss": 1.0452,
      "step": 5293
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7768862247467041,
      "learning_rate": 9.85873522435389e-05,
      "loss": 0.9636,
      "step": 5294
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7459428310394287,
      "learning_rate": 9.855729797204061e-05,
      "loss": 1.0354,
      "step": 5295
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7940539121627808,
      "learning_rate": 9.852724383088231e-05,
      "loss": 1.0558,
      "step": 5296
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7851496338844299,
      "learning_rate": 9.849718982277913e-05,
      "loss": 0.869,
      "step": 5297
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8199241757392883,
      "learning_rate": 9.846713595044638e-05,
      "loss": 1.0221,
      "step": 5298
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8193984031677246,
      "learning_rate": 9.843708221659924e-05,
      "loss": 0.972,
      "step": 5299
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8176395297050476,
      "learning_rate": 9.840702862395281e-05,
      "loss": 0.8572,
      "step": 5300
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8419879674911499,
      "learning_rate": 9.83769751752224e-05,
      "loss": 0.9904,
      "step": 5301
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.778895914554596,
      "learning_rate": 9.834692187312301e-05,
      "loss": 0.9784,
      "step": 5302
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7293362617492676,
      "learning_rate": 9.831686872036992e-05,
      "loss": 1.0606,
      "step": 5303
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7587986588478088,
      "learning_rate": 9.828681571967823e-05,
      "loss": 0.9385,
      "step": 5304
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7802038788795471,
      "learning_rate": 9.825676287376298e-05,
      "loss": 1.011,
      "step": 5305
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.77179354429245,
      "learning_rate": 9.822671018533941e-05,
      "loss": 1.1325,
      "step": 5306
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7506158351898193,
      "learning_rate": 9.819665765712248e-05,
      "loss": 1.0097,
      "step": 5307
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7466879487037659,
      "learning_rate": 9.81666052918273e-05,
      "loss": 0.918,
      "step": 5308
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7179660797119141,
      "learning_rate": 9.813655309216903e-05,
      "loss": 0.8989,
      "step": 5309
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8438458442687988,
      "learning_rate": 9.810650106086257e-05,
      "loss": 0.9791,
      "step": 5310
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7376074194908142,
      "learning_rate": 9.807644920062306e-05,
      "loss": 1.0319,
      "step": 5311
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8553178310394287,
      "learning_rate": 9.804639751416544e-05,
      "loss": 1.054,
      "step": 5312
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7286335825920105,
      "learning_rate": 9.801634600420476e-05,
      "loss": 0.7195,
      "step": 5313
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7107179760932922,
      "learning_rate": 9.798629467345599e-05,
      "loss": 0.9612,
      "step": 5314
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.720973789691925,
      "learning_rate": 9.795624352463407e-05,
      "loss": 0.9992,
      "step": 5315
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7459086775779724,
      "learning_rate": 9.792619256045402e-05,
      "loss": 0.9668,
      "step": 5316
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8582184314727783,
      "learning_rate": 9.789614178363069e-05,
      "loss": 1.0098,
      "step": 5317
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.746880054473877,
      "learning_rate": 9.786609119687904e-05,
      "loss": 0.9713,
      "step": 5318
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7928895950317383,
      "learning_rate": 9.7836040802914e-05,
      "loss": 0.9543,
      "step": 5319
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8589850068092346,
      "learning_rate": 9.780599060445039e-05,
      "loss": 0.8259,
      "step": 5320
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7704259753227234,
      "learning_rate": 9.777594060420312e-05,
      "loss": 0.916,
      "step": 5321
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7357114553451538,
      "learning_rate": 9.774589080488702e-05,
      "loss": 0.9275,
      "step": 5322
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8844978213310242,
      "learning_rate": 9.771584120921692e-05,
      "loss": 1.2205,
      "step": 5323
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.832973062992096,
      "learning_rate": 9.768579181990764e-05,
      "loss": 1.0149,
      "step": 5324
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8065707683563232,
      "learning_rate": 9.765574263967396e-05,
      "loss": 0.979,
      "step": 5325
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7501281499862671,
      "learning_rate": 9.762569367123067e-05,
      "loss": 0.9781,
      "step": 5326
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9292058348655701,
      "learning_rate": 9.75956449172925e-05,
      "loss": 0.9833,
      "step": 5327
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8146618604660034,
      "learning_rate": 9.75655963805742e-05,
      "loss": 0.9133,
      "step": 5328
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.760799765586853,
      "learning_rate": 9.753554806379051e-05,
      "loss": 1.1945,
      "step": 5329
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8132180571556091,
      "learning_rate": 9.750549996965607e-05,
      "loss": 1.1495,
      "step": 5330
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8438681960105896,
      "learning_rate": 9.74754521008856e-05,
      "loss": 1.0982,
      "step": 5331
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7601552605628967,
      "learning_rate": 9.744540446019372e-05,
      "loss": 0.9861,
      "step": 5332
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8696185350418091,
      "learning_rate": 9.741535705029508e-05,
      "loss": 1.069,
      "step": 5333
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7651448845863342,
      "learning_rate": 9.738530987390433e-05,
      "loss": 1.0204,
      "step": 5334
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.5807250738143921,
      "learning_rate": 9.735526293373599e-05,
      "loss": 0.6925,
      "step": 5335
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9367736577987671,
      "learning_rate": 9.732521623250473e-05,
      "loss": 0.7211,
      "step": 5336
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7551747560501099,
      "learning_rate": 9.729516977292496e-05,
      "loss": 0.8202,
      "step": 5337
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7845062017440796,
      "learning_rate": 9.726512355771132e-05,
      "loss": 0.7973,
      "step": 5338
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7183378338813782,
      "learning_rate": 9.723507758957832e-05,
      "loss": 0.8383,
      "step": 5339
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.700423538684845,
      "learning_rate": 9.720503187124034e-05,
      "loss": 0.8643,
      "step": 5340
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7689960598945618,
      "learning_rate": 9.717498640541196e-05,
      "loss": 0.8637,
      "step": 5341
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7119944095611572,
      "learning_rate": 9.714494119480753e-05,
      "loss": 0.9275,
      "step": 5342
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.719994068145752,
      "learning_rate": 9.711489624214147e-05,
      "loss": 0.8171,
      "step": 5343
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9013209939002991,
      "learning_rate": 9.708485155012827e-05,
      "loss": 0.709,
      "step": 5344
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7073320746421814,
      "learning_rate": 9.70548071214822e-05,
      "loss": 0.7942,
      "step": 5345
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.711128830909729,
      "learning_rate": 9.702476295891763e-05,
      "loss": 0.7392,
      "step": 5346
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8236702680587769,
      "learning_rate": 9.699471906514887e-05,
      "loss": 0.7073,
      "step": 5347
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7535871863365173,
      "learning_rate": 9.696467544289018e-05,
      "loss": 0.8832,
      "step": 5348
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.749582827091217,
      "learning_rate": 9.693463209485597e-05,
      "loss": 0.8044,
      "step": 5349
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8349359631538391,
      "learning_rate": 9.690458902376036e-05,
      "loss": 0.9137,
      "step": 5350
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8222732543945312,
      "learning_rate": 9.687454623231759e-05,
      "loss": 0.7883,
      "step": 5351
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8505678772926331,
      "learning_rate": 9.684450372324188e-05,
      "loss": 0.8457,
      "step": 5352
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8476554155349731,
      "learning_rate": 9.681446149924737e-05,
      "loss": 1.0079,
      "step": 5353
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8563003540039062,
      "learning_rate": 9.678441956304827e-05,
      "loss": 0.8173,
      "step": 5354
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.773239254951477,
      "learning_rate": 9.675437791735862e-05,
      "loss": 0.8251,
      "step": 5355
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7639120221138,
      "learning_rate": 9.672433656489258e-05,
      "loss": 0.8415,
      "step": 5356
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7628129720687866,
      "learning_rate": 9.669429550836416e-05,
      "loss": 0.799,
      "step": 5357
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8292572498321533,
      "learning_rate": 9.666425475048742e-05,
      "loss": 0.7951,
      "step": 5358
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7805098295211792,
      "learning_rate": 9.66342142939764e-05,
      "loss": 0.8922,
      "step": 5359
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.713222324848175,
      "learning_rate": 9.660417414154504e-05,
      "loss": 0.9142,
      "step": 5360
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7346338033676147,
      "learning_rate": 9.657413429590735e-05,
      "loss": 0.8735,
      "step": 5361
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7653969526290894,
      "learning_rate": 9.654409475977719e-05,
      "loss": 0.8297,
      "step": 5362
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.9530655741691589,
      "learning_rate": 9.651405553586851e-05,
      "loss": 0.8566,
      "step": 5363
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7035889029502869,
      "learning_rate": 9.648401662689521e-05,
      "loss": 0.827,
      "step": 5364
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8551998734474182,
      "learning_rate": 9.645397803557106e-05,
      "loss": 0.7667,
      "step": 5365
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8492952585220337,
      "learning_rate": 9.642393976460994e-05,
      "loss": 0.775,
      "step": 5366
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7752746343612671,
      "learning_rate": 9.63939018167256e-05,
      "loss": 0.826,
      "step": 5367
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7863603830337524,
      "learning_rate": 9.63638641946318e-05,
      "loss": 0.8421,
      "step": 5368
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8763649463653564,
      "learning_rate": 9.633382690104232e-05,
      "loss": 0.7575,
      "step": 5369
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7970002293586731,
      "learning_rate": 9.63037899386708e-05,
      "loss": 0.8454,
      "step": 5370
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8034215569496155,
      "learning_rate": 9.627375331023096e-05,
      "loss": 0.8343,
      "step": 5371
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8347408175468445,
      "learning_rate": 9.624371701843636e-05,
      "loss": 0.7685,
      "step": 5372
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.821174144744873,
      "learning_rate": 9.621368106600067e-05,
      "loss": 0.8305,
      "step": 5373
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.9249157905578613,
      "learning_rate": 9.61836454556375e-05,
      "loss": 0.8789,
      "step": 5374
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.9051786065101624,
      "learning_rate": 9.61536101900603e-05,
      "loss": 0.8459,
      "step": 5375
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7552970051765442,
      "learning_rate": 9.612357527198271e-05,
      "loss": 0.6945,
      "step": 5376
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8446747660636902,
      "learning_rate": 9.609354070411807e-05,
      "loss": 0.8362,
      "step": 5377
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7395488619804382,
      "learning_rate": 9.606350648917993e-05,
      "loss": 0.6508,
      "step": 5378
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7817243933677673,
      "learning_rate": 9.603347262988173e-05,
      "loss": 0.8063,
      "step": 5379
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9117684364318848,
      "learning_rate": 9.600343912893677e-05,
      "loss": 0.8674,
      "step": 5380
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.756833553314209,
      "learning_rate": 9.597340598905852e-05,
      "loss": 0.6625,
      "step": 5381
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7395047545433044,
      "learning_rate": 9.594337321296018e-05,
      "loss": 0.7593,
      "step": 5382
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9986709952354431,
      "learning_rate": 9.591334080335509e-05,
      "loss": 0.8484,
      "step": 5383
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8254805207252502,
      "learning_rate": 9.58833087629566e-05,
      "loss": 0.7311,
      "step": 5384
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7360950112342834,
      "learning_rate": 9.58532770944778e-05,
      "loss": 0.7988,
      "step": 5385
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8294689655303955,
      "learning_rate": 9.582324580063194e-05,
      "loss": 0.763,
      "step": 5386
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.861795961856842,
      "learning_rate": 9.579321488413217e-05,
      "loss": 0.9623,
      "step": 5387
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.785373866558075,
      "learning_rate": 9.576318434769161e-05,
      "loss": 0.8922,
      "step": 5388
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.773952066898346,
      "learning_rate": 9.573315419402338e-05,
      "loss": 0.7433,
      "step": 5389
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7986981868743896,
      "learning_rate": 9.570312442584048e-05,
      "loss": 0.8083,
      "step": 5390
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7751654982566833,
      "learning_rate": 9.567309504585598e-05,
      "loss": 0.7328,
      "step": 5391
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8302088379859924,
      "learning_rate": 9.564306605678284e-05,
      "loss": 0.8564,
      "step": 5392
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7000736594200134,
      "learning_rate": 9.5613037461334e-05,
      "loss": 0.7825,
      "step": 5393
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.778109610080719,
      "learning_rate": 9.558300926222241e-05,
      "loss": 0.7689,
      "step": 5394
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8596503734588623,
      "learning_rate": 9.55529814621609e-05,
      "loss": 0.8508,
      "step": 5395
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8834884166717529,
      "learning_rate": 9.552295406386238e-05,
      "loss": 0.8606,
      "step": 5396
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.840469479560852,
      "learning_rate": 9.549292707003956e-05,
      "loss": 0.8142,
      "step": 5397
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9146082401275635,
      "learning_rate": 9.546290048340526e-05,
      "loss": 0.8043,
      "step": 5398
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7533279061317444,
      "learning_rate": 9.543287430667224e-05,
      "loss": 0.7517,
      "step": 5399
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.7766563296318054,
      "learning_rate": 9.540284854255316e-05,
      "loss": 0.7442,
      "step": 5400
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8230363726615906,
      "learning_rate": 9.537282319376071e-05,
      "loss": 0.7892,
      "step": 5401
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0036050081253052,
      "learning_rate": 9.534279826300744e-05,
      "loss": 0.8415,
      "step": 5402
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.7596420049667358,
      "learning_rate": 9.531277375300599e-05,
      "loss": 0.8648,
      "step": 5403
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8162840008735657,
      "learning_rate": 9.528274966646892e-05,
      "loss": 0.7478,
      "step": 5404
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8774458169937134,
      "learning_rate": 9.525272600610867e-05,
      "loss": 0.916,
      "step": 5405
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9435096383094788,
      "learning_rate": 9.522270277463778e-05,
      "loss": 0.8668,
      "step": 5406
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9227274060249329,
      "learning_rate": 9.519267997476862e-05,
      "loss": 0.739,
      "step": 5407
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8549928069114685,
      "learning_rate": 9.51626576092136e-05,
      "loss": 0.7067,
      "step": 5408
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.7779527306556702,
      "learning_rate": 9.513263568068512e-05,
      "loss": 0.7269,
      "step": 5409
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9053125381469727,
      "learning_rate": 9.510261419189541e-05,
      "loss": 0.946,
      "step": 5410
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8333085775375366,
      "learning_rate": 9.507259314555681e-05,
      "loss": 0.897,
      "step": 5411
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.7893860936164856,
      "learning_rate": 9.504257254438148e-05,
      "loss": 0.7189,
      "step": 5412
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8207554817199707,
      "learning_rate": 9.501255239108167e-05,
      "loss": 0.7308,
      "step": 5413
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9493604898452759,
      "learning_rate": 9.498253268836954e-05,
      "loss": 0.7729,
      "step": 5414
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.7668083310127258,
      "learning_rate": 9.495251343895712e-05,
      "loss": 0.6177,
      "step": 5415
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8166645765304565,
      "learning_rate": 9.492249464555658e-05,
      "loss": 0.7806,
      "step": 5416
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8242104053497314,
      "learning_rate": 9.489247631087987e-05,
      "loss": 0.6867,
      "step": 5417
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8475474715232849,
      "learning_rate": 9.4862458437639e-05,
      "loss": 0.8646,
      "step": 5418
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8722558617591858,
      "learning_rate": 9.483244102854596e-05,
      "loss": 0.8745,
      "step": 5419
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.977030873298645,
      "learning_rate": 9.480242408631258e-05,
      "loss": 0.997,
      "step": 5420
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8375899195671082,
      "learning_rate": 9.477240761365078e-05,
      "loss": 0.866,
      "step": 5421
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8502902388572693,
      "learning_rate": 9.474239161327234e-05,
      "loss": 0.7542,
      "step": 5422
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.7692695260047913,
      "learning_rate": 9.4712376087889e-05,
      "loss": 0.758,
      "step": 5423
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8772521615028381,
      "learning_rate": 9.468236104021263e-05,
      "loss": 0.8157,
      "step": 5424
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.811822772026062,
      "learning_rate": 9.465234647295479e-05,
      "loss": 0.8019,
      "step": 5425
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.820911169052124,
      "learning_rate": 9.462233238882719e-05,
      "loss": 0.7661,
      "step": 5426
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.7837050557136536,
      "learning_rate": 9.459231879054137e-05,
      "loss": 0.8949,
      "step": 5427
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8992356657981873,
      "learning_rate": 9.456230568080893e-05,
      "loss": 0.7137,
      "step": 5428
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9014803767204285,
      "learning_rate": 9.453229306234143e-05,
      "loss": 0.7944,
      "step": 5429
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9340708255767822,
      "learning_rate": 9.450228093785025e-05,
      "loss": 0.8931,
      "step": 5430
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9139885306358337,
      "learning_rate": 9.447226931004689e-05,
      "loss": 0.8763,
      "step": 5431
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.7462260127067566,
      "learning_rate": 9.444225818164268e-05,
      "loss": 0.7181,
      "step": 5432
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8099203109741211,
      "learning_rate": 9.441224755534896e-05,
      "loss": 0.817,
      "step": 5433
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.84419846534729,
      "learning_rate": 9.438223743387709e-05,
      "loss": 0.8604,
      "step": 5434
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8317033648490906,
      "learning_rate": 9.435222781993821e-05,
      "loss": 0.7222,
      "step": 5435
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.81523197889328,
      "learning_rate": 9.43222187162436e-05,
      "loss": 0.902,
      "step": 5436
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8841376900672913,
      "learning_rate": 9.429221012550436e-05,
      "loss": 0.7186,
      "step": 5437
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.7745022773742676,
      "learning_rate": 9.426220205043162e-05,
      "loss": 0.8232,
      "step": 5438
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.7608467936515808,
      "learning_rate": 9.423219449373645e-05,
      "loss": 0.7752,
      "step": 5439
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.7767959237098694,
      "learning_rate": 9.420218745812983e-05,
      "loss": 0.7236,
      "step": 5440
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9423446655273438,
      "learning_rate": 9.417218094632278e-05,
      "loss": 0.7962,
      "step": 5441
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8715342283248901,
      "learning_rate": 9.414217496102614e-05,
      "loss": 0.8702,
      "step": 5442
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9314033389091492,
      "learning_rate": 9.411216950495084e-05,
      "loss": 1.0354,
      "step": 5443
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9917663931846619,
      "learning_rate": 9.408216458080772e-05,
      "loss": 0.9122,
      "step": 5444
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.7052233815193176,
      "learning_rate": 9.40521601913075e-05,
      "loss": 0.6633,
      "step": 5445
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8609520196914673,
      "learning_rate": 9.402215633916095e-05,
      "loss": 0.7533,
      "step": 5446
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8447232246398926,
      "learning_rate": 9.399215302707868e-05,
      "loss": 0.8033,
      "step": 5447
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8273617029190063,
      "learning_rate": 9.396215025777139e-05,
      "loss": 0.7448,
      "step": 5448
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9455866813659668,
      "learning_rate": 9.393214803394968e-05,
      "loss": 1.0166,
      "step": 5449
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8604373931884766,
      "learning_rate": 9.390214635832398e-05,
      "loss": 0.6915,
      "step": 5450
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.968650221824646,
      "learning_rate": 9.387214523360489e-05,
      "loss": 0.8138,
      "step": 5451
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8356905579566956,
      "learning_rate": 9.384214466250273e-05,
      "loss": 0.7836,
      "step": 5452
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0817919969558716,
      "learning_rate": 9.381214464772792e-05,
      "loss": 0.9852,
      "step": 5453
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8433053493499756,
      "learning_rate": 9.378214519199087e-05,
      "loss": 0.8783,
      "step": 5454
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9286566972732544,
      "learning_rate": 9.375214629800173e-05,
      "loss": 0.9176,
      "step": 5455
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0043586492538452,
      "learning_rate": 9.372214796847086e-05,
      "loss": 0.7709,
      "step": 5456
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8386707901954651,
      "learning_rate": 9.369215020610834e-05,
      "loss": 0.7308,
      "step": 5457
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9132738709449768,
      "learning_rate": 9.366215301362431e-05,
      "loss": 0.7898,
      "step": 5458
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8323861956596375,
      "learning_rate": 9.363215639372893e-05,
      "loss": 0.7991,
      "step": 5459
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8959023952484131,
      "learning_rate": 9.360216034913213e-05,
      "loss": 0.8941,
      "step": 5460
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8528773784637451,
      "learning_rate": 9.357216488254393e-05,
      "loss": 0.884,
      "step": 5461
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0008320808410645,
      "learning_rate": 9.354216999667423e-05,
      "loss": 0.7154,
      "step": 5462
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9015179872512817,
      "learning_rate": 9.35121756942329e-05,
      "loss": 0.8794,
      "step": 5463
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8274909853935242,
      "learning_rate": 9.348218197792978e-05,
      "loss": 0.8211,
      "step": 5464
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.7812042832374573,
      "learning_rate": 9.345218885047459e-05,
      "loss": 0.7786,
      "step": 5465
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1789311170578003,
      "learning_rate": 9.342219631457709e-05,
      "loss": 0.8566,
      "step": 5466
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9687840342521667,
      "learning_rate": 9.339220437294689e-05,
      "loss": 0.8258,
      "step": 5467
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.970503032207489,
      "learning_rate": 9.33622130282936e-05,
      "loss": 0.7277,
      "step": 5468
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8009697198867798,
      "learning_rate": 9.333222228332683e-05,
      "loss": 0.8979,
      "step": 5469
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8295863270759583,
      "learning_rate": 9.330223214075598e-05,
      "loss": 0.8943,
      "step": 5470
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.7602128982543945,
      "learning_rate": 9.327224260329056e-05,
      "loss": 0.7821,
      "step": 5471
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9184693694114685,
      "learning_rate": 9.32422536736399e-05,
      "loss": 1.0241,
      "step": 5472
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8481183648109436,
      "learning_rate": 9.321226535451337e-05,
      "loss": 0.7563,
      "step": 5473
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8175013065338135,
      "learning_rate": 9.318227764862026e-05,
      "loss": 0.748,
      "step": 5474
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.858511209487915,
      "learning_rate": 9.315229055866971e-05,
      "loss": 0.7585,
      "step": 5475
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8355712890625,
      "learning_rate": 9.312230408737099e-05,
      "loss": 0.8897,
      "step": 5476
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9581938982009888,
      "learning_rate": 9.309231823743312e-05,
      "loss": 0.7681,
      "step": 5477
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.7755740880966187,
      "learning_rate": 9.306233301156518e-05,
      "loss": 0.786,
      "step": 5478
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9166685342788696,
      "learning_rate": 9.303234841247621e-05,
      "loss": 0.8348,
      "step": 5479
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9045132994651794,
      "learning_rate": 9.300236444287508e-05,
      "loss": 0.9487,
      "step": 5480
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8675636053085327,
      "learning_rate": 9.297238110547074e-05,
      "loss": 0.7866,
      "step": 5481
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8145818710327148,
      "learning_rate": 9.294239840297191e-05,
      "loss": 0.797,
      "step": 5482
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8576038479804993,
      "learning_rate": 9.291241633808745e-05,
      "loss": 0.9114,
      "step": 5483
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.7921628355979919,
      "learning_rate": 9.288243491352607e-05,
      "loss": 0.7517,
      "step": 5484
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9601874351501465,
      "learning_rate": 9.285245413199637e-05,
      "loss": 0.933,
      "step": 5485
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8971057534217834,
      "learning_rate": 9.282247399620703e-05,
      "loss": 0.7563,
      "step": 5486
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9189906120300293,
      "learning_rate": 9.279249450886643e-05,
      "loss": 0.7375,
      "step": 5487
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8775677680969238,
      "learning_rate": 9.276251567268317e-05,
      "loss": 0.7742,
      "step": 5488
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9255915880203247,
      "learning_rate": 9.273253749036571e-05,
      "loss": 0.8354,
      "step": 5489
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9448326826095581,
      "learning_rate": 9.270255996462226e-05,
      "loss": 0.947,
      "step": 5490
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.871783971786499,
      "learning_rate": 9.267258309816126e-05,
      "loss": 0.7042,
      "step": 5491
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1187124252319336,
      "learning_rate": 9.264260689369085e-05,
      "loss": 0.918,
      "step": 5492
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8499632477760315,
      "learning_rate": 9.261263135391922e-05,
      "loss": 0.8241,
      "step": 5493
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9623643755912781,
      "learning_rate": 9.25826564815546e-05,
      "loss": 0.7799,
      "step": 5494
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8706701397895813,
      "learning_rate": 9.255268227930492e-05,
      "loss": 0.8103,
      "step": 5495
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.7522194385528564,
      "learning_rate": 9.252270874987824e-05,
      "loss": 0.7191,
      "step": 5496
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9570392370223999,
      "learning_rate": 9.249273589598248e-05,
      "loss": 0.8638,
      "step": 5497
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9298526048660278,
      "learning_rate": 9.246276372032552e-05,
      "loss": 0.8942,
      "step": 5498
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8214056491851807,
      "learning_rate": 9.24327922256152e-05,
      "loss": 0.7832,
      "step": 5499
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8473328948020935,
      "learning_rate": 9.240282141455924e-05,
      "loss": 0.8746,
      "step": 5500
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.953040599822998,
      "learning_rate": 9.237285128986537e-05,
      "loss": 0.8873,
      "step": 5501
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8263955116271973,
      "learning_rate": 9.234288185424118e-05,
      "loss": 0.8502,
      "step": 5502
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8837599754333496,
      "learning_rate": 9.231291311039426e-05,
      "loss": 0.8835,
      "step": 5503
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8725624084472656,
      "learning_rate": 9.228294506103214e-05,
      "loss": 0.874,
      "step": 5504
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9597254395484924,
      "learning_rate": 9.225297770886222e-05,
      "loss": 0.8396,
      "step": 5505
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9080321192741394,
      "learning_rate": 9.222301105659192e-05,
      "loss": 0.7449,
      "step": 5506
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.7567276358604431,
      "learning_rate": 9.219304510692852e-05,
      "loss": 0.5697,
      "step": 5507
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8823275566101074,
      "learning_rate": 9.216307986257928e-05,
      "loss": 0.8734,
      "step": 5508
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8761025071144104,
      "learning_rate": 9.213311532625144e-05,
      "loss": 0.9315,
      "step": 5509
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9233946204185486,
      "learning_rate": 9.210315150065205e-05,
      "loss": 0.7459,
      "step": 5510
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8146713376045227,
      "learning_rate": 9.207318838848824e-05,
      "loss": 0.8198,
      "step": 5511
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8557800650596619,
      "learning_rate": 9.204322599246695e-05,
      "loss": 0.7685,
      "step": 5512
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.7936701774597168,
      "learning_rate": 9.201326431529516e-05,
      "loss": 0.7421,
      "step": 5513
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8390325903892517,
      "learning_rate": 9.198330335967972e-05,
      "loss": 0.6661,
      "step": 5514
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8884896636009216,
      "learning_rate": 9.195334312832742e-05,
      "loss": 0.8321,
      "step": 5515
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9207426905632019,
      "learning_rate": 9.192338362394501e-05,
      "loss": 0.7698,
      "step": 5516
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9065296053886414,
      "learning_rate": 9.189342484923916e-05,
      "loss": 0.7976,
      "step": 5517
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8725468516349792,
      "learning_rate": 9.186346680691647e-05,
      "loss": 0.915,
      "step": 5518
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8427053689956665,
      "learning_rate": 9.183350949968352e-05,
      "loss": 0.7072,
      "step": 5519
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0602563619613647,
      "learning_rate": 9.180355293024669e-05,
      "loss": 0.9017,
      "step": 5520
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8901358246803284,
      "learning_rate": 9.177359710131253e-05,
      "loss": 0.9988,
      "step": 5521
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.785239577293396,
      "learning_rate": 9.17436420155872e-05,
      "loss": 0.72,
      "step": 5522
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9820289015769958,
      "learning_rate": 9.171368767577709e-05,
      "loss": 0.8722,
      "step": 5523
    },
    {
      "epoch": 2.07,
      "grad_norm": Infinity,
      "learning_rate": 9.171368767577709e-05,
      "loss": 0.8213,
      "step": 5524
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.87762850522995,
      "learning_rate": 9.168373408458844e-05,
      "loss": 0.8056,
      "step": 5525
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.7401769161224365,
      "learning_rate": 9.165378124472726e-05,
      "loss": 0.8221,
      "step": 5526
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.7532601356506348,
      "learning_rate": 9.162382915889974e-05,
      "loss": 0.7237,
      "step": 5527
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0011965036392212,
      "learning_rate": 9.15938778298118e-05,
      "loss": 0.7385,
      "step": 5528
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8625580668449402,
      "learning_rate": 9.156392726016938e-05,
      "loss": 0.7943,
      "step": 5529
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8107515573501587,
      "learning_rate": 9.153397745267843e-05,
      "loss": 0.8285,
      "step": 5530
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9454732537269592,
      "learning_rate": 9.150402841004462e-05,
      "loss": 0.8911,
      "step": 5531
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9602572321891785,
      "learning_rate": 9.14740801349738e-05,
      "loss": 0.7222,
      "step": 5532
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9064531922340393,
      "learning_rate": 9.144413263017152e-05,
      "loss": 0.8751,
      "step": 5533
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.749590277671814,
      "learning_rate": 9.141418589834339e-05,
      "loss": 0.6223,
      "step": 5534
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.7996737957000732,
      "learning_rate": 9.138423994219502e-05,
      "loss": 0.7685,
      "step": 5535
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9227060079574585,
      "learning_rate": 9.135429476443174e-05,
      "loss": 0.8372,
      "step": 5536
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9365580081939697,
      "learning_rate": 9.132435036775901e-05,
      "loss": 0.8447,
      "step": 5537
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8712472915649414,
      "learning_rate": 9.129440675488208e-05,
      "loss": 0.7561,
      "step": 5538
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0136624574661255,
      "learning_rate": 9.12644639285062e-05,
      "loss": 1.0203,
      "step": 5539
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8654160499572754,
      "learning_rate": 9.123452189133657e-05,
      "loss": 0.7715,
      "step": 5540
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9060373902320862,
      "learning_rate": 9.120458064607825e-05,
      "loss": 0.788,
      "step": 5541
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9991627931594849,
      "learning_rate": 9.117464019543627e-05,
      "loss": 0.7239,
      "step": 5542
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8686838150024414,
      "learning_rate": 9.114470054211556e-05,
      "loss": 0.9138,
      "step": 5543
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8661286234855652,
      "learning_rate": 9.111476168882102e-05,
      "loss": 0.8007,
      "step": 5544
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8300328254699707,
      "learning_rate": 9.108482363825749e-05,
      "loss": 0.7168,
      "step": 5545
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9890086054801941,
      "learning_rate": 9.105488639312962e-05,
      "loss": 0.7787,
      "step": 5546
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9272122979164124,
      "learning_rate": 9.102494995614215e-05,
      "loss": 0.7401,
      "step": 5547
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9605776071548462,
      "learning_rate": 9.099501432999962e-05,
      "loss": 0.7843,
      "step": 5548
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9065970778465271,
      "learning_rate": 9.096507951740656e-05,
      "loss": 0.8953,
      "step": 5549
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9268360137939453,
      "learning_rate": 9.093514552106743e-05,
      "loss": 0.7946,
      "step": 5550
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8034109473228455,
      "learning_rate": 9.090521234368654e-05,
      "loss": 0.7973,
      "step": 5551
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9844200015068054,
      "learning_rate": 9.087527998796825e-05,
      "loss": 0.9567,
      "step": 5552
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8318657279014587,
      "learning_rate": 9.084534845661673e-05,
      "loss": 0.8085,
      "step": 5553
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9146385192871094,
      "learning_rate": 9.081541775233615e-05,
      "loss": 0.8379,
      "step": 5554
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8079643845558167,
      "learning_rate": 9.078548787783059e-05,
      "loss": 0.8194,
      "step": 5555
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.7781199812889099,
      "learning_rate": 9.0755558835804e-05,
      "loss": 0.7889,
      "step": 5556
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.7658002972602844,
      "learning_rate": 9.072563062896038e-05,
      "loss": 0.8061,
      "step": 5557
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8341374397277832,
      "learning_rate": 9.069570326000344e-05,
      "loss": 0.8124,
      "step": 5558
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9441350698471069,
      "learning_rate": 9.066577673163707e-05,
      "loss": 0.8503,
      "step": 5559
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8286848664283752,
      "learning_rate": 9.063585104656493e-05,
      "loss": 0.7851,
      "step": 5560
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8983969688415527,
      "learning_rate": 9.06059262074906e-05,
      "loss": 0.9856,
      "step": 5561
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9185383915901184,
      "learning_rate": 9.057600221711772e-05,
      "loss": 0.8021,
      "step": 5562
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9250062108039856,
      "learning_rate": 9.05460790781496e-05,
      "loss": 0.8806,
      "step": 5563
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8666012287139893,
      "learning_rate": 9.051615679328974e-05,
      "loss": 0.9271,
      "step": 5564
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8660537004470825,
      "learning_rate": 9.048623536524143e-05,
      "loss": 0.7203,
      "step": 5565
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9456992745399475,
      "learning_rate": 9.045631479670784e-05,
      "loss": 0.6684,
      "step": 5566
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9416536688804626,
      "learning_rate": 9.042639509039225e-05,
      "loss": 0.8448,
      "step": 5567
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8631331324577332,
      "learning_rate": 9.03964762489976e-05,
      "loss": 0.7658,
      "step": 5568
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9214067459106445,
      "learning_rate": 9.036655827522693e-05,
      "loss": 0.914,
      "step": 5569
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8794437646865845,
      "learning_rate": 9.033664117178323e-05,
      "loss": 0.8155,
      "step": 5570
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9746697545051575,
      "learning_rate": 9.030672494136926e-05,
      "loss": 0.9867,
      "step": 5571
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.878689169883728,
      "learning_rate": 9.027680958668781e-05,
      "loss": 0.8291,
      "step": 5572
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8388155102729797,
      "learning_rate": 9.024689511044155e-05,
      "loss": 0.9092,
      "step": 5573
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8395091891288757,
      "learning_rate": 9.02169815153331e-05,
      "loss": 0.804,
      "step": 5574
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8379384875297546,
      "learning_rate": 9.0187068804065e-05,
      "loss": 1.0142,
      "step": 5575
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0258941650390625,
      "learning_rate": 9.015715697933964e-05,
      "loss": 0.9823,
      "step": 5576
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8639580011367798,
      "learning_rate": 9.012724604385945e-05,
      "loss": 0.8602,
      "step": 5577
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9037371277809143,
      "learning_rate": 9.009733600032666e-05,
      "loss": 0.7436,
      "step": 5578
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9421469569206238,
      "learning_rate": 9.006742685144348e-05,
      "loss": 0.8686,
      "step": 5579
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8989153504371643,
      "learning_rate": 9.003751859991208e-05,
      "loss": 0.9292,
      "step": 5580
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8879985809326172,
      "learning_rate": 9.000761124843444e-05,
      "loss": 0.8379,
      "step": 5581
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9677584171295166,
      "learning_rate": 8.997770479971257e-05,
      "loss": 0.9089,
      "step": 5582
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8893254399299622,
      "learning_rate": 8.994779925644831e-05,
      "loss": 0.9109,
      "step": 5583
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8634993433952332,
      "learning_rate": 8.991789462134346e-05,
      "loss": 0.8429,
      "step": 5584
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9128538966178894,
      "learning_rate": 8.988799089709977e-05,
      "loss": 0.8627,
      "step": 5585
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8412069082260132,
      "learning_rate": 8.985808808641883e-05,
      "loss": 0.8108,
      "step": 5586
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9386151432991028,
      "learning_rate": 8.982818619200223e-05,
      "loss": 0.8765,
      "step": 5587
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8001821637153625,
      "learning_rate": 8.97982852165514e-05,
      "loss": 0.8323,
      "step": 5588
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8746115565299988,
      "learning_rate": 8.976838516276773e-05,
      "loss": 0.8639,
      "step": 5589
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8036655187606812,
      "learning_rate": 8.973848603335255e-05,
      "loss": 0.769,
      "step": 5590
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8981305360794067,
      "learning_rate": 8.970858783100704e-05,
      "loss": 0.7686,
      "step": 5591
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8977667689323425,
      "learning_rate": 8.967869055843237e-05,
      "loss": 0.7699,
      "step": 5592
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8719660043716431,
      "learning_rate": 8.964879421832955e-05,
      "loss": 0.7822,
      "step": 5593
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9096459150314331,
      "learning_rate": 8.961889881339956e-05,
      "loss": 0.9042,
      "step": 5594
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9188867211341858,
      "learning_rate": 8.95890043463433e-05,
      "loss": 0.7447,
      "step": 5595
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8987383246421814,
      "learning_rate": 8.955911081986152e-05,
      "loss": 0.7457,
      "step": 5596
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8577207922935486,
      "learning_rate": 8.952921823665503e-05,
      "loss": 0.7674,
      "step": 5597
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8443433046340942,
      "learning_rate": 8.949932659942431e-05,
      "loss": 0.8067,
      "step": 5598
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8788016438484192,
      "learning_rate": 8.946943591087e-05,
      "loss": 0.6933,
      "step": 5599
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9981191754341125,
      "learning_rate": 8.943954617369259e-05,
      "loss": 0.8573,
      "step": 5600
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8851143717765808,
      "learning_rate": 8.940965739059229e-05,
      "loss": 0.7064,
      "step": 5601
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8657859563827515,
      "learning_rate": 8.937976956426958e-05,
      "loss": 0.9163,
      "step": 5602
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9790180325508118,
      "learning_rate": 8.934988269742449e-05,
      "loss": 0.8596,
      "step": 5603
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8869326114654541,
      "learning_rate": 8.931999679275718e-05,
      "loss": 0.7649,
      "step": 5604
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9774800539016724,
      "learning_rate": 8.929011185296777e-05,
      "loss": 0.8307,
      "step": 5605
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.890600860118866,
      "learning_rate": 8.926022788075605e-05,
      "loss": 0.9566,
      "step": 5606
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8603239059448242,
      "learning_rate": 8.923034487882199e-05,
      "loss": 0.693,
      "step": 5607
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9072414040565491,
      "learning_rate": 8.920046284986525e-05,
      "loss": 0.8165,
      "step": 5608
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9458053112030029,
      "learning_rate": 8.917058179658552e-05,
      "loss": 0.9318,
      "step": 5609
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.015374779701233,
      "learning_rate": 8.914070172168247e-05,
      "loss": 0.8877,
      "step": 5610
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8642667531967163,
      "learning_rate": 8.91108226278555e-05,
      "loss": 0.853,
      "step": 5611
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9268933534622192,
      "learning_rate": 8.908094451780408e-05,
      "loss": 0.8249,
      "step": 5612
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8148179650306702,
      "learning_rate": 8.905106739422746e-05,
      "loss": 0.7413,
      "step": 5613
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9090878963470459,
      "learning_rate": 8.902119125982493e-05,
      "loss": 0.7992,
      "step": 5614
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8897498250007629,
      "learning_rate": 8.899131611729559e-05,
      "loss": 0.7685,
      "step": 5615
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.7701549530029297,
      "learning_rate": 8.896144196933852e-05,
      "loss": 0.6195,
      "step": 5616
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0228956937789917,
      "learning_rate": 8.89315688186527e-05,
      "loss": 0.99,
      "step": 5617
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.018775224685669,
      "learning_rate": 8.89016966679369e-05,
      "loss": 0.8547,
      "step": 5618
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9844170808792114,
      "learning_rate": 8.887182551989e-05,
      "loss": 0.7201,
      "step": 5619
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9265791773796082,
      "learning_rate": 8.884195537721068e-05,
      "loss": 0.9062,
      "step": 5620
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9837334752082825,
      "learning_rate": 8.88120862425975e-05,
      "loss": 0.7844,
      "step": 5621
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9729360342025757,
      "learning_rate": 8.878221811874898e-05,
      "loss": 0.7753,
      "step": 5622
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8806496262550354,
      "learning_rate": 8.875235100836352e-05,
      "loss": 0.757,
      "step": 5623
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8464471101760864,
      "learning_rate": 8.872248491413946e-05,
      "loss": 0.7898,
      "step": 5624
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9058859944343567,
      "learning_rate": 8.869261983877508e-05,
      "loss": 0.829,
      "step": 5625
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.006860375404358,
      "learning_rate": 8.866275578496845e-05,
      "loss": 0.8457,
      "step": 5626
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9947961568832397,
      "learning_rate": 8.863289275541763e-05,
      "loss": 1.0457,
      "step": 5627
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8027253746986389,
      "learning_rate": 8.860303075282061e-05,
      "loss": 0.7683,
      "step": 5628
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8853110074996948,
      "learning_rate": 8.857316977987521e-05,
      "loss": 0.9286,
      "step": 5629
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8881149291992188,
      "learning_rate": 8.854330983927926e-05,
      "loss": 0.6816,
      "step": 5630
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.076610803604126,
      "learning_rate": 8.851345093373036e-05,
      "loss": 0.8351,
      "step": 5631
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.968830406665802,
      "learning_rate": 8.848359306592618e-05,
      "loss": 0.9351,
      "step": 5632
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9844539165496826,
      "learning_rate": 8.84537362385641e-05,
      "loss": 0.896,
      "step": 5633
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0906575918197632,
      "learning_rate": 8.842388045434157e-05,
      "loss": 1.071,
      "step": 5634
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8878204226493835,
      "learning_rate": 8.839402571595596e-05,
      "loss": 0.9571,
      "step": 5635
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8960426449775696,
      "learning_rate": 8.836417202610436e-05,
      "loss": 0.7548,
      "step": 5636
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8859007358551025,
      "learning_rate": 8.8334319387484e-05,
      "loss": 0.9642,
      "step": 5637
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8490790128707886,
      "learning_rate": 8.830446780279176e-05,
      "loss": 0.8531,
      "step": 5638
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8480827808380127,
      "learning_rate": 8.827461727472466e-05,
      "loss": 0.8374,
      "step": 5639
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.816676676273346,
      "learning_rate": 8.824476780597954e-05,
      "loss": 0.8253,
      "step": 5640
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9016271829605103,
      "learning_rate": 8.821491939925301e-05,
      "loss": 0.8563,
      "step": 5641
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9887182712554932,
      "learning_rate": 8.818507205724187e-05,
      "loss": 0.8933,
      "step": 5642
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9692315459251404,
      "learning_rate": 8.815522578264253e-05,
      "loss": 0.811,
      "step": 5643
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8304370641708374,
      "learning_rate": 8.812538057815145e-05,
      "loss": 0.8215,
      "step": 5644
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8376574516296387,
      "learning_rate": 8.809553644646507e-05,
      "loss": 0.7718,
      "step": 5645
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.7389733195304871,
      "learning_rate": 8.806569339027951e-05,
      "loss": 0.6607,
      "step": 5646
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9403268098831177,
      "learning_rate": 8.803585141229102e-05,
      "loss": 0.9106,
      "step": 5647
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8031682372093201,
      "learning_rate": 8.800601051519558e-05,
      "loss": 0.7065,
      "step": 5648
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9023293256759644,
      "learning_rate": 8.797617070168919e-05,
      "loss": 0.7907,
      "step": 5649
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8338594436645508,
      "learning_rate": 8.79463319744677e-05,
      "loss": 0.8555,
      "step": 5650
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8909322023391724,
      "learning_rate": 8.791649433622688e-05,
      "loss": 0.8734,
      "step": 5651
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8399606347084045,
      "learning_rate": 8.788665778966238e-05,
      "loss": 0.9424,
      "step": 5652
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8470072150230408,
      "learning_rate": 8.785682233746975e-05,
      "loss": 0.7431,
      "step": 5653
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8998028635978699,
      "learning_rate": 8.782698798234446e-05,
      "loss": 0.6582,
      "step": 5654
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8875757455825806,
      "learning_rate": 8.779715472698192e-05,
      "loss": 0.8404,
      "step": 5655
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.0471546649932861,
      "learning_rate": 8.776732257407731e-05,
      "loss": 0.7382,
      "step": 5656
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8708950281143188,
      "learning_rate": 8.773749152632588e-05,
      "loss": 0.7478,
      "step": 5657
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.7622387409210205,
      "learning_rate": 8.770766158642264e-05,
      "loss": 0.701,
      "step": 5658
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8494791984558105,
      "learning_rate": 8.767783275706257e-05,
      "loss": 0.6878,
      "step": 5659
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8960793018341064,
      "learning_rate": 8.764800504094058e-05,
      "loss": 0.8773,
      "step": 5660
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.765622615814209,
      "learning_rate": 8.761817844075137e-05,
      "loss": 0.6615,
      "step": 5661
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9101631045341492,
      "learning_rate": 8.758835295918963e-05,
      "loss": 0.746,
      "step": 5662
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.816622257232666,
      "learning_rate": 8.755852859894993e-05,
      "loss": 0.8962,
      "step": 5663
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9631803631782532,
      "learning_rate": 8.752870536272673e-05,
      "loss": 0.8178,
      "step": 5664
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8514527678489685,
      "learning_rate": 8.749888325321439e-05,
      "loss": 0.8543,
      "step": 5665
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.857399046421051,
      "learning_rate": 8.746906227310716e-05,
      "loss": 0.9169,
      "step": 5666
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8268188238143921,
      "learning_rate": 8.743924242509922e-05,
      "loss": 0.7523,
      "step": 5667
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.7412222623825073,
      "learning_rate": 8.740942371188459e-05,
      "loss": 0.6822,
      "step": 5668
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8526499271392822,
      "learning_rate": 8.737960613615724e-05,
      "loss": 0.6934,
      "step": 5669
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.7676637768745422,
      "learning_rate": 8.734978970061104e-05,
      "loss": 0.9051,
      "step": 5670
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.954738438129425,
      "learning_rate": 8.731997440793968e-05,
      "loss": 0.8721,
      "step": 5671
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8336813449859619,
      "learning_rate": 8.72901602608369e-05,
      "loss": 0.7827,
      "step": 5672
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9843933582305908,
      "learning_rate": 8.72603472619961e-05,
      "loss": 0.9216,
      "step": 5673
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9803519248962402,
      "learning_rate": 8.723053541411082e-05,
      "loss": 0.8238,
      "step": 5674
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.884221076965332,
      "learning_rate": 8.720072471987442e-05,
      "loss": 0.7043,
      "step": 5675
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.850408136844635,
      "learning_rate": 8.717091518198e-05,
      "loss": 0.8706,
      "step": 5676
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8717281222343445,
      "learning_rate": 8.71411068031208e-05,
      "loss": 0.7843,
      "step": 5677
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.0415698289871216,
      "learning_rate": 8.711129958598976e-05,
      "loss": 0.8823,
      "step": 5678
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.7701334357261658,
      "learning_rate": 8.708149353327982e-05,
      "loss": 0.6824,
      "step": 5679
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8792080283164978,
      "learning_rate": 8.705168864768384e-05,
      "loss": 0.8244,
      "step": 5680
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.895536482334137,
      "learning_rate": 8.702188493189447e-05,
      "loss": 0.873,
      "step": 5681
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9660058617591858,
      "learning_rate": 8.69920823886043e-05,
      "loss": 0.9203,
      "step": 5682
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8742905855178833,
      "learning_rate": 8.696228102050583e-05,
      "loss": 1.0514,
      "step": 5683
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8971613049507141,
      "learning_rate": 8.693248083029142e-05,
      "loss": 0.7705,
      "step": 5684
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.81290203332901,
      "learning_rate": 8.690268182065346e-05,
      "loss": 0.7759,
      "step": 5685
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9770933389663696,
      "learning_rate": 8.687288399428397e-05,
      "loss": 0.7779,
      "step": 5686
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8299341201782227,
      "learning_rate": 8.684308735387513e-05,
      "loss": 0.8691,
      "step": 5687
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8290414810180664,
      "learning_rate": 8.681329190211882e-05,
      "loss": 0.7148,
      "step": 5688
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9326844215393066,
      "learning_rate": 8.678349764170693e-05,
      "loss": 0.8495,
      "step": 5689
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9444479942321777,
      "learning_rate": 8.675370457533121e-05,
      "loss": 0.7214,
      "step": 5690
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8702495098114014,
      "learning_rate": 8.672391270568328e-05,
      "loss": 0.7651,
      "step": 5691
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1098989248275757,
      "learning_rate": 8.669412203545468e-05,
      "loss": 0.9304,
      "step": 5692
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0180435180664062,
      "learning_rate": 8.666433256733678e-05,
      "loss": 0.9392,
      "step": 5693
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0562876462936401,
      "learning_rate": 8.663454430402093e-05,
      "loss": 0.7962,
      "step": 5694
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9325870275497437,
      "learning_rate": 8.660475724819836e-05,
      "loss": 0.8123,
      "step": 5695
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1128060817718506,
      "learning_rate": 8.65749714025601e-05,
      "loss": 0.9257,
      "step": 5696
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8302357792854309,
      "learning_rate": 8.65451867697972e-05,
      "loss": 0.7225,
      "step": 5697
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.985501229763031,
      "learning_rate": 8.651540335260045e-05,
      "loss": 0.6855,
      "step": 5698
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9869139194488525,
      "learning_rate": 8.648562115366066e-05,
      "loss": 0.7976,
      "step": 5699
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9399576187133789,
      "learning_rate": 8.645584017566851e-05,
      "loss": 0.6966,
      "step": 5700
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8851685523986816,
      "learning_rate": 8.642606042131449e-05,
      "loss": 0.8309,
      "step": 5701
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8776306509971619,
      "learning_rate": 8.639628189328907e-05,
      "loss": 0.7605,
      "step": 5702
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9954470992088318,
      "learning_rate": 8.636650459428254e-05,
      "loss": 0.811,
      "step": 5703
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9385342001914978,
      "learning_rate": 8.633672852698514e-05,
      "loss": 0.8488,
      "step": 5704
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8304972052574158,
      "learning_rate": 8.630695369408697e-05,
      "loss": 0.8209,
      "step": 5705
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8644033670425415,
      "learning_rate": 8.6277180098278e-05,
      "loss": 0.8915,
      "step": 5706
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0068514347076416,
      "learning_rate": 8.62474077422481e-05,
      "loss": 0.8138,
      "step": 5707
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.023825764656067,
      "learning_rate": 8.621763662868703e-05,
      "loss": 0.9294,
      "step": 5708
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.938262939453125,
      "learning_rate": 8.618786676028446e-05,
      "loss": 0.8874,
      "step": 5709
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9379661679267883,
      "learning_rate": 8.615809813972996e-05,
      "loss": 0.9787,
      "step": 5710
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0042370557785034,
      "learning_rate": 8.612833076971286e-05,
      "loss": 0.8372,
      "step": 5711
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9768930077552795,
      "learning_rate": 8.609856465292259e-05,
      "loss": 0.9507,
      "step": 5712
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8598830699920654,
      "learning_rate": 8.606879979204825e-05,
      "loss": 0.8188,
      "step": 5713
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8076317310333252,
      "learning_rate": 8.603903618977898e-05,
      "loss": 0.7945,
      "step": 5714
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8813288807868958,
      "learning_rate": 8.600927384880378e-05,
      "loss": 0.8432,
      "step": 5715
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8992553353309631,
      "learning_rate": 8.597951277181142e-05,
      "loss": 0.8134,
      "step": 5716
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0259615182876587,
      "learning_rate": 8.594975296149076e-05,
      "loss": 0.7824,
      "step": 5717
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.868723452091217,
      "learning_rate": 8.591999442053032e-05,
      "loss": 0.8704,
      "step": 5718
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8588517904281616,
      "learning_rate": 8.589023715161865e-05,
      "loss": 0.8156,
      "step": 5719
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9276012182235718,
      "learning_rate": 8.586048115744422e-05,
      "loss": 0.8223,
      "step": 5720
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9439911842346191,
      "learning_rate": 8.583072644069522e-05,
      "loss": 0.8857,
      "step": 5721
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9122838377952576,
      "learning_rate": 8.580097300405988e-05,
      "loss": 0.8467,
      "step": 5722
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0123558044433594,
      "learning_rate": 8.577122085022621e-05,
      "loss": 1.0072,
      "step": 5723
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.921371340751648,
      "learning_rate": 8.574146998188217e-05,
      "loss": 0.8298,
      "step": 5724
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.879334568977356,
      "learning_rate": 8.571172040171562e-05,
      "loss": 0.8259,
      "step": 5725
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0288796424865723,
      "learning_rate": 8.56819721124142e-05,
      "loss": 0.8428,
      "step": 5726
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9684762954711914,
      "learning_rate": 8.565222511666557e-05,
      "loss": 0.9385,
      "step": 5727
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9912325739860535,
      "learning_rate": 8.562247941715712e-05,
      "loss": 0.7712,
      "step": 5728
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9355105757713318,
      "learning_rate": 8.559273501657625e-05,
      "loss": 1.0681,
      "step": 5729
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.7969098091125488,
      "learning_rate": 8.556299191761024e-05,
      "loss": 0.7973,
      "step": 5730
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8423822522163391,
      "learning_rate": 8.553325012294613e-05,
      "loss": 0.7859,
      "step": 5731
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8393206000328064,
      "learning_rate": 8.550350963527098e-05,
      "loss": 0.7169,
      "step": 5732
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8980963230133057,
      "learning_rate": 8.547377045727164e-05,
      "loss": 0.8089,
      "step": 5733
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9337926506996155,
      "learning_rate": 8.54440325916349e-05,
      "loss": 0.7806,
      "step": 5734
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8664975166320801,
      "learning_rate": 8.541429604104741e-05,
      "loss": 0.7795,
      "step": 5735
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9418028593063354,
      "learning_rate": 8.538456080819567e-05,
      "loss": 0.9126,
      "step": 5736
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9349313378334045,
      "learning_rate": 8.535482689576613e-05,
      "loss": 0.8214,
      "step": 5737
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9468107223510742,
      "learning_rate": 8.532509430644505e-05,
      "loss": 0.8085,
      "step": 5738
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9069304466247559,
      "learning_rate": 8.529536304291859e-05,
      "loss": 0.7516,
      "step": 5739
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9387087821960449,
      "learning_rate": 8.526563310787284e-05,
      "loss": 0.84,
      "step": 5740
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9777936339378357,
      "learning_rate": 8.52359045039937e-05,
      "loss": 0.8842,
      "step": 5741
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9157954454421997,
      "learning_rate": 8.520617723396702e-05,
      "loss": 0.7169,
      "step": 5742
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9201861023902893,
      "learning_rate": 8.517645130047845e-05,
      "loss": 0.8396,
      "step": 5743
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.8513765335083008,
      "learning_rate": 8.514672670621357e-05,
      "loss": 0.8651,
      "step": 5744
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9212194681167603,
      "learning_rate": 8.511700345385785e-05,
      "loss": 0.7591,
      "step": 5745
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9415310025215149,
      "learning_rate": 8.508728154609657e-05,
      "loss": 0.7863,
      "step": 5746
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0088706016540527,
      "learning_rate": 8.505756098561501e-05,
      "loss": 0.8768,
      "step": 5747
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.8022462129592896,
      "learning_rate": 8.502784177509818e-05,
      "loss": 0.831,
      "step": 5748
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9014273881912231,
      "learning_rate": 8.499812391723105e-05,
      "loss": 0.7601,
      "step": 5749
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9626069664955139,
      "learning_rate": 8.496840741469855e-05,
      "loss": 0.7964,
      "step": 5750
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.979386031627655,
      "learning_rate": 8.493869227018527e-05,
      "loss": 0.8051,
      "step": 5751
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.8098431825637817,
      "learning_rate": 8.490897848637592e-05,
      "loss": 0.7284,
      "step": 5752
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.8875352740287781,
      "learning_rate": 8.487926606595488e-05,
      "loss": 0.8896,
      "step": 5753
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0015044212341309,
      "learning_rate": 8.484955501160652e-05,
      "loss": 0.8197,
      "step": 5754
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9683628678321838,
      "learning_rate": 8.481984532601515e-05,
      "loss": 0.8576,
      "step": 5755
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0050989389419556,
      "learning_rate": 8.479013701186475e-05,
      "loss": 0.8112,
      "step": 5756
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.894388735294342,
      "learning_rate": 8.476043007183938e-05,
      "loss": 0.8936,
      "step": 5757
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9694968461990356,
      "learning_rate": 8.473072450862285e-05,
      "loss": 0.9196,
      "step": 5758
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9153710007667542,
      "learning_rate": 8.470102032489888e-05,
      "loss": 0.668,
      "step": 5759
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.8825878500938416,
      "learning_rate": 8.467131752335116e-05,
      "loss": 0.854,
      "step": 5760
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9378197193145752,
      "learning_rate": 8.464161610666307e-05,
      "loss": 0.7902,
      "step": 5761
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.049019455909729,
      "learning_rate": 8.461191607751804e-05,
      "loss": 0.8789,
      "step": 5762
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.958395779132843,
      "learning_rate": 8.458221743859924e-05,
      "loss": 0.7378,
      "step": 5763
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0207874774932861,
      "learning_rate": 8.45525201925898e-05,
      "loss": 0.8735,
      "step": 5764
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9180783033370972,
      "learning_rate": 8.452282434217272e-05,
      "loss": 0.8038,
      "step": 5765
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9972503185272217,
      "learning_rate": 8.449312989003078e-05,
      "loss": 0.9059,
      "step": 5766
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.864944577217102,
      "learning_rate": 8.446343683884681e-05,
      "loss": 0.6363,
      "step": 5767
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8935273885726929,
      "learning_rate": 8.443374519130331e-05,
      "loss": 1.0042,
      "step": 5768
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.7741546630859375,
      "learning_rate": 8.440405495008281e-05,
      "loss": 0.81,
      "step": 5769
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9032292366027832,
      "learning_rate": 8.437436611786766e-05,
      "loss": 0.8566,
      "step": 5770
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9659150242805481,
      "learning_rate": 8.434467869734005e-05,
      "loss": 0.9859,
      "step": 5771
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9433410167694092,
      "learning_rate": 8.431499269118207e-05,
      "loss": 0.7365,
      "step": 5772
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8648310303688049,
      "learning_rate": 8.428530810207568e-05,
      "loss": 0.8615,
      "step": 5773
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9430176019668579,
      "learning_rate": 8.425562493270275e-05,
      "loss": 0.8104,
      "step": 5774
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9665706157684326,
      "learning_rate": 8.422594318574497e-05,
      "loss": 0.8714,
      "step": 5775
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9146357178688049,
      "learning_rate": 8.419626286388388e-05,
      "loss": 0.688,
      "step": 5776
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9882055521011353,
      "learning_rate": 8.416658396980099e-05,
      "loss": 0.7365,
      "step": 5777
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9143183827400208,
      "learning_rate": 8.413690650617757e-05,
      "loss": 0.832,
      "step": 5778
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0234034061431885,
      "learning_rate": 8.410723047569483e-05,
      "loss": 0.995,
      "step": 5779
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8566288948059082,
      "learning_rate": 8.407755588103386e-05,
      "loss": 0.7319,
      "step": 5780
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.841599702835083,
      "learning_rate": 8.404788272487553e-05,
      "loss": 0.8235,
      "step": 5781
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8171674609184265,
      "learning_rate": 8.401821100990072e-05,
      "loss": 0.7784,
      "step": 5782
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1559754610061646,
      "learning_rate": 8.398854073878999e-05,
      "loss": 0.9221,
      "step": 5783
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0499789714813232,
      "learning_rate": 8.395887191422397e-05,
      "loss": 0.8645,
      "step": 5784
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3504812717437744,
      "learning_rate": 8.392920453888309e-05,
      "loss": 0.906,
      "step": 5785
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8913654685020447,
      "learning_rate": 8.389953861544752e-05,
      "loss": 0.6512,
      "step": 5786
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8633145689964294,
      "learning_rate": 8.386987414659756e-05,
      "loss": 0.841,
      "step": 5787
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.025991678237915,
      "learning_rate": 8.384021113501306e-05,
      "loss": 0.8721,
      "step": 5788
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8738961815834045,
      "learning_rate": 8.381054958337402e-05,
      "loss": 0.8539,
      "step": 5789
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8748875856399536,
      "learning_rate": 8.37808894943602e-05,
      "loss": 0.6967,
      "step": 5790
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9729429483413696,
      "learning_rate": 8.37512308706511e-05,
      "loss": 0.8826,
      "step": 5791
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8592857718467712,
      "learning_rate": 8.372157371492638e-05,
      "loss": 0.7537,
      "step": 5792
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8431957960128784,
      "learning_rate": 8.369191802986524e-05,
      "loss": 0.8575,
      "step": 5793
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9647923111915588,
      "learning_rate": 8.366226381814697e-05,
      "loss": 0.8773,
      "step": 5794
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8477747440338135,
      "learning_rate": 8.363261108245072e-05,
      "loss": 0.762,
      "step": 5795
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8961308598518372,
      "learning_rate": 8.360295982545534e-05,
      "loss": 0.8828,
      "step": 5796
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8899791240692139,
      "learning_rate": 8.357331004983973e-05,
      "loss": 0.8546,
      "step": 5797
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9492402672767639,
      "learning_rate": 8.354366175828251e-05,
      "loss": 0.9796,
      "step": 5798
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.0668731927871704,
      "learning_rate": 8.351401495346228e-05,
      "loss": 0.7995,
      "step": 5799
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8782174587249756,
      "learning_rate": 8.348436963805748e-05,
      "loss": 0.7835,
      "step": 5800
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8887962698936462,
      "learning_rate": 8.345472581474636e-05,
      "loss": 0.7315,
      "step": 5801
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9190962910652161,
      "learning_rate": 8.342508348620709e-05,
      "loss": 0.9353,
      "step": 5802
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8968420624732971,
      "learning_rate": 8.339544265511766e-05,
      "loss": 0.8095,
      "step": 5803
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9461013078689575,
      "learning_rate": 8.336580332415597e-05,
      "loss": 0.9363,
      "step": 5804
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8945879340171814,
      "learning_rate": 8.33361654959998e-05,
      "loss": 0.8483,
      "step": 5805
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9249940514564514,
      "learning_rate": 8.33065291733267e-05,
      "loss": 0.8835,
      "step": 5806
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9324250221252441,
      "learning_rate": 8.32768943588142e-05,
      "loss": 0.7102,
      "step": 5807
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9545156955718994,
      "learning_rate": 8.324726105513958e-05,
      "loss": 0.7636,
      "step": 5808
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8187214136123657,
      "learning_rate": 8.321762926498006e-05,
      "loss": 0.7617,
      "step": 5809
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9916725158691406,
      "learning_rate": 8.318799899101274e-05,
      "loss": 0.793,
      "step": 5810
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.0003262758255005,
      "learning_rate": 8.315837023591452e-05,
      "loss": 0.8795,
      "step": 5811
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.0816155672073364,
      "learning_rate": 8.31287430023622e-05,
      "loss": 0.746,
      "step": 5812
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8276634216308594,
      "learning_rate": 8.309911729303241e-05,
      "loss": 0.8372,
      "step": 5813
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.907431960105896,
      "learning_rate": 8.306949311060167e-05,
      "loss": 0.8159,
      "step": 5814
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9165525436401367,
      "learning_rate": 8.30398704577464e-05,
      "loss": 0.7821,
      "step": 5815
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8705980181694031,
      "learning_rate": 8.301024933714276e-05,
      "loss": 0.857,
      "step": 5816
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.0065878629684448,
      "learning_rate": 8.298062975146694e-05,
      "loss": 1.0165,
      "step": 5817
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1356894969940186,
      "learning_rate": 8.295101170339483e-05,
      "loss": 0.8244,
      "step": 5818
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.0193732976913452,
      "learning_rate": 8.292139519560229e-05,
      "loss": 0.904,
      "step": 5819
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8636108636856079,
      "learning_rate": 8.289178023076501e-05,
      "loss": 0.8576,
      "step": 5820
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9237553477287292,
      "learning_rate": 8.28621668115585e-05,
      "loss": 0.8175,
      "step": 5821
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9538276195526123,
      "learning_rate": 8.283255494065823e-05,
      "loss": 0.7589,
      "step": 5822
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8187211155891418,
      "learning_rate": 8.280294462073935e-05,
      "loss": 0.6035,
      "step": 5823
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.835979163646698,
      "learning_rate": 8.277333585447706e-05,
      "loss": 0.8628,
      "step": 5824
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8352502584457397,
      "learning_rate": 8.274372864454641e-05,
      "loss": 0.8145,
      "step": 5825
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9201905727386475,
      "learning_rate": 8.271412299362209e-05,
      "loss": 0.6937,
      "step": 5826
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9547684192657471,
      "learning_rate": 8.268451890437896e-05,
      "loss": 0.7486,
      "step": 5827
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.962713360786438,
      "learning_rate": 8.265491637949145e-05,
      "loss": 0.8643,
      "step": 5828
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.858282208442688,
      "learning_rate": 8.262531542163403e-05,
      "loss": 0.8332,
      "step": 5829
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9314063787460327,
      "learning_rate": 8.259571603348105e-05,
      "loss": 0.7792,
      "step": 5830
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8805924654006958,
      "learning_rate": 8.256611821770653e-05,
      "loss": 0.6915,
      "step": 5831
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9726827144622803,
      "learning_rate": 8.253652197698455e-05,
      "loss": 0.8928,
      "step": 5832
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.0675846338272095,
      "learning_rate": 8.25069273139889e-05,
      "loss": 0.9119,
      "step": 5833
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9909031987190247,
      "learning_rate": 8.247733423139331e-05,
      "loss": 0.955,
      "step": 5834
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9767759442329407,
      "learning_rate": 8.244774273187141e-05,
      "loss": 1.1025,
      "step": 5835
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.0358740091323853,
      "learning_rate": 8.241815281809655e-05,
      "loss": 0.7835,
      "step": 5836
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8870111107826233,
      "learning_rate": 8.238856449274204e-05,
      "loss": 0.8062,
      "step": 5837
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8954660296440125,
      "learning_rate": 8.2358977758481e-05,
      "loss": 0.8221,
      "step": 5838
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8845124244689941,
      "learning_rate": 8.232939261798644e-05,
      "loss": 0.8322,
      "step": 5839
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8227827548980713,
      "learning_rate": 8.229980907393122e-05,
      "loss": 0.8015,
      "step": 5840
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9672027826309204,
      "learning_rate": 8.227022712898802e-05,
      "loss": 0.9339,
      "step": 5841
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9234362244606018,
      "learning_rate": 8.224064678582943e-05,
      "loss": 0.8332,
      "step": 5842
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.871410608291626,
      "learning_rate": 8.221106804712782e-05,
      "loss": 0.7074,
      "step": 5843
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8701865673065186,
      "learning_rate": 8.21814909155555e-05,
      "loss": 0.793,
      "step": 5844
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9916341304779053,
      "learning_rate": 8.21519153937846e-05,
      "loss": 0.9548,
      "step": 5845
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.0512160062789917,
      "learning_rate": 8.212234148448707e-05,
      "loss": 0.9336,
      "step": 5846
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8245749473571777,
      "learning_rate": 8.20927691903348e-05,
      "loss": 0.7318,
      "step": 5847
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9027788639068604,
      "learning_rate": 8.206319851399939e-05,
      "loss": 0.8975,
      "step": 5848
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.0249130725860596,
      "learning_rate": 8.203362945815248e-05,
      "loss": 0.7925,
      "step": 5849
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.0170999765396118,
      "learning_rate": 8.200406202546541e-05,
      "loss": 0.712,
      "step": 5850
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8836989402770996,
      "learning_rate": 8.197449621860943e-05,
      "loss": 0.765,
      "step": 5851
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9385777115821838,
      "learning_rate": 8.194493204025568e-05,
      "loss": 0.7421,
      "step": 5852
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9420427680015564,
      "learning_rate": 8.191536949307508e-05,
      "loss": 0.7595,
      "step": 5853
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.023455023765564,
      "learning_rate": 8.188580857973844e-05,
      "loss": 0.7513,
      "step": 5854
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.0047972202301025,
      "learning_rate": 8.185624930291646e-05,
      "loss": 0.8433,
      "step": 5855
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9844880104064941,
      "learning_rate": 8.18266916652796e-05,
      "loss": 1.0306,
      "step": 5856
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9397380948066711,
      "learning_rate": 8.179713566949829e-05,
      "loss": 0.7583,
      "step": 5857
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.134047269821167,
      "learning_rate": 8.176758131824265e-05,
      "loss": 0.8072,
      "step": 5858
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9175476431846619,
      "learning_rate": 8.173802861418282e-05,
      "loss": 0.852,
      "step": 5859
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9957419633865356,
      "learning_rate": 8.170847755998876e-05,
      "loss": 0.7181,
      "step": 5860
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9451197981834412,
      "learning_rate": 8.167892815833011e-05,
      "loss": 0.8333,
      "step": 5861
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9104927778244019,
      "learning_rate": 8.164938041187665e-05,
      "loss": 0.8253,
      "step": 5862
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.7333707809448242,
      "learning_rate": 8.161983432329771e-05,
      "loss": 0.6399,
      "step": 5863
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.916006863117218,
      "learning_rate": 8.159028989526264e-05,
      "loss": 0.8331,
      "step": 5864
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9831569194793701,
      "learning_rate": 8.156074713044074e-05,
      "loss": 0.7078,
      "step": 5865
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9709320068359375,
      "learning_rate": 8.153120603150084e-05,
      "loss": 0.8273,
      "step": 5866
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9606887698173523,
      "learning_rate": 8.150166660111199e-05,
      "loss": 0.8458,
      "step": 5867
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.0189917087554932,
      "learning_rate": 8.147212884194278e-05,
      "loss": 1.0005,
      "step": 5868
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8855880498886108,
      "learning_rate": 8.144259275666179e-05,
      "loss": 0.8847,
      "step": 5869
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9294763803482056,
      "learning_rate": 8.141305834793753e-05,
      "loss": 0.8239,
      "step": 5870
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9067355990409851,
      "learning_rate": 8.138352561843818e-05,
      "loss": 0.9331,
      "step": 5871
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8585739731788635,
      "learning_rate": 8.13539945708319e-05,
      "loss": 0.7268,
      "step": 5872
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9075150489807129,
      "learning_rate": 8.132446520778662e-05,
      "loss": 0.7771,
      "step": 5873
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8397543430328369,
      "learning_rate": 8.129493753197017e-05,
      "loss": 0.7479,
      "step": 5874
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9425328969955444,
      "learning_rate": 8.126541154605022e-05,
      "loss": 0.7846,
      "step": 5875
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9679790735244751,
      "learning_rate": 8.123588725269427e-05,
      "loss": 0.9231,
      "step": 5876
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9444148540496826,
      "learning_rate": 8.120636465456965e-05,
      "loss": 0.8576,
      "step": 5877
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9487553238868713,
      "learning_rate": 8.117684375434357e-05,
      "loss": 0.9,
      "step": 5878
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9532830119132996,
      "learning_rate": 8.114732455468307e-05,
      "loss": 0.7992,
      "step": 5879
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8672592043876648,
      "learning_rate": 8.111780705825508e-05,
      "loss": 0.7887,
      "step": 5880
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.836396336555481,
      "learning_rate": 8.10882912677263e-05,
      "loss": 0.7639,
      "step": 5881
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9721764326095581,
      "learning_rate": 8.105877718576332e-05,
      "loss": 0.8115,
      "step": 5882
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9433866739273071,
      "learning_rate": 8.102926481503257e-05,
      "loss": 0.81,
      "step": 5883
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8358859419822693,
      "learning_rate": 8.099975415820034e-05,
      "loss": 0.6957,
      "step": 5884
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9138021469116211,
      "learning_rate": 8.097024521793276e-05,
      "loss": 0.8474,
      "step": 5885
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.0327115058898926,
      "learning_rate": 8.094073799689575e-05,
      "loss": 0.7758,
      "step": 5886
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9728330373764038,
      "learning_rate": 8.091123249775519e-05,
      "loss": 0.7119,
      "step": 5887
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9047760367393494,
      "learning_rate": 8.088172872317665e-05,
      "loss": 0.7763,
      "step": 5888
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8901060223579407,
      "learning_rate": 8.085222667582568e-05,
      "loss": 0.6684,
      "step": 5889
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8689963817596436,
      "learning_rate": 8.082272635836762e-05,
      "loss": 0.72,
      "step": 5890
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9310135841369629,
      "learning_rate": 8.079322777346766e-05,
      "loss": 0.7216,
      "step": 5891
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9189056158065796,
      "learning_rate": 8.076373092379087e-05,
      "loss": 0.6409,
      "step": 5892
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9765612483024597,
      "learning_rate": 8.073423581200198e-05,
      "loss": 0.8479,
      "step": 5893
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8745375871658325,
      "learning_rate": 8.070474244076584e-05,
      "loss": 0.7851,
      "step": 5894
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.0411179065704346,
      "learning_rate": 8.067525081274698e-05,
      "loss": 0.8547,
      "step": 5895
    },
    {
      "epoch": 2.21,
      "eval_loss": 0.9851978421211243,
      "eval_runtime": 17.6363,
      "eval_samples_per_second": 154.738,
      "eval_steps_per_second": 38.727,
      "step": 5895
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.846912145614624,
      "learning_rate": 8.064576093060978e-05,
      "loss": 0.7711,
      "step": 5896
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.7630794048309326,
      "learning_rate": 8.061627279701854e-05,
      "loss": 0.6304,
      "step": 5897
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9502009749412537,
      "learning_rate": 8.058678641463724e-05,
      "loss": 0.929,
      "step": 5898
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9916005730628967,
      "learning_rate": 8.05573017861299e-05,
      "loss": 0.8879,
      "step": 5899
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9297499656677246,
      "learning_rate": 8.052781891416028e-05,
      "loss": 0.7929,
      "step": 5900
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9243143200874329,
      "learning_rate": 8.049833780139192e-05,
      "loss": 0.7962,
      "step": 5901
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.7968193292617798,
      "learning_rate": 8.046885845048838e-05,
      "loss": 0.6354,
      "step": 5902
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9880951046943665,
      "learning_rate": 8.043938086411285e-05,
      "loss": 0.758,
      "step": 5903
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.8543862700462341,
      "learning_rate": 8.04099050449285e-05,
      "loss": 0.8533,
      "step": 5904
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9380577802658081,
      "learning_rate": 8.038043099559837e-05,
      "loss": 0.8214,
      "step": 5905
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.8662536144256592,
      "learning_rate": 8.035095871878516e-05,
      "loss": 0.7062,
      "step": 5906
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.920100748538971,
      "learning_rate": 8.032148821715161e-05,
      "loss": 0.8069,
      "step": 5907
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.8904909491539001,
      "learning_rate": 8.029201949336014e-05,
      "loss": 0.8413,
      "step": 5908
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.053934931755066,
      "learning_rate": 8.026255255007315e-05,
      "loss": 0.888,
      "step": 5909
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.8953412771224976,
      "learning_rate": 8.02330873899528e-05,
      "loss": 0.7457,
      "step": 5910
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9340875148773193,
      "learning_rate": 8.020362401566103e-05,
      "loss": 0.7853,
      "step": 5911
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9482661485671997,
      "learning_rate": 8.01741624298598e-05,
      "loss": 0.8173,
      "step": 5912
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.929085373878479,
      "learning_rate": 8.01447026352107e-05,
      "loss": 0.8231,
      "step": 5913
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9424161911010742,
      "learning_rate": 8.011524463437531e-05,
      "loss": 0.9281,
      "step": 5914
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9352458119392395,
      "learning_rate": 8.008578843001499e-05,
      "loss": 0.8635,
      "step": 5915
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9547236561775208,
      "learning_rate": 8.005633402479091e-05,
      "loss": 0.8775,
      "step": 5916
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9422927498817444,
      "learning_rate": 8.002688142136417e-05,
      "loss": 0.8793,
      "step": 5917
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1358392238616943,
      "learning_rate": 7.999743062239557e-05,
      "loss": 0.6435,
      "step": 5918
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.8697951436042786,
      "learning_rate": 7.996798163054587e-05,
      "loss": 0.8809,
      "step": 5919
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.8541561365127563,
      "learning_rate": 7.993853444847561e-05,
      "loss": 0.7999,
      "step": 5920
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9206909537315369,
      "learning_rate": 7.990908907884517e-05,
      "loss": 0.9123,
      "step": 5921
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9707465767860413,
      "learning_rate": 7.98796455243148e-05,
      "loss": 0.7083,
      "step": 5922
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9484550356864929,
      "learning_rate": 7.98502037875445e-05,
      "loss": 0.8058,
      "step": 5923
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.8958969712257385,
      "learning_rate": 7.982076387119421e-05,
      "loss": 0.6587,
      "step": 5924
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.06850004196167,
      "learning_rate": 7.979132577792366e-05,
      "loss": 0.9313,
      "step": 5925
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.043208360671997,
      "learning_rate": 7.97618895103924e-05,
      "loss": 0.8027,
      "step": 5926
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8443194627761841,
      "learning_rate": 7.973245507125985e-05,
      "loss": 0.7943,
      "step": 5927
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9484256505966187,
      "learning_rate": 7.970302246318521e-05,
      "loss": 0.8429,
      "step": 5928
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9223771095275879,
      "learning_rate": 7.967359168882755e-05,
      "loss": 0.7763,
      "step": 5929
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0735764503479004,
      "learning_rate": 7.964416275084584e-05,
      "loss": 0.9217,
      "step": 5930
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9296501874923706,
      "learning_rate": 7.961473565189875e-05,
      "loss": 0.7216,
      "step": 5931
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.09675133228302,
      "learning_rate": 7.95853103946449e-05,
      "loss": 0.942,
      "step": 5932
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8760204911231995,
      "learning_rate": 7.955588698174262e-05,
      "loss": 0.833,
      "step": 5933
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0340945720672607,
      "learning_rate": 7.95264654158502e-05,
      "loss": 0.8102,
      "step": 5934
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9970924854278564,
      "learning_rate": 7.949704569962575e-05,
      "loss": 0.8074,
      "step": 5935
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9114053249359131,
      "learning_rate": 7.946762783572708e-05,
      "loss": 0.6665,
      "step": 5936
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9341669082641602,
      "learning_rate": 7.943821182681205e-05,
      "loss": 0.8213,
      "step": 5937
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0555737018585205,
      "learning_rate": 7.94087976755381e-05,
      "loss": 0.8724,
      "step": 5938
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0389922857284546,
      "learning_rate": 7.937938538456268e-05,
      "loss": 0.7858,
      "step": 5939
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0340293645858765,
      "learning_rate": 7.934997495654311e-05,
      "loss": 0.9236,
      "step": 5940
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0396037101745605,
      "learning_rate": 7.932056639413632e-05,
      "loss": 0.9321,
      "step": 5941
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8729332089424133,
      "learning_rate": 7.929115969999933e-05,
      "loss": 0.9121,
      "step": 5942
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9289164543151855,
      "learning_rate": 7.926175487678876e-05,
      "loss": 0.7855,
      "step": 5943
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0263704061508179,
      "learning_rate": 7.92323519271612e-05,
      "loss": 0.9414,
      "step": 5944
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8594925999641418,
      "learning_rate": 7.920295085377311e-05,
      "loss": 0.7034,
      "step": 5945
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9330661296844482,
      "learning_rate": 7.917355165928064e-05,
      "loss": 0.7885,
      "step": 5946
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8596360087394714,
      "learning_rate": 7.914415434633985e-05,
      "loss": 0.7892,
      "step": 5947
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9123426675796509,
      "learning_rate": 7.911475891760663e-05,
      "loss": 0.7658,
      "step": 5948
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8126939535140991,
      "learning_rate": 7.908536537573668e-05,
      "loss": 0.7057,
      "step": 5949
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9995397925376892,
      "learning_rate": 7.905597372338558e-05,
      "loss": 0.9061,
      "step": 5950
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8321855068206787,
      "learning_rate": 7.902658396320864e-05,
      "loss": 0.7429,
      "step": 5951
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8316742777824402,
      "learning_rate": 7.899719609786111e-05,
      "loss": 0.7331,
      "step": 5952
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.0233700275421143,
      "learning_rate": 7.896781012999798e-05,
      "loss": 0.8818,
      "step": 5953
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8726819753646851,
      "learning_rate": 7.893842606227411e-05,
      "loss": 0.737,
      "step": 5954
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9957520365715027,
      "learning_rate": 7.890904389734423e-05,
      "loss": 0.8524,
      "step": 5955
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1021480560302734,
      "learning_rate": 7.887966363786279e-05,
      "loss": 0.8759,
      "step": 5956
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.0580075979232788,
      "learning_rate": 7.885028528648417e-05,
      "loss": 0.7908,
      "step": 5957
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9442081451416016,
      "learning_rate": 7.882090884586251e-05,
      "loss": 0.7812,
      "step": 5958
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8994975090026855,
      "learning_rate": 7.879153431865182e-05,
      "loss": 0.8549,
      "step": 5959
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9029251337051392,
      "learning_rate": 7.876216170750595e-05,
      "loss": 0.7865,
      "step": 5960
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8510790467262268,
      "learning_rate": 7.873279101507849e-05,
      "loss": 0.7632,
      "step": 5961
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9053831696510315,
      "learning_rate": 7.870342224402296e-05,
      "loss": 0.8503,
      "step": 5962
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.838259220123291,
      "learning_rate": 7.867405539699266e-05,
      "loss": 0.8749,
      "step": 5963
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9268372654914856,
      "learning_rate": 7.864469047664067e-05,
      "loss": 0.8889,
      "step": 5964
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.0869096517562866,
      "learning_rate": 7.861532748562002e-05,
      "loss": 0.8642,
      "step": 5965
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9542982578277588,
      "learning_rate": 7.858596642658343e-05,
      "loss": 0.8358,
      "step": 5966
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9124200344085693,
      "learning_rate": 7.855660730218358e-05,
      "loss": 0.9071,
      "step": 5967
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.0706727504730225,
      "learning_rate": 7.852725011507278e-05,
      "loss": 0.8898,
      "step": 5968
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.048890233039856,
      "learning_rate": 7.849789486790336e-05,
      "loss": 0.9103,
      "step": 5969
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9156789779663086,
      "learning_rate": 7.846854156332744e-05,
      "loss": 0.731,
      "step": 5970
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8907057642936707,
      "learning_rate": 7.843919020399684e-05,
      "loss": 0.8293,
      "step": 5971
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.887760579586029,
      "learning_rate": 7.840984079256337e-05,
      "loss": 0.8317,
      "step": 5972
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9877274632453918,
      "learning_rate": 7.838049333167848e-05,
      "loss": 0.8994,
      "step": 5973
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8505722880363464,
      "learning_rate": 7.835114782399364e-05,
      "loss": 0.7855,
      "step": 5974
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.943242073059082,
      "learning_rate": 7.832180427216004e-05,
      "loss": 0.8798,
      "step": 5975
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9101446866989136,
      "learning_rate": 7.829246267882864e-05,
      "loss": 0.8875,
      "step": 5976
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9752018451690674,
      "learning_rate": 7.826312304665039e-05,
      "loss": 0.8363,
      "step": 5977
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8358737826347351,
      "learning_rate": 7.823378537827585e-05,
      "loss": 0.9858,
      "step": 5978
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9272959232330322,
      "learning_rate": 7.820444967635554e-05,
      "loss": 0.7891,
      "step": 5979
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.876106321811676,
      "learning_rate": 7.817511594353986e-05,
      "loss": 0.8222,
      "step": 5980
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9235019683837891,
      "learning_rate": 7.814578418247884e-05,
      "loss": 0.8567,
      "step": 5981
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9984934329986572,
      "learning_rate": 7.81164543958225e-05,
      "loss": 0.8099,
      "step": 5982
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8538726568222046,
      "learning_rate": 7.808712658622058e-05,
      "loss": 0.8877,
      "step": 5983
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8751855492591858,
      "learning_rate": 7.80578007563227e-05,
      "loss": 0.9108,
      "step": 5984
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.013213038444519,
      "learning_rate": 7.802847690877832e-05,
      "loss": 0.8479,
      "step": 5985
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9892585873603821,
      "learning_rate": 7.799915504623662e-05,
      "loss": 0.7561,
      "step": 5986
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8966041207313538,
      "learning_rate": 7.796983517134671e-05,
      "loss": 0.8457,
      "step": 5987
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9851561784744263,
      "learning_rate": 7.794051728675744e-05,
      "loss": 0.9167,
      "step": 5988
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9364020228385925,
      "learning_rate": 7.791120139511752e-05,
      "loss": 0.8239,
      "step": 5989
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9969381093978882,
      "learning_rate": 7.788188749907553e-05,
      "loss": 0.802,
      "step": 5990
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.869059681892395,
      "learning_rate": 7.785257560127975e-05,
      "loss": 0.6809,
      "step": 5991
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8653062582015991,
      "learning_rate": 7.782326570437837e-05,
      "loss": 0.8977,
      "step": 5992
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9574717283248901,
      "learning_rate": 7.779395781101936e-05,
      "loss": 0.8889,
      "step": 5993
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9621526002883911,
      "learning_rate": 7.776465192385051e-05,
      "loss": 0.7451,
      "step": 5994
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9248753190040588,
      "learning_rate": 7.773534804551952e-05,
      "loss": 0.7704,
      "step": 5995
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8614605069160461,
      "learning_rate": 7.770604617867374e-05,
      "loss": 0.8527,
      "step": 5996
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9433735013008118,
      "learning_rate": 7.767674632596047e-05,
      "loss": 0.7901,
      "step": 5997
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1110789775848389,
      "learning_rate": 7.764744849002676e-05,
      "loss": 0.8102,
      "step": 5998
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8785794377326965,
      "learning_rate": 7.761815267351951e-05,
      "loss": 0.7737,
      "step": 5999
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8256083130836487,
      "learning_rate": 7.758885887908546e-05,
      "loss": 0.674,
      "step": 6000
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9222911596298218,
      "learning_rate": 7.755956710937111e-05,
      "loss": 0.8507,
      "step": 6001
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.0109745264053345,
      "learning_rate": 7.753027736702283e-05,
      "loss": 0.9074,
      "step": 6002
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9208554029464722,
      "learning_rate": 7.750098965468675e-05,
      "loss": 0.8569,
      "step": 6003
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9506418108940125,
      "learning_rate": 7.747170397500887e-05,
      "loss": 0.8253,
      "step": 6004
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.880961537361145,
      "learning_rate": 7.744242033063498e-05,
      "loss": 0.8595,
      "step": 6005
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9308170676231384,
      "learning_rate": 7.741313872421069e-05,
      "loss": 0.6725,
      "step": 6006
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.7863349318504333,
      "learning_rate": 7.738385915838148e-05,
      "loss": 0.8121,
      "step": 6007
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9947253465652466,
      "learning_rate": 7.735458163579246e-05,
      "loss": 0.7986,
      "step": 6008
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9812553524971008,
      "learning_rate": 7.73253061590888e-05,
      "loss": 0.9588,
      "step": 6009
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8369360566139221,
      "learning_rate": 7.729603273091539e-05,
      "loss": 0.71,
      "step": 6010
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1110618114471436,
      "learning_rate": 7.72667613539168e-05,
      "loss": 1.0222,
      "step": 6011
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9721248149871826,
      "learning_rate": 7.723749203073768e-05,
      "loss": 0.836,
      "step": 6012
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9773927927017212,
      "learning_rate": 7.720822476402222e-05,
      "loss": 0.8335,
      "step": 6013
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8430916666984558,
      "learning_rate": 7.71789595564146e-05,
      "loss": 0.7719,
      "step": 6014
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8690534830093384,
      "learning_rate": 7.714969641055883e-05,
      "loss": 0.8635,
      "step": 6015
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.0574071407318115,
      "learning_rate": 7.712043532909855e-05,
      "loss": 0.8388,
      "step": 6016
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9497845768928528,
      "learning_rate": 7.709117631467746e-05,
      "loss": 0.865,
      "step": 6017
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9590959548950195,
      "learning_rate": 7.706191936993883e-05,
      "loss": 0.8476,
      "step": 6018
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9530377388000488,
      "learning_rate": 7.703266449752588e-05,
      "loss": 0.8631,
      "step": 6019
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8704661726951599,
      "learning_rate": 7.700341170008174e-05,
      "loss": 0.8122,
      "step": 6020
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9610298871994019,
      "learning_rate": 7.697416098024908e-05,
      "loss": 0.8681,
      "step": 6021
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1411447525024414,
      "learning_rate": 7.694491234067064e-05,
      "loss": 0.9467,
      "step": 6022
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8619931936264038,
      "learning_rate": 7.69156657839888e-05,
      "loss": 0.8491,
      "step": 6023
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8454144597053528,
      "learning_rate": 7.688642131284585e-05,
      "loss": 0.7364,
      "step": 6024
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8475390076637268,
      "learning_rate": 7.685717892988392e-05,
      "loss": 0.8304,
      "step": 6025
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9204264283180237,
      "learning_rate": 7.682793863774477e-05,
      "loss": 0.9058,
      "step": 6026
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9191638231277466,
      "learning_rate": 7.679870043907022e-05,
      "loss": 0.759,
      "step": 6027
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.0018426179885864,
      "learning_rate": 7.67694643365017e-05,
      "loss": 0.8978,
      "step": 6028
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9855759739875793,
      "learning_rate": 7.674023033268053e-05,
      "loss": 0.7762,
      "step": 6029
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9066625833511353,
      "learning_rate": 7.671099843024788e-05,
      "loss": 0.7192,
      "step": 6030
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9252528548240662,
      "learning_rate": 7.668176863184463e-05,
      "loss": 0.8426,
      "step": 6031
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.051020622253418,
      "learning_rate": 7.665254094011159e-05,
      "loss": 1.0451,
      "step": 6032
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.8911116719245911,
      "learning_rate": 7.662331535768925e-05,
      "loss": 0.7787,
      "step": 6033
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.8512126207351685,
      "learning_rate": 7.659409188721803e-05,
      "loss": 1.0123,
      "step": 6034
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.0443404912948608,
      "learning_rate": 7.656487053133809e-05,
      "loss": 0.8443,
      "step": 6035
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.0287353992462158,
      "learning_rate": 7.653565129268937e-05,
      "loss": 0.8305,
      "step": 6036
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9822664260864258,
      "learning_rate": 7.650643417391175e-05,
      "loss": 0.8467,
      "step": 6037
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9871828556060791,
      "learning_rate": 7.647721917764475e-05,
      "loss": 0.9288,
      "step": 6038
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.8477451205253601,
      "learning_rate": 7.644800630652782e-05,
      "loss": 0.8031,
      "step": 6039
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9076310396194458,
      "learning_rate": 7.641879556320018e-05,
      "loss": 0.7784,
      "step": 6040
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.0930908918380737,
      "learning_rate": 7.638958695030083e-05,
      "loss": 0.7368,
      "step": 6041
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.925722062587738,
      "learning_rate": 7.636038047046866e-05,
      "loss": 0.8651,
      "step": 6042
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.955289900302887,
      "learning_rate": 7.63311761263422e-05,
      "loss": 0.946,
      "step": 6043
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9356592893600464,
      "learning_rate": 7.630197392055998e-05,
      "loss": 0.8366,
      "step": 6044
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.8755605816841125,
      "learning_rate": 7.627277385576027e-05,
      "loss": 0.7718,
      "step": 6045
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9093108177185059,
      "learning_rate": 7.624357593458107e-05,
      "loss": 0.6729,
      "step": 6046
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9349961876869202,
      "learning_rate": 7.621438015966033e-05,
      "loss": 0.666,
      "step": 6047
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.891871452331543,
      "learning_rate": 7.618518653363559e-05,
      "loss": 0.9378,
      "step": 6048
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.037710189819336,
      "learning_rate": 7.615599505914443e-05,
      "loss": 1.0047,
      "step": 6049
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9317094087600708,
      "learning_rate": 7.612680573882417e-05,
      "loss": 0.9075,
      "step": 6050
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.0549111366271973,
      "learning_rate": 7.609761857531177e-05,
      "loss": 0.9851,
      "step": 6051
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.819863498210907,
      "learning_rate": 7.606843357124426e-05,
      "loss": 0.7983,
      "step": 6052
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9261680245399475,
      "learning_rate": 7.603925072925823e-05,
      "loss": 0.7862,
      "step": 6053
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.155320167541504,
      "learning_rate": 7.601007005199021e-05,
      "loss": 0.6894,
      "step": 6054
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9081116318702698,
      "learning_rate": 7.598089154207659e-05,
      "loss": 0.7276,
      "step": 6055
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9812401533126831,
      "learning_rate": 7.595171520215339e-05,
      "loss": 0.6783,
      "step": 6056
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8422990441322327,
      "learning_rate": 7.592254103485659e-05,
      "loss": 0.6947,
      "step": 6057
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9618236422538757,
      "learning_rate": 7.589336904282184e-05,
      "loss": 0.7969,
      "step": 6058
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1555843353271484,
      "learning_rate": 7.586419922868471e-05,
      "loss": 0.9942,
      "step": 6059
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8606289029121399,
      "learning_rate": 7.583503159508055e-05,
      "loss": 0.8605,
      "step": 6060
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9539014101028442,
      "learning_rate": 7.580586614464445e-05,
      "loss": 0.8388,
      "step": 6061
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9737402200698853,
      "learning_rate": 7.577670288001137e-05,
      "loss": 0.9167,
      "step": 6062
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9719823598861694,
      "learning_rate": 7.574754180381601e-05,
      "loss": 0.9387,
      "step": 6063
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9214046001434326,
      "learning_rate": 7.571838291869294e-05,
      "loss": 0.78,
      "step": 6064
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9994566440582275,
      "learning_rate": 7.56892262272765e-05,
      "loss": 0.8266,
      "step": 6065
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8975732326507568,
      "learning_rate": 7.56600717322008e-05,
      "loss": 0.8277,
      "step": 6066
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8384100198745728,
      "learning_rate": 7.563091943609984e-05,
      "loss": 0.6732,
      "step": 6067
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.018539547920227,
      "learning_rate": 7.56017693416073e-05,
      "loss": 0.8992,
      "step": 6068
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.864441990852356,
      "learning_rate": 7.557262145135677e-05,
      "loss": 0.8602,
      "step": 6069
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1134114265441895,
      "learning_rate": 7.55434757679816e-05,
      "loss": 0.6978,
      "step": 6070
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9138038754463196,
      "learning_rate": 7.551433229411488e-05,
      "loss": 0.7928,
      "step": 6071
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8474975228309631,
      "learning_rate": 7.548519103238965e-05,
      "loss": 0.6306,
      "step": 6072
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9360517263412476,
      "learning_rate": 7.545605198543858e-05,
      "loss": 0.7843,
      "step": 6073
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.052502155303955,
      "learning_rate": 7.542691515589424e-05,
      "loss": 0.8009,
      "step": 6074
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8930253386497498,
      "learning_rate": 7.539778054638901e-05,
      "loss": 0.7445,
      "step": 6075
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9840304851531982,
      "learning_rate": 7.536864815955497e-05,
      "loss": 0.6513,
      "step": 6076
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8417231440544128,
      "learning_rate": 7.533951799802416e-05,
      "loss": 0.8765,
      "step": 6077
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9773772954940796,
      "learning_rate": 7.531039006442822e-05,
      "loss": 0.8144,
      "step": 6078
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8340346813201904,
      "learning_rate": 7.528126436139878e-05,
      "loss": 0.8377,
      "step": 6079
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9493737816810608,
      "learning_rate": 7.525214089156713e-05,
      "loss": 0.8816,
      "step": 6080
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9480802416801453,
      "learning_rate": 7.522301965756443e-05,
      "loss": 0.8511,
      "step": 6081
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8603459000587463,
      "learning_rate": 7.519390066202166e-05,
      "loss": 0.8228,
      "step": 6082
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.8750146627426147,
      "learning_rate": 7.516478390756945e-05,
      "loss": 0.721,
      "step": 6083
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.0141137838363647,
      "learning_rate": 7.51356693968384e-05,
      "loss": 0.679,
      "step": 6084
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9597448706626892,
      "learning_rate": 7.51065571324589e-05,
      "loss": 0.8363,
      "step": 6085
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9455132484436035,
      "learning_rate": 7.507744711706095e-05,
      "loss": 0.8185,
      "step": 6086
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.0338767766952515,
      "learning_rate": 7.50483393532746e-05,
      "loss": 0.9558,
      "step": 6087
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9663766026496887,
      "learning_rate": 7.501923384372948e-05,
      "loss": 0.7432,
      "step": 6088
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9102634191513062,
      "learning_rate": 7.499013059105511e-05,
      "loss": 0.7822,
      "step": 6089
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9623690843582153,
      "learning_rate": 7.49610295978809e-05,
      "loss": 0.8291,
      "step": 6090
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.846402108669281,
      "learning_rate": 7.493193086683585e-05,
      "loss": 0.6874,
      "step": 6091
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.0376150608062744,
      "learning_rate": 7.490283440054895e-05,
      "loss": 0.9145,
      "step": 6092
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.0824878215789795,
      "learning_rate": 7.487374020164883e-05,
      "loss": 0.7842,
      "step": 6093
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.0418875217437744,
      "learning_rate": 7.484464827276399e-05,
      "loss": 0.8283,
      "step": 6094
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9080113172531128,
      "learning_rate": 7.481555861652282e-05,
      "loss": 0.7545,
      "step": 6095
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.0601459741592407,
      "learning_rate": 7.478647123555329e-05,
      "loss": 0.9723,
      "step": 6096
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.8856122493743896,
      "learning_rate": 7.475738613248336e-05,
      "loss": 0.7313,
      "step": 6097
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9679635167121887,
      "learning_rate": 7.472830330994063e-05,
      "loss": 0.6892,
      "step": 6098
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.8435798287391663,
      "learning_rate": 7.469922277055261e-05,
      "loss": 0.6811,
      "step": 6099
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.0226726531982422,
      "learning_rate": 7.467014451694659e-05,
      "loss": 0.8369,
      "step": 6100
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9256264567375183,
      "learning_rate": 7.464106855174957e-05,
      "loss": 0.8316,
      "step": 6101
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.8727557063102722,
      "learning_rate": 7.461199487758845e-05,
      "loss": 0.8111,
      "step": 6102
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.0694643259048462,
      "learning_rate": 7.458292349708982e-05,
      "loss": 0.8187,
      "step": 6103
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.8906984925270081,
      "learning_rate": 7.455385441288014e-05,
      "loss": 0.8122,
      "step": 6104
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.979741096496582,
      "learning_rate": 7.452478762758566e-05,
      "loss": 0.8765,
      "step": 6105
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.277288556098938,
      "learning_rate": 7.449572314383237e-05,
      "loss": 0.9184,
      "step": 6106
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9827014207839966,
      "learning_rate": 7.446666096424609e-05,
      "loss": 0.9565,
      "step": 6107
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9648321866989136,
      "learning_rate": 7.443760109145242e-05,
      "loss": 0.9719,
      "step": 6108
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9090179800987244,
      "learning_rate": 7.440854352807674e-05,
      "loss": 0.8128,
      "step": 6109
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9141748547554016,
      "learning_rate": 7.437948827674427e-05,
      "loss": 0.7347,
      "step": 6110
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9909433722496033,
      "learning_rate": 7.435043534007998e-05,
      "loss": 0.7174,
      "step": 6111
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.0806602239608765,
      "learning_rate": 7.432138472070863e-05,
      "loss": 0.9631,
      "step": 6112
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9663843512535095,
      "learning_rate": 7.429233642125476e-05,
      "loss": 0.8731,
      "step": 6113
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9923156499862671,
      "learning_rate": 7.426329044434273e-05,
      "loss": 0.8832,
      "step": 6114
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.8782480955123901,
      "learning_rate": 7.423424679259673e-05,
      "loss": 0.7994,
      "step": 6115
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.8621048927307129,
      "learning_rate": 7.420520546864061e-05,
      "loss": 0.7628,
      "step": 6116
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9524788856506348,
      "learning_rate": 7.417616647509817e-05,
      "loss": 0.6741,
      "step": 6117
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9254405498504639,
      "learning_rate": 7.414712981459283e-05,
      "loss": 0.7433,
      "step": 6118
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.8311590552330017,
      "learning_rate": 7.411809548974792e-05,
      "loss": 0.7736,
      "step": 6119
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.8999873399734497,
      "learning_rate": 7.408906350318661e-05,
      "loss": 0.7722,
      "step": 6120
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9228675961494446,
      "learning_rate": 7.406003385753163e-05,
      "loss": 0.9079,
      "step": 6121
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9783831834793091,
      "learning_rate": 7.403100655540578e-05,
      "loss": 0.8174,
      "step": 6122
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9533256888389587,
      "learning_rate": 7.400198159943137e-05,
      "loss": 0.9856,
      "step": 6123
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.8786800503730774,
      "learning_rate": 7.397295899223078e-05,
      "loss": 0.7757,
      "step": 6124
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9071011543273926,
      "learning_rate": 7.3943938736426e-05,
      "loss": 0.8467,
      "step": 6125
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.799500048160553,
      "learning_rate": 7.391492083463878e-05,
      "loss": 0.8191,
      "step": 6126
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.0251702070236206,
      "learning_rate": 7.388590528949082e-05,
      "loss": 0.8052,
      "step": 6127
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.0316787958145142,
      "learning_rate": 7.385689210360341e-05,
      "loss": 0.895,
      "step": 6128
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9390791654586792,
      "learning_rate": 7.382788127959777e-05,
      "loss": 0.9537,
      "step": 6129
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9090555310249329,
      "learning_rate": 7.379887282009493e-05,
      "loss": 0.9402,
      "step": 6130
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.94757479429245,
      "learning_rate": 7.376986672771554e-05,
      "loss": 0.8821,
      "step": 6131
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9730098247528076,
      "learning_rate": 7.374086300508018e-05,
      "loss": 0.8498,
      "step": 6132
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.8473596572875977,
      "learning_rate": 7.371186165480916e-05,
      "loss": 0.6427,
      "step": 6133
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.8505036234855652,
      "learning_rate": 7.368286267952262e-05,
      "loss": 0.857,
      "step": 6134
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9205848574638367,
      "learning_rate": 7.365386608184042e-05,
      "loss": 0.8458,
      "step": 6135
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.019980788230896,
      "learning_rate": 7.362487186438224e-05,
      "loss": 0.65,
      "step": 6136
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9564799070358276,
      "learning_rate": 7.359588002976757e-05,
      "loss": 0.8687,
      "step": 6137
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8387205600738525,
      "learning_rate": 7.356689058061561e-05,
      "loss": 0.7064,
      "step": 6138
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8889574408531189,
      "learning_rate": 7.353790351954543e-05,
      "loss": 0.7578,
      "step": 6139
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9078226089477539,
      "learning_rate": 7.350891884917585e-05,
      "loss": 0.867,
      "step": 6140
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8757694959640503,
      "learning_rate": 7.347993657212547e-05,
      "loss": 0.7596,
      "step": 6141
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1252766847610474,
      "learning_rate": 7.345095669101265e-05,
      "loss": 0.7715,
      "step": 6142
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8675466179847717,
      "learning_rate": 7.342197920845557e-05,
      "loss": 0.6649,
      "step": 6143
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9147344827651978,
      "learning_rate": 7.339300412707217e-05,
      "loss": 0.7686,
      "step": 6144
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8982990980148315,
      "learning_rate": 7.336403144948022e-05,
      "loss": 0.8823,
      "step": 6145
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9413564205169678,
      "learning_rate": 7.333506117829723e-05,
      "loss": 0.8101,
      "step": 6146
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.0050358772277832,
      "learning_rate": 7.330609331614047e-05,
      "loss": 0.9291,
      "step": 6147
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8738914132118225,
      "learning_rate": 7.327712786562703e-05,
      "loss": 0.801,
      "step": 6148
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9076246023178101,
      "learning_rate": 7.324816482937378e-05,
      "loss": 0.8289,
      "step": 6149
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8470820188522339,
      "learning_rate": 7.32192042099974e-05,
      "loss": 0.815,
      "step": 6150
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9740384817123413,
      "learning_rate": 7.319024601011427e-05,
      "loss": 0.7987,
      "step": 6151
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8639681935310364,
      "learning_rate": 7.31612902323406e-05,
      "loss": 0.714,
      "step": 6152
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8703543543815613,
      "learning_rate": 7.313233687929242e-05,
      "loss": 0.9243,
      "step": 6153
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9259352684020996,
      "learning_rate": 7.310338595358545e-05,
      "loss": 0.691,
      "step": 6154
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8001659512519836,
      "learning_rate": 7.307443745783531e-05,
      "loss": 0.8958,
      "step": 6155
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.925732433795929,
      "learning_rate": 7.304549139465727e-05,
      "loss": 0.85,
      "step": 6156
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.963634729385376,
      "learning_rate": 7.301654776666649e-05,
      "loss": 0.7251,
      "step": 6157
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8718309998512268,
      "learning_rate": 7.298760657647779e-05,
      "loss": 0.7499,
      "step": 6158
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.0094935894012451,
      "learning_rate": 7.295866782670591e-05,
      "loss": 0.8979,
      "step": 6159
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9438912272453308,
      "learning_rate": 7.292973151996534e-05,
      "loss": 0.92,
      "step": 6160
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.966501772403717,
      "learning_rate": 7.290079765887017e-05,
      "loss": 0.7664,
      "step": 6161
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9071841239929199,
      "learning_rate": 7.287186624603457e-05,
      "loss": 0.7872,
      "step": 6162
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9368190765380859,
      "learning_rate": 7.284293728407222e-05,
      "loss": 0.7777,
      "step": 6163
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9625749588012695,
      "learning_rate": 7.28140107755967e-05,
      "loss": 0.7089,
      "step": 6164
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9289863705635071,
      "learning_rate": 7.278508672322145e-05,
      "loss": 0.8647,
      "step": 6165
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9649459719657898,
      "learning_rate": 7.275616512955945e-05,
      "loss": 0.8963,
      "step": 6166
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9218540787696838,
      "learning_rate": 7.272724599722374e-05,
      "loss": 0.7968,
      "step": 6167
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.866667628288269,
      "learning_rate": 7.269832932882688e-05,
      "loss": 0.8718,
      "step": 6168
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.7932524085044861,
      "learning_rate": 7.266941512698138e-05,
      "loss": 0.6353,
      "step": 6169
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.100878119468689,
      "learning_rate": 7.264050339429953e-05,
      "loss": 0.9142,
      "step": 6170
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9313275218009949,
      "learning_rate": 7.261159413339327e-05,
      "loss": 0.7655,
      "step": 6171
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.8792120218276978,
      "learning_rate": 7.258268734687441e-05,
      "loss": 0.7902,
      "step": 6172
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.0168558359146118,
      "learning_rate": 7.255378303735449e-05,
      "loss": 0.799,
      "step": 6173
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.0159275531768799,
      "learning_rate": 7.252488120744488e-05,
      "loss": 0.8926,
      "step": 6174
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9590809941291809,
      "learning_rate": 7.24959818597567e-05,
      "loss": 0.9289,
      "step": 6175
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9937403202056885,
      "learning_rate": 7.246708499690083e-05,
      "loss": 0.826,
      "step": 6176
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.8797158002853394,
      "learning_rate": 7.243819062148793e-05,
      "loss": 0.7315,
      "step": 6177
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.0171879529953003,
      "learning_rate": 7.240929873612846e-05,
      "loss": 0.829,
      "step": 6178
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.8982298374176025,
      "learning_rate": 7.238040934343263e-05,
      "loss": 0.8955,
      "step": 6179
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.8887124061584473,
      "learning_rate": 7.235152244601043e-05,
      "loss": 0.8379,
      "step": 6180
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.8284779191017151,
      "learning_rate": 7.232263804647163e-05,
      "loss": 0.6919,
      "step": 6181
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9982441663742065,
      "learning_rate": 7.22937561474258e-05,
      "loss": 0.8215,
      "step": 6182
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9693446159362793,
      "learning_rate": 7.226487675148219e-05,
      "loss": 0.7419,
      "step": 6183
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.8807844519615173,
      "learning_rate": 7.223599986124994e-05,
      "loss": 0.8292,
      "step": 6184
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.174902319908142,
      "learning_rate": 7.220712547933792e-05,
      "loss": 0.845,
      "step": 6185
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.024556040763855,
      "learning_rate": 7.217825360835473e-05,
      "loss": 0.8049,
      "step": 6186
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.8520997762680054,
      "learning_rate": 7.214938425090883e-05,
      "loss": 0.7491,
      "step": 6187
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.026688575744629,
      "learning_rate": 7.212051740960834e-05,
      "loss": 0.8712,
      "step": 6188
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.001364827156067,
      "learning_rate": 7.209165308706128e-05,
      "loss": 0.892,
      "step": 6189
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.914002001285553,
      "learning_rate": 7.206279128587533e-05,
      "loss": 0.9089,
      "step": 6190
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.8959076404571533,
      "learning_rate": 7.2033932008658e-05,
      "loss": 0.8051,
      "step": 6191
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.94645756483078,
      "learning_rate": 7.20050752580166e-05,
      "loss": 0.8904,
      "step": 6192
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.051952600479126,
      "learning_rate": 7.19762210365581e-05,
      "loss": 0.9046,
      "step": 6193
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.0830702781677246,
      "learning_rate": 7.194736934688937e-05,
      "loss": 0.751,
      "step": 6194
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9208510518074036,
      "learning_rate": 7.191852019161703e-05,
      "loss": 0.7596,
      "step": 6195
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.8309812545776367,
      "learning_rate": 7.188967357334733e-05,
      "loss": 0.6636,
      "step": 6196
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9025148749351501,
      "learning_rate": 7.186082949468653e-05,
      "loss": 1.0101,
      "step": 6197
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9140446782112122,
      "learning_rate": 7.18319879582404e-05,
      "loss": 0.8467,
      "step": 6198
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.8626208901405334,
      "learning_rate": 7.180314896661468e-05,
      "loss": 0.743,
      "step": 6199
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9373636841773987,
      "learning_rate": 7.177431252241484e-05,
      "loss": 0.8312,
      "step": 6200
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.7314283847808838,
      "learning_rate": 7.174547862824602e-05,
      "loss": 0.6439,
      "step": 6201
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9718741178512573,
      "learning_rate": 7.171664728671326e-05,
      "loss": 1.0423,
      "step": 6202
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.8401091694831848,
      "learning_rate": 7.168781850042124e-05,
      "loss": 0.8433,
      "step": 6203
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.8812562227249146,
      "learning_rate": 7.16589922719745e-05,
      "loss": 0.7756,
      "step": 6204
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9585669636726379,
      "learning_rate": 7.16301686039774e-05,
      "loss": 0.8077,
      "step": 6205
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9120372533798218,
      "learning_rate": 7.160134749903391e-05,
      "loss": 0.7966,
      "step": 6206
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9285467267036438,
      "learning_rate": 7.157252895974788e-05,
      "loss": 0.885,
      "step": 6207
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9309242367744446,
      "learning_rate": 7.154371298872289e-05,
      "loss": 0.89,
      "step": 6208
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9329833984375,
      "learning_rate": 7.151489958856233e-05,
      "loss": 0.7847,
      "step": 6209
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1782058477401733,
      "learning_rate": 7.148608876186931e-05,
      "loss": 0.963,
      "step": 6210
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.139611840248108,
      "learning_rate": 7.145728051124671e-05,
      "loss": 0.8676,
      "step": 6211
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.0557208061218262,
      "learning_rate": 7.142847483929722e-05,
      "loss": 0.7978,
      "step": 6212
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.181353211402893,
      "learning_rate": 7.139967174862325e-05,
      "loss": 0.8221,
      "step": 6213
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9601356983184814,
      "learning_rate": 7.1370871241827e-05,
      "loss": 0.7475,
      "step": 6214
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9548810124397278,
      "learning_rate": 7.134207332151044e-05,
      "loss": 0.7456,
      "step": 6215
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.893725574016571,
      "learning_rate": 7.131327799027529e-05,
      "loss": 0.7539,
      "step": 6216
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9347209334373474,
      "learning_rate": 7.128448525072307e-05,
      "loss": 0.8808,
      "step": 6217
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.972072958946228,
      "learning_rate": 7.1255695105455e-05,
      "loss": 0.8101,
      "step": 6218
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.0365705490112305,
      "learning_rate": 7.122690755707212e-05,
      "loss": 0.8564,
      "step": 6219
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9098747372627258,
      "learning_rate": 7.119812260817526e-05,
      "loss": 0.7796,
      "step": 6220
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.8695220947265625,
      "learning_rate": 7.116934026136492e-05,
      "loss": 0.8706,
      "step": 6221
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.883747398853302,
      "learning_rate": 7.114056051924147e-05,
      "loss": 0.8148,
      "step": 6222
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.0084387063980103,
      "learning_rate": 7.111178338440496e-05,
      "loss": 0.8343,
      "step": 6223
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.022931694984436,
      "learning_rate": 7.108300885945525e-05,
      "loss": 0.7656,
      "step": 6224
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.0235340595245361,
      "learning_rate": 7.105423694699198e-05,
      "loss": 0.9164,
      "step": 6225
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.965325117111206,
      "learning_rate": 7.102546764961449e-05,
      "loss": 0.8846,
      "step": 6226
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9229762554168701,
      "learning_rate": 7.099670096992197e-05,
      "loss": 0.8729,
      "step": 6227
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.8393018245697021,
      "learning_rate": 7.096793691051328e-05,
      "loss": 0.7467,
      "step": 6228
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1935091018676758,
      "learning_rate": 7.093917547398711e-05,
      "loss": 0.8479,
      "step": 6229
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9041988849639893,
      "learning_rate": 7.091041666294191e-05,
      "loss": 0.8091,
      "step": 6230
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2488301992416382,
      "learning_rate": 7.088166047997585e-05,
      "loss": 0.8565,
      "step": 6231
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9754186868667603,
      "learning_rate": 7.085290692768694e-05,
      "loss": 0.679,
      "step": 6232
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.8768190145492554,
      "learning_rate": 7.082415600867278e-05,
      "loss": 0.5734,
      "step": 6233
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.0176365375518799,
      "learning_rate": 7.079540772553097e-05,
      "loss": 0.9716,
      "step": 6234
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9428842067718506,
      "learning_rate": 7.076666208085876e-05,
      "loss": 0.9053,
      "step": 6235
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9657464623451233,
      "learning_rate": 7.073791907725303e-05,
      "loss": 0.9522,
      "step": 6236
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9140335917472839,
      "learning_rate": 7.070917871731072e-05,
      "loss": 0.7942,
      "step": 6237
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.8639563322067261,
      "learning_rate": 7.06804410036282e-05,
      "loss": 0.7278,
      "step": 6238
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9554201364517212,
      "learning_rate": 7.065170593880181e-05,
      "loss": 0.7474,
      "step": 6239
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.923717200756073,
      "learning_rate": 7.062297352542769e-05,
      "loss": 0.819,
      "step": 6240
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9834179878234863,
      "learning_rate": 7.059424376610154e-05,
      "loss": 0.7734,
      "step": 6241
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.185652732849121,
      "learning_rate": 7.0565516663419e-05,
      "loss": 0.82,
      "step": 6242
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9571309685707092,
      "learning_rate": 7.053679221997533e-05,
      "loss": 0.7691,
      "step": 6243
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9570879340171814,
      "learning_rate": 7.050807043836565e-05,
      "loss": 0.8137,
      "step": 6244
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9898768067359924,
      "learning_rate": 7.047935132118489e-05,
      "loss": 0.7749,
      "step": 6245
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9095770716667175,
      "learning_rate": 7.045063487102754e-05,
      "loss": 0.7205,
      "step": 6246
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.8524069786071777,
      "learning_rate": 7.042192109048806e-05,
      "loss": 0.7058,
      "step": 6247
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.0311510562896729,
      "learning_rate": 7.03932099821605e-05,
      "loss": 0.9257,
      "step": 6248
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9623306393623352,
      "learning_rate": 7.03645015486388e-05,
      "loss": 0.912,
      "step": 6249
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.0326757431030273,
      "learning_rate": 7.033579579251658e-05,
      "loss": 0.9746,
      "step": 6250
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.035431146621704,
      "learning_rate": 7.030709271638726e-05,
      "loss": 0.8395,
      "step": 6251
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.8906907439231873,
      "learning_rate": 7.0278392322844e-05,
      "loss": 0.9186,
      "step": 6252
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.022970199584961,
      "learning_rate": 7.024969461447972e-05,
      "loss": 1.0297,
      "step": 6253
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.8895883560180664,
      "learning_rate": 7.022099959388706e-05,
      "loss": 0.8014,
      "step": 6254
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.0652027130126953,
      "learning_rate": 7.019230726365851e-05,
      "loss": 0.71,
      "step": 6255
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.8727202415466309,
      "learning_rate": 7.016361762638619e-05,
      "loss": 0.8772,
      "step": 6256
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.8981159925460815,
      "learning_rate": 7.013493068466211e-05,
      "loss": 0.7551,
      "step": 6257
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.0146751403808594,
      "learning_rate": 7.010624644107792e-05,
      "loss": 0.9903,
      "step": 6258
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.0026201009750366,
      "learning_rate": 7.007756489822512e-05,
      "loss": 0.7714,
      "step": 6259
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9356503486633301,
      "learning_rate": 7.004888605869494e-05,
      "loss": 0.7073,
      "step": 6260
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.8520841002464294,
      "learning_rate": 7.00202099250783e-05,
      "loss": 0.7475,
      "step": 6261
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.0225168466567993,
      "learning_rate": 6.999153649996595e-05,
      "loss": 0.8935,
      "step": 6262
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9664372205734253,
      "learning_rate": 6.996286578594835e-05,
      "loss": 0.8549,
      "step": 6263
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.165625810623169,
      "learning_rate": 6.993419778561576e-05,
      "loss": 0.8893,
      "step": 6264
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.8771733641624451,
      "learning_rate": 6.99055325015582e-05,
      "loss": 0.8416,
      "step": 6265
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9531654119491577,
      "learning_rate": 6.987686993636536e-05,
      "loss": 0.8766,
      "step": 6266
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.977652370929718,
      "learning_rate": 6.98482100926268e-05,
      "loss": 0.7669,
      "step": 6267
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8846288323402405,
      "learning_rate": 6.981955297293168e-05,
      "loss": 0.8237,
      "step": 6268
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9223921298980713,
      "learning_rate": 6.979089857986909e-05,
      "loss": 0.8811,
      "step": 6269
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8725709319114685,
      "learning_rate": 6.97622469160278e-05,
      "loss": 0.7829,
      "step": 6270
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9503806233406067,
      "learning_rate": 6.973359798399624e-05,
      "loss": 0.9937,
      "step": 6271
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.17890465259552,
      "learning_rate": 6.97049517863628e-05,
      "loss": 0.7642,
      "step": 6272
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9978543519973755,
      "learning_rate": 6.967630832571538e-05,
      "loss": 0.8972,
      "step": 6273
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9675953984260559,
      "learning_rate": 6.964766760464181e-05,
      "loss": 0.8366,
      "step": 6274
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8403088450431824,
      "learning_rate": 6.961902962572968e-05,
      "loss": 0.8428,
      "step": 6275
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8614692091941833,
      "learning_rate": 6.959039439156613e-05,
      "loss": 0.7333,
      "step": 6276
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9656548500061035,
      "learning_rate": 6.956176190473835e-05,
      "loss": 0.783,
      "step": 6277
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8785561323165894,
      "learning_rate": 6.953313216783297e-05,
      "loss": 0.7772,
      "step": 6278
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.1046223640441895,
      "learning_rate": 6.950450518343659e-05,
      "loss": 0.7358,
      "step": 6279
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8370841145515442,
      "learning_rate": 6.947588095413558e-05,
      "loss": 0.8043,
      "step": 6280
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9674923419952393,
      "learning_rate": 6.944725948251584e-05,
      "loss": 0.8088,
      "step": 6281
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9207295775413513,
      "learning_rate": 6.941864077116324e-05,
      "loss": 0.7477,
      "step": 6282
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9472736120223999,
      "learning_rate": 6.939002482266326e-05,
      "loss": 0.6687,
      "step": 6283
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.0192410945892334,
      "learning_rate": 6.936141163960125e-05,
      "loss": 0.8621,
      "step": 6284
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9424362778663635,
      "learning_rate": 6.933280122456222e-05,
      "loss": 0.9222,
      "step": 6285
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2017186880111694,
      "learning_rate": 6.930419358013096e-05,
      "loss": 0.8287,
      "step": 6286
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9377169013023376,
      "learning_rate": 6.927558870889202e-05,
      "loss": 0.9123,
      "step": 6287
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.0218063592910767,
      "learning_rate": 6.924698661342967e-05,
      "loss": 0.7705,
      "step": 6288
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.914851725101471,
      "learning_rate": 6.921838729632794e-05,
      "loss": 0.88,
      "step": 6289
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8921465277671814,
      "learning_rate": 6.918979076017067e-05,
      "loss": 0.7619,
      "step": 6290
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.0365058183670044,
      "learning_rate": 6.916119700754134e-05,
      "loss": 0.8236,
      "step": 6291
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8989388346672058,
      "learning_rate": 6.913260604102325e-05,
      "loss": 0.8934,
      "step": 6292
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9946595430374146,
      "learning_rate": 6.910401786319944e-05,
      "loss": 0.8658,
      "step": 6293
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.0774812698364258,
      "learning_rate": 6.907543247665266e-05,
      "loss": 1.0516,
      "step": 6294
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.1482843160629272,
      "learning_rate": 6.904684988396549e-05,
      "loss": 0.8861,
      "step": 6295
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8799174427986145,
      "learning_rate": 6.901827008772016e-05,
      "loss": 0.5139,
      "step": 6296
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.0195943117141724,
      "learning_rate": 6.898969309049872e-05,
      "loss": 0.7631,
      "step": 6297
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8641614317893982,
      "learning_rate": 6.896111889488293e-05,
      "loss": 0.6998,
      "step": 6298
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8745054006576538,
      "learning_rate": 6.893254750345429e-05,
      "loss": 0.7707,
      "step": 6299
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8836456537246704,
      "learning_rate": 6.89039789187941e-05,
      "loss": 0.846,
      "step": 6300
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.0164562463760376,
      "learning_rate": 6.887541314348333e-05,
      "loss": 0.8814,
      "step": 6301
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8903396725654602,
      "learning_rate": 6.884685018010279e-05,
      "loss": 0.9822,
      "step": 6302
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8403976559638977,
      "learning_rate": 6.881829003123291e-05,
      "loss": 0.8607,
      "step": 6303
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8526760339736938,
      "learning_rate": 6.878973269945397e-05,
      "loss": 0.603,
      "step": 6304
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9646632075309753,
      "learning_rate": 6.8761178187346e-05,
      "loss": 0.8983,
      "step": 6305
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8441158533096313,
      "learning_rate": 6.873262649748869e-05,
      "loss": 0.7552,
      "step": 6306
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9267090559005737,
      "learning_rate": 6.870407763246156e-05,
      "loss": 0.8428,
      "step": 6307
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9881188273429871,
      "learning_rate": 6.867553159484378e-05,
      "loss": 0.8206,
      "step": 6308
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.0087049007415771,
      "learning_rate": 6.864698838721438e-05,
      "loss": 0.8071,
      "step": 6309
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8286829590797424,
      "learning_rate": 6.861844801215209e-05,
      "loss": 0.6532,
      "step": 6310
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.0032931566238403,
      "learning_rate": 6.858991047223529e-05,
      "loss": 0.8791,
      "step": 6311
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9459896683692932,
      "learning_rate": 6.856137577004229e-05,
      "loss": 0.7575,
      "step": 6312
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8851755857467651,
      "learning_rate": 6.853284390815093e-05,
      "loss": 0.6871,
      "step": 6313
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9469976425170898,
      "learning_rate": 6.850431488913895e-05,
      "loss": 0.8843,
      "step": 6314
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.894963800907135,
      "learning_rate": 6.847578871558386e-05,
      "loss": 0.6569,
      "step": 6315
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9813401699066162,
      "learning_rate": 6.844726539006273e-05,
      "loss": 0.728,
      "step": 6316
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9641947746276855,
      "learning_rate": 6.841874491515253e-05,
      "loss": 0.893,
      "step": 6317
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9524853825569153,
      "learning_rate": 6.83902272934299e-05,
      "loss": 0.7931,
      "step": 6318
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9399118423461914,
      "learning_rate": 6.836171252747126e-05,
      "loss": 0.862,
      "step": 6319
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9244645833969116,
      "learning_rate": 6.833320061985277e-05,
      "loss": 0.7847,
      "step": 6320
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9315648674964905,
      "learning_rate": 6.83046915731503e-05,
      "loss": 0.7555,
      "step": 6321
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9111561179161072,
      "learning_rate": 6.827618538993949e-05,
      "loss": 0.7881,
      "step": 6322
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9781177639961243,
      "learning_rate": 6.82476820727957e-05,
      "loss": 0.9014,
      "step": 6323
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.8731727004051208,
      "learning_rate": 6.821918162429406e-05,
      "loss": 0.6333,
      "step": 6324
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9178487658500671,
      "learning_rate": 6.819068404700943e-05,
      "loss": 0.7831,
      "step": 6325
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9033799767494202,
      "learning_rate": 6.816218934351637e-05,
      "loss": 0.9553,
      "step": 6326
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.068293809890747,
      "learning_rate": 6.813369751638927e-05,
      "loss": 0.8229,
      "step": 6327
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.8811349272727966,
      "learning_rate": 6.810520856820215e-05,
      "loss": 0.712,
      "step": 6328
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9240093231201172,
      "learning_rate": 6.807672250152884e-05,
      "loss": 0.8913,
      "step": 6329
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9028269648551941,
      "learning_rate": 6.804823931894293e-05,
      "loss": 0.8418,
      "step": 6330
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9800744652748108,
      "learning_rate": 6.801975902301766e-05,
      "loss": 0.72,
      "step": 6331
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.1193130016326904,
      "learning_rate": 6.79912816163261e-05,
      "loss": 0.9756,
      "step": 6332
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9098719358444214,
      "learning_rate": 6.796280710144102e-05,
      "loss": 0.8013,
      "step": 6333
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9596406817436218,
      "learning_rate": 6.79343354809349e-05,
      "loss": 0.9296,
      "step": 6334
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9250086545944214,
      "learning_rate": 6.790586675738006e-05,
      "loss": 0.7661,
      "step": 6335
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9604171514511108,
      "learning_rate": 6.78774009333484e-05,
      "loss": 0.8262,
      "step": 6336
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9391502141952515,
      "learning_rate": 6.784893801141173e-05,
      "loss": 0.9353,
      "step": 6337
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.983830988407135,
      "learning_rate": 6.782047799414142e-05,
      "loss": 0.7698,
      "step": 6338
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9738194942474365,
      "learning_rate": 6.779202088410874e-05,
      "loss": 1.0251,
      "step": 6339
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9980114698410034,
      "learning_rate": 6.776356668388464e-05,
      "loss": 0.7539,
      "step": 6340
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.0469671487808228,
      "learning_rate": 6.773511539603974e-05,
      "loss": 0.8284,
      "step": 6341
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9466899037361145,
      "learning_rate": 6.770666702314451e-05,
      "loss": 0.8088,
      "step": 6342
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.8917349576950073,
      "learning_rate": 6.767822156776903e-05,
      "loss": 0.8833,
      "step": 6343
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.0219058990478516,
      "learning_rate": 6.764977903248325e-05,
      "loss": 0.838,
      "step": 6344
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.989513099193573,
      "learning_rate": 6.762133941985677e-05,
      "loss": 0.8165,
      "step": 6345
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.0653127431869507,
      "learning_rate": 6.759290273245892e-05,
      "loss": 0.8184,
      "step": 6346
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8858362436294556,
      "learning_rate": 6.756446897285887e-05,
      "loss": 0.8765,
      "step": 6347
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8181782364845276,
      "learning_rate": 6.753603814362537e-05,
      "loss": 0.7808,
      "step": 6348
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9460709095001221,
      "learning_rate": 6.750761024732697e-05,
      "loss": 0.8992,
      "step": 6349
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9978798627853394,
      "learning_rate": 6.747918528653208e-05,
      "loss": 0.8909,
      "step": 6350
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.0180226564407349,
      "learning_rate": 6.745076326380861e-05,
      "loss": 0.7091,
      "step": 6351
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9715395569801331,
      "learning_rate": 6.742234418172447e-05,
      "loss": 0.9012,
      "step": 6352
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.0103256702423096,
      "learning_rate": 6.739392804284701e-05,
      "loss": 0.7673,
      "step": 6353
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.971220850944519,
      "learning_rate": 6.736551484974352e-05,
      "loss": 0.912,
      "step": 6354
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8784416317939758,
      "learning_rate": 6.733710460498105e-05,
      "loss": 0.8383,
      "step": 6355
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.087942123413086,
      "learning_rate": 6.73086973111262e-05,
      "loss": 0.7772,
      "step": 6356
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.1349889039993286,
      "learning_rate": 6.728029297074548e-05,
      "loss": 0.7805,
      "step": 6357
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9298689365386963,
      "learning_rate": 6.7251891586405e-05,
      "loss": 0.8245,
      "step": 6358
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9284396767616272,
      "learning_rate": 6.72234931606707e-05,
      "loss": 0.8692,
      "step": 6359
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9341880679130554,
      "learning_rate": 6.719509769610824e-05,
      "loss": 0.9235,
      "step": 6360
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9478971362113953,
      "learning_rate": 6.716670519528293e-05,
      "loss": 0.8835,
      "step": 6361
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8818480968475342,
      "learning_rate": 6.713831566075992e-05,
      "loss": 0.7478,
      "step": 6362
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.0238244533538818,
      "learning_rate": 6.710992909510403e-05,
      "loss": 0.9158,
      "step": 6363
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8776931762695312,
      "learning_rate": 6.708154550087981e-05,
      "loss": 0.7703,
      "step": 6364
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.0171411037445068,
      "learning_rate": 6.705316488065158e-05,
      "loss": 0.6749,
      "step": 6365
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8269084095954895,
      "learning_rate": 6.702478723698335e-05,
      "loss": 0.7913,
      "step": 6366
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9567999839782715,
      "learning_rate": 6.699641257243892e-05,
      "loss": 0.7906,
      "step": 6367
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.0641324520111084,
      "learning_rate": 6.69680408895817e-05,
      "loss": 0.874,
      "step": 6368
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9358649849891663,
      "learning_rate": 6.693967219097499e-05,
      "loss": 0.8519,
      "step": 6369
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.0261650085449219,
      "learning_rate": 6.691130647918171e-05,
      "loss": 0.8156,
      "step": 6370
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8204389810562134,
      "learning_rate": 6.688294375676452e-05,
      "loss": 0.7975,
      "step": 6371
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9277483224868774,
      "learning_rate": 6.685458402628588e-05,
      "loss": 0.8408,
      "step": 6372
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9820665121078491,
      "learning_rate": 6.68262272903079e-05,
      "loss": 0.8115,
      "step": 6373
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.0119231939315796,
      "learning_rate": 6.679787355139243e-05,
      "loss": 0.7165,
      "step": 6374
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8702779412269592,
      "learning_rate": 6.676952281210112e-05,
      "loss": 0.6677,
      "step": 6375
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8298000693321228,
      "learning_rate": 6.674117507499526e-05,
      "loss": 0.7829,
      "step": 6376
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9961859583854675,
      "learning_rate": 6.671283034263595e-05,
      "loss": 0.8379,
      "step": 6377
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8349009156227112,
      "learning_rate": 6.668448861758391e-05,
      "loss": 0.7891,
      "step": 6378
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.897449791431427,
      "learning_rate": 6.665614990239969e-05,
      "loss": 0.742,
      "step": 6379
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2056703567504883,
      "learning_rate": 6.662781419964356e-05,
      "loss": 0.9689,
      "step": 6380
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9727599024772644,
      "learning_rate": 6.659948151187545e-05,
      "loss": 0.8687,
      "step": 6381
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9134826064109802,
      "learning_rate": 6.65711518416551e-05,
      "loss": 0.8012,
      "step": 6382
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8090587258338928,
      "learning_rate": 6.654282519154188e-05,
      "loss": 0.8075,
      "step": 6383
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9396087527275085,
      "learning_rate": 6.651450156409494e-05,
      "loss": 0.7411,
      "step": 6384
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8785601258277893,
      "learning_rate": 6.648618096187327e-05,
      "loss": 0.8069,
      "step": 6385
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.0169916152954102,
      "learning_rate": 6.645786338743532e-05,
      "loss": 0.6988,
      "step": 6386
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9737557172775269,
      "learning_rate": 6.642954884333955e-05,
      "loss": 0.7938,
      "step": 6387
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8810718059539795,
      "learning_rate": 6.640123733214394e-05,
      "loss": 0.7119,
      "step": 6388
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8219184279441833,
      "learning_rate": 6.637292885640627e-05,
      "loss": 0.7745,
      "step": 6389
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.0556600093841553,
      "learning_rate": 6.634462341868415e-05,
      "loss": 0.5811,
      "step": 6390
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9419564008712769,
      "learning_rate": 6.63163210215347e-05,
      "loss": 0.7965,
      "step": 6391
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9729017019271851,
      "learning_rate": 6.628802166751495e-05,
      "loss": 0.9151,
      "step": 6392
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9615692496299744,
      "learning_rate": 6.625972535918156e-05,
      "loss": 0.8875,
      "step": 6393
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9165303707122803,
      "learning_rate": 6.623143209909093e-05,
      "loss": 0.7975,
      "step": 6394
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.049666404724121,
      "learning_rate": 6.620314188979923e-05,
      "loss": 0.9224,
      "step": 6395
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.837774395942688,
      "learning_rate": 6.617485473386231e-05,
      "loss": 0.8395,
      "step": 6396
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.095869541168213,
      "learning_rate": 6.614657063383573e-05,
      "loss": 0.8483,
      "step": 6397
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8801032304763794,
      "learning_rate": 6.611828959227484e-05,
      "loss": 0.8255,
      "step": 6398
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8234226703643799,
      "learning_rate": 6.60900116117346e-05,
      "loss": 0.636,
      "step": 6399
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9350499510765076,
      "learning_rate": 6.606173669476987e-05,
      "loss": 0.8332,
      "step": 6400
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8862643837928772,
      "learning_rate": 6.603346484393503e-05,
      "loss": 0.8474,
      "step": 6401
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.1416571140289307,
      "learning_rate": 6.600519606178436e-05,
      "loss": 0.9151,
      "step": 6402
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8747037649154663,
      "learning_rate": 6.59769303508717e-05,
      "loss": 0.7405,
      "step": 6403
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9327594637870789,
      "learning_rate": 6.594866771375077e-05,
      "loss": 0.9668,
      "step": 6404
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8626515865325928,
      "learning_rate": 6.592040815297494e-05,
      "loss": 0.7923,
      "step": 6405
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9083289504051208,
      "learning_rate": 6.589215167109723e-05,
      "loss": 0.9212,
      "step": 6406
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9849241971969604,
      "learning_rate": 6.586389827067055e-05,
      "loss": 0.8201,
      "step": 6407
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8996176719665527,
      "learning_rate": 6.583564795424734e-05,
      "loss": 0.9115,
      "step": 6408
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.931928277015686,
      "learning_rate": 6.58074007243799e-05,
      "loss": 0.9022,
      "step": 6409
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9910814166069031,
      "learning_rate": 6.577915658362022e-05,
      "loss": 1.0597,
      "step": 6410
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9780309200286865,
      "learning_rate": 6.575091553451998e-05,
      "loss": 0.7982,
      "step": 6411
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8526566028594971,
      "learning_rate": 6.572267757963062e-05,
      "loss": 0.8036,
      "step": 6412
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8941213488578796,
      "learning_rate": 6.569444272150324e-05,
      "loss": 0.694,
      "step": 6413
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9897906184196472,
      "learning_rate": 6.566621096268873e-05,
      "loss": 0.8173,
      "step": 6414
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9929688572883606,
      "learning_rate": 6.563798230573768e-05,
      "loss": 0.9246,
      "step": 6415
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.0461108684539795,
      "learning_rate": 6.560975675320033e-05,
      "loss": 0.7577,
      "step": 6416
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8602179288864136,
      "learning_rate": 6.558153430762681e-05,
      "loss": 0.7353,
      "step": 6417
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9322593212127686,
      "learning_rate": 6.555331497156672e-05,
      "loss": 0.7462,
      "step": 6418
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.1136105060577393,
      "learning_rate": 6.55250987475696e-05,
      "loss": 1.0152,
      "step": 6419
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8647451996803284,
      "learning_rate": 6.549688563818465e-05,
      "loss": 0.5787,
      "step": 6420
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.021077036857605,
      "learning_rate": 6.546867564596067e-05,
      "loss": 0.8033,
      "step": 6421
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9715837240219116,
      "learning_rate": 6.54404687734464e-05,
      "loss": 0.9326,
      "step": 6422
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.1517680883407593,
      "learning_rate": 6.541226502319004e-05,
      "loss": 0.7178,
      "step": 6423
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.0519474744796753,
      "learning_rate": 6.53840643977397e-05,
      "loss": 0.8789,
      "step": 6424
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9460950493812561,
      "learning_rate": 6.535586689964319e-05,
      "loss": 0.8135,
      "step": 6425
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.893166184425354,
      "learning_rate": 6.53276725314479e-05,
      "loss": 0.9067,
      "step": 6426
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8801971077919006,
      "learning_rate": 6.529948129570114e-05,
      "loss": 0.7642,
      "step": 6427
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9635523557662964,
      "learning_rate": 6.527129319494973e-05,
      "loss": 0.7357,
      "step": 6428
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8950421810150146,
      "learning_rate": 6.524310823174034e-05,
      "loss": 0.712,
      "step": 6429
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.019730567932129,
      "learning_rate": 6.521492640861938e-05,
      "loss": 0.9549,
      "step": 6430
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9443743228912354,
      "learning_rate": 6.518674772813281e-05,
      "loss": 0.8513,
      "step": 6431
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.0187928676605225,
      "learning_rate": 6.51585721928265e-05,
      "loss": 0.8161,
      "step": 6432
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.224259376525879,
      "learning_rate": 6.513039980524591e-05,
      "loss": 0.8436,
      "step": 6433
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8516876697540283,
      "learning_rate": 6.510223056793626e-05,
      "loss": 0.7481,
      "step": 6434
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9022040367126465,
      "learning_rate": 6.50740644834425e-05,
      "loss": 0.8293,
      "step": 6435
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9249104261398315,
      "learning_rate": 6.504590155430926e-05,
      "loss": 0.7153,
      "step": 6436
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8732777833938599,
      "learning_rate": 6.501774178308093e-05,
      "loss": 0.7169,
      "step": 6437
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8999826312065125,
      "learning_rate": 6.498958517230151e-05,
      "loss": 0.7997,
      "step": 6438
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9368231296539307,
      "learning_rate": 6.496143172451487e-05,
      "loss": 0.7556,
      "step": 6439
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.0797761678695679,
      "learning_rate": 6.49332814422645e-05,
      "loss": 0.9234,
      "step": 6440
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.1074094772338867,
      "learning_rate": 6.490513432809357e-05,
      "loss": 0.9232,
      "step": 6441
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8619398474693298,
      "learning_rate": 6.487699038454508e-05,
      "loss": 0.9323,
      "step": 6442
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.0510265827178955,
      "learning_rate": 6.484884961416162e-05,
      "loss": 0.8953,
      "step": 6443
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9262064099311829,
      "learning_rate": 6.482071201948556e-05,
      "loss": 0.8965,
      "step": 6444
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.952705442905426,
      "learning_rate": 6.479257760305904e-05,
      "loss": 0.7822,
      "step": 6445
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8337555527687073,
      "learning_rate": 6.476444636742374e-05,
      "loss": 0.7466,
      "step": 6446
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8132683634757996,
      "learning_rate": 6.473631831512122e-05,
      "loss": 0.878,
      "step": 6447
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8945775628089905,
      "learning_rate": 6.470819344869268e-05,
      "loss": 0.8282,
      "step": 6448
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.8810907006263733,
      "learning_rate": 6.468007177067902e-05,
      "loss": 0.8646,
      "step": 6449
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9152600765228271,
      "learning_rate": 6.465195328362092e-05,
      "loss": 0.8233,
      "step": 6450
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.956823468208313,
      "learning_rate": 6.46238379900587e-05,
      "loss": 0.8271,
      "step": 6451
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.0312672853469849,
      "learning_rate": 6.459572589253244e-05,
      "loss": 0.9571,
      "step": 6452
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.8972964286804199,
      "learning_rate": 6.456761699358184e-05,
      "loss": 0.9124,
      "step": 6453
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9931522607803345,
      "learning_rate": 6.453951129574644e-05,
      "loss": 0.853,
      "step": 6454
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.0029653310775757,
      "learning_rate": 6.451140880156545e-05,
      "loss": 0.8918,
      "step": 6455
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9245167970657349,
      "learning_rate": 6.448330951357769e-05,
      "loss": 0.8279,
      "step": 6456
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9320852756500244,
      "learning_rate": 6.445521343432188e-05,
      "loss": 0.8517,
      "step": 6457
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9748640060424805,
      "learning_rate": 6.442712056633622e-05,
      "loss": 0.7116,
      "step": 6458
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.8715943694114685,
      "learning_rate": 6.439903091215882e-05,
      "loss": 0.8655,
      "step": 6459
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9386141896247864,
      "learning_rate": 6.437094447432744e-05,
      "loss": 0.7586,
      "step": 6460
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9181474447250366,
      "learning_rate": 6.434286125537945e-05,
      "loss": 0.8425,
      "step": 6461
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9631593227386475,
      "learning_rate": 6.431478125785211e-05,
      "loss": 0.8938,
      "step": 6462
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.0355615615844727,
      "learning_rate": 6.428670448428218e-05,
      "loss": 0.7401,
      "step": 6463
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9288381934165955,
      "learning_rate": 6.425863093720628e-05,
      "loss": 0.7514,
      "step": 6464
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.1878407001495361,
      "learning_rate": 6.423056061916077e-05,
      "loss": 0.7921,
      "step": 6465
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9922305941581726,
      "learning_rate": 6.420249353268155e-05,
      "loss": 0.9356,
      "step": 6466
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.0492583513259888,
      "learning_rate": 6.417442968030437e-05,
      "loss": 1.0956,
      "step": 6467
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.0709989070892334,
      "learning_rate": 6.414636906456459e-05,
      "loss": 0.8801,
      "step": 6468
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.0354478359222412,
      "learning_rate": 6.411831168799738e-05,
      "loss": 0.7756,
      "step": 6469
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9077548384666443,
      "learning_rate": 6.409025755313757e-05,
      "loss": 0.7185,
      "step": 6470
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9772162437438965,
      "learning_rate": 6.406220666251964e-05,
      "loss": 0.8176,
      "step": 6471
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9777350425720215,
      "learning_rate": 6.403415901867788e-05,
      "loss": 0.8075,
      "step": 6472
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.8921540975570679,
      "learning_rate": 6.400611462414619e-05,
      "loss": 0.8218,
      "step": 6473
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.906919002532959,
      "learning_rate": 6.397807348145825e-05,
      "loss": 0.7705,
      "step": 6474
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.965469479560852,
      "learning_rate": 6.395003559314744e-05,
      "loss": 0.7674,
      "step": 6475
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.979746401309967,
      "learning_rate": 6.392200096174678e-05,
      "loss": 0.7292,
      "step": 6476
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9564318060874939,
      "learning_rate": 6.389396958978908e-05,
      "loss": 0.7416,
      "step": 6477
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9637297987937927,
      "learning_rate": 6.386594147980678e-05,
      "loss": 0.8265,
      "step": 6478
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.0443305969238281,
      "learning_rate": 6.383791663433206e-05,
      "loss": 0.7137,
      "step": 6479
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9801715016365051,
      "learning_rate": 6.380989505589686e-05,
      "loss": 0.8093,
      "step": 6480
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.8080372214317322,
      "learning_rate": 6.37818767470327e-05,
      "loss": 0.6817,
      "step": 6481
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9429090619087219,
      "learning_rate": 6.375386171027094e-05,
      "loss": 0.7979,
      "step": 6482
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9921965003013611,
      "learning_rate": 6.372584994814252e-05,
      "loss": 0.8558,
      "step": 6483
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9155974388122559,
      "learning_rate": 6.369784146317819e-05,
      "loss": 0.8999,
      "step": 6484
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9985141754150391,
      "learning_rate": 6.366983625790834e-05,
      "loss": 0.8533,
      "step": 6485
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9655118584632874,
      "learning_rate": 6.364183433486308e-05,
      "loss": 0.8803,
      "step": 6486
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9485723972320557,
      "learning_rate": 6.361383569657223e-05,
      "loss": 0.8161,
      "step": 6487
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9708414673805237,
      "learning_rate": 6.35858403455653e-05,
      "loss": 0.8069,
      "step": 6488
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9146203398704529,
      "learning_rate": 6.355784828437153e-05,
      "loss": 0.7338,
      "step": 6489
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9346987009048462,
      "learning_rate": 6.352985951551983e-05,
      "loss": 0.9185,
      "step": 6490
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9882563948631287,
      "learning_rate": 6.350187404153883e-05,
      "loss": 0.844,
      "step": 6491
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9026610851287842,
      "learning_rate": 6.347389186495688e-05,
      "loss": 0.8361,
      "step": 6492
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.008806824684143,
      "learning_rate": 6.344591298830193e-05,
      "loss": 0.8584,
      "step": 6493
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.0522782802581787,
      "learning_rate": 6.34179374141018e-05,
      "loss": 0.7946,
      "step": 6494
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9980722069740295,
      "learning_rate": 6.338996514488395e-05,
      "loss": 1.0012,
      "step": 6495
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.0276881456375122,
      "learning_rate": 6.336199618317537e-05,
      "loss": 0.8609,
      "step": 6496
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.0530096292495728,
      "learning_rate": 6.333403053150308e-05,
      "loss": 0.8101,
      "step": 6497
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.0483702421188354,
      "learning_rate": 6.330606819239348e-05,
      "loss": 0.8043,
      "step": 6498
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.0610220432281494,
      "learning_rate": 6.327810916837284e-05,
      "loss": 0.7935,
      "step": 6499
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9744018316268921,
      "learning_rate": 6.325015346196719e-05,
      "loss": 0.9037,
      "step": 6500
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.0621654987335205,
      "learning_rate": 6.322220107570205e-05,
      "loss": 0.921,
      "step": 6501
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.1827298402786255,
      "learning_rate": 6.319425201210281e-05,
      "loss": 0.7848,
      "step": 6502
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.225909948348999,
      "learning_rate": 6.316630627369451e-05,
      "loss": 1.0015,
      "step": 6503
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.996768057346344,
      "learning_rate": 6.313836386300185e-05,
      "loss": 0.9625,
      "step": 6504
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9362988471984863,
      "learning_rate": 6.311042478254938e-05,
      "loss": 0.8137,
      "step": 6505
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9462225437164307,
      "learning_rate": 6.308248903486111e-05,
      "loss": 0.775,
      "step": 6506
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.0419329404830933,
      "learning_rate": 6.305455662246096e-05,
      "loss": 0.9317,
      "step": 6507
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.862458348274231,
      "learning_rate": 6.302662754787241e-05,
      "loss": 0.8419,
      "step": 6508
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9084261655807495,
      "learning_rate": 6.29987018136187e-05,
      "loss": 0.8696,
      "step": 6509
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.018864393234253,
      "learning_rate": 6.297077942222282e-05,
      "loss": 0.8147,
      "step": 6510
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.935609757900238,
      "learning_rate": 6.294286037620734e-05,
      "loss": 0.7524,
      "step": 6511
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.941339910030365,
      "learning_rate": 6.291494467809462e-05,
      "loss": 0.9219,
      "step": 6512
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.002559781074524,
      "learning_rate": 6.288703233040667e-05,
      "loss": 0.8783,
      "step": 6513
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9348935484886169,
      "learning_rate": 6.28591233356652e-05,
      "loss": 0.9559,
      "step": 6514
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9044655561447144,
      "learning_rate": 6.283121769639167e-05,
      "loss": 0.7447,
      "step": 6515
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9007348418235779,
      "learning_rate": 6.280331541510716e-05,
      "loss": 0.7183,
      "step": 6516
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9882896542549133,
      "learning_rate": 6.277541649433249e-05,
      "loss": 0.6538,
      "step": 6517
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.8573456406593323,
      "learning_rate": 6.274752093658818e-05,
      "loss": 0.846,
      "step": 6518
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.8707038164138794,
      "learning_rate": 6.271962874439442e-05,
      "loss": 0.7129,
      "step": 6519
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.0396522283554077,
      "learning_rate": 6.269173992027114e-05,
      "loss": 0.8879,
      "step": 6520
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9969940781593323,
      "learning_rate": 6.26638544667379e-05,
      "loss": 0.8554,
      "step": 6521
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9481281638145447,
      "learning_rate": 6.263597238631405e-05,
      "loss": 0.8698,
      "step": 6522
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.0395705699920654,
      "learning_rate": 6.26080936815185e-05,
      "loss": 0.6371,
      "step": 6523
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9268097877502441,
      "learning_rate": 6.258021835486998e-05,
      "loss": 0.94,
      "step": 6524
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9226809740066528,
      "learning_rate": 6.255234640888688e-05,
      "loss": 0.7378,
      "step": 6525
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.973568856716156,
      "learning_rate": 6.252447784608724e-05,
      "loss": 0.9353,
      "step": 6526
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9399523735046387,
      "learning_rate": 6.249661266898887e-05,
      "loss": 0.9005,
      "step": 6527
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.8221200704574585,
      "learning_rate": 6.246875088010915e-05,
      "loss": 0.7824,
      "step": 6528
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.91694176197052,
      "learning_rate": 6.244089248196529e-05,
      "loss": 0.719,
      "step": 6529
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9563716650009155,
      "learning_rate": 6.24130374770742e-05,
      "loss": 0.9026,
      "step": 6530
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.0524191856384277,
      "learning_rate": 6.238518586795229e-05,
      "loss": 0.7846,
      "step": 6531
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9351217746734619,
      "learning_rate": 6.235733765711592e-05,
      "loss": 0.8142,
      "step": 6532
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.8990921974182129,
      "learning_rate": 6.23294928470809e-05,
      "loss": 0.8337,
      "step": 6533
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.148982286453247,
      "learning_rate": 6.230165144036293e-05,
      "loss": 0.866,
      "step": 6534
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9928149580955505,
      "learning_rate": 6.227381343947733e-05,
      "loss": 0.9086,
      "step": 6535
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.0408509969711304,
      "learning_rate": 6.224597884693905e-05,
      "loss": 0.8129,
      "step": 6536
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9203112125396729,
      "learning_rate": 6.221814766526286e-05,
      "loss": 0.8097,
      "step": 6537
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9963998198509216,
      "learning_rate": 6.219031989696307e-05,
      "loss": 0.81,
      "step": 6538
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.8776340484619141,
      "learning_rate": 6.216249554455377e-05,
      "loss": 0.7176,
      "step": 6539
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9041548371315002,
      "learning_rate": 6.213467461054885e-05,
      "loss": 0.7137,
      "step": 6540
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9312465786933899,
      "learning_rate": 6.210685709746163e-05,
      "loss": 0.7677,
      "step": 6541
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9554211497306824,
      "learning_rate": 6.207904300780534e-05,
      "loss": 0.7852,
      "step": 6542
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9075998663902283,
      "learning_rate": 6.20512323440928e-05,
      "loss": 0.8632,
      "step": 6543
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.8768923282623291,
      "learning_rate": 6.202342510883653e-05,
      "loss": 0.8232,
      "step": 6544
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9523442983627319,
      "learning_rate": 6.199562130454882e-05,
      "loss": 0.8195,
      "step": 6545
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9885941743850708,
      "learning_rate": 6.196782093374149e-05,
      "loss": 0.8468,
      "step": 6546
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9235928058624268,
      "learning_rate": 6.194002399892625e-05,
      "loss": 0.6699,
      "step": 6547
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9114676117897034,
      "learning_rate": 6.191223050261432e-05,
      "loss": 0.9604,
      "step": 6548
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.0495816469192505,
      "learning_rate": 6.188444044731671e-05,
      "loss": 0.7926,
      "step": 6549
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.954431414604187,
      "learning_rate": 6.185665383554414e-05,
      "loss": 0.8283,
      "step": 6550
    },
    {
      "epoch": 2.46,
      "eval_loss": 0.983707070350647,
      "eval_runtime": 17.7959,
      "eval_samples_per_second": 153.35,
      "eval_steps_per_second": 38.38,
      "step": 6550
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9318667650222778,
      "learning_rate": 6.182887066980687e-05,
      "loss": 0.7753,
      "step": 6551
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.8867954611778259,
      "learning_rate": 6.180109095261505e-05,
      "loss": 0.6673,
      "step": 6552
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.917029082775116,
      "learning_rate": 6.177331468647835e-05,
      "loss": 0.7575,
      "step": 6553
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9577431678771973,
      "learning_rate": 6.174554187390625e-05,
      "loss": 0.8405,
      "step": 6554
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.0297268629074097,
      "learning_rate": 6.171777251740785e-05,
      "loss": 0.8071,
      "step": 6555
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.033363699913025,
      "learning_rate": 6.169000661949194e-05,
      "loss": 0.9187,
      "step": 6556
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8895694613456726,
      "learning_rate": 6.166224418266701e-05,
      "loss": 0.7009,
      "step": 6557
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.0453028678894043,
      "learning_rate": 6.163448520944128e-05,
      "loss": 0.9334,
      "step": 6558
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.1392394304275513,
      "learning_rate": 6.160672970232255e-05,
      "loss": 0.9162,
      "step": 6559
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9568962454795837,
      "learning_rate": 6.157897766381845e-05,
      "loss": 0.7592,
      "step": 6560
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9768176674842834,
      "learning_rate": 6.155122909643615e-05,
      "loss": 0.9007,
      "step": 6561
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9717034697532654,
      "learning_rate": 6.152348400268259e-05,
      "loss": 0.9142,
      "step": 6562
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9644777774810791,
      "learning_rate": 6.149574238506444e-05,
      "loss": 0.8193,
      "step": 6563
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9324485659599304,
      "learning_rate": 6.146800424608793e-05,
      "loss": 0.9476,
      "step": 6564
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.04218590259552,
      "learning_rate": 6.144026958825909e-05,
      "loss": 0.8632,
      "step": 6565
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8786495327949524,
      "learning_rate": 6.141253841408354e-05,
      "loss": 0.9254,
      "step": 6566
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.947974443435669,
      "learning_rate": 6.138481072606665e-05,
      "loss": 0.7575,
      "step": 6567
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.0358220338821411,
      "learning_rate": 6.135708652671353e-05,
      "loss": 0.8487,
      "step": 6568
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.0438722372055054,
      "learning_rate": 6.13293658185288e-05,
      "loss": 0.8242,
      "step": 6569
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9584254622459412,
      "learning_rate": 6.130164860401695e-05,
      "loss": 0.6414,
      "step": 6570
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3015732765197754,
      "learning_rate": 6.127393488568199e-05,
      "loss": 0.7602,
      "step": 6571
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.850447416305542,
      "learning_rate": 6.124622466602777e-05,
      "loss": 0.8707,
      "step": 6572
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8629624247550964,
      "learning_rate": 6.121851794755776e-05,
      "loss": 0.8217,
      "step": 6573
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9459384083747864,
      "learning_rate": 6.119081473277501e-05,
      "loss": 0.8037,
      "step": 6574
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.0278830528259277,
      "learning_rate": 6.116311502418249e-05,
      "loss": 0.8145,
      "step": 6575
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.0946799516677856,
      "learning_rate": 6.113541882428259e-05,
      "loss": 0.8438,
      "step": 6576
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8945231437683105,
      "learning_rate": 6.110772613557751e-05,
      "loss": 0.7793,
      "step": 6577
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9611460566520691,
      "learning_rate": 6.108003696056924e-05,
      "loss": 0.8309,
      "step": 6578
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.0656673908233643,
      "learning_rate": 6.105235130175922e-05,
      "loss": 0.7861,
      "step": 6579
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9813756346702576,
      "learning_rate": 6.102466916164877e-05,
      "loss": 0.8286,
      "step": 6580
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9764003753662109,
      "learning_rate": 6.099699054273875e-05,
      "loss": 0.9217,
      "step": 6581
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.0544682741165161,
      "learning_rate": 6.09693154475298e-05,
      "loss": 0.8863,
      "step": 6582
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.0966427326202393,
      "learning_rate": 6.094164387852225e-05,
      "loss": 0.863,
      "step": 6583
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8946224451065063,
      "learning_rate": 6.091397583821599e-05,
      "loss": 0.8078,
      "step": 6584
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9691383242607117,
      "learning_rate": 6.088631132911073e-05,
      "loss": 0.88,
      "step": 6585
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.0494399070739746,
      "learning_rate": 6.085865035370577e-05,
      "loss": 0.8208,
      "step": 6586
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9896689057350159,
      "learning_rate": 6.0830992914500117e-05,
      "loss": 0.9572,
      "step": 6587
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.0214627981185913,
      "learning_rate": 6.080333901399251e-05,
      "loss": 0.9282,
      "step": 6588
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9828228950500488,
      "learning_rate": 6.077568865468126e-05,
      "loss": 0.931,
      "step": 6589
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.065006971359253,
      "learning_rate": 6.074804183906448e-05,
      "loss": 0.8782,
      "step": 6590
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9665316939353943,
      "learning_rate": 6.072039856963987e-05,
      "loss": 0.901,
      "step": 6591
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8945008516311646,
      "learning_rate": 6.069275884890483e-05,
      "loss": 0.8144,
      "step": 6592
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8783544301986694,
      "learning_rate": 6.0665122679356504e-05,
      "loss": 0.8429,
      "step": 6593
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8953959345817566,
      "learning_rate": 6.06374900634916e-05,
      "loss": 0.8561,
      "step": 6594
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8966822624206543,
      "learning_rate": 6.0609861003806624e-05,
      "loss": 0.9078,
      "step": 6595
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.0480576753616333,
      "learning_rate": 6.0582235502797646e-05,
      "loss": 0.7269,
      "step": 6596
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8792452216148376,
      "learning_rate": 6.0554613562960526e-05,
      "loss": 0.7717,
      "step": 6597
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.1262755393981934,
      "learning_rate": 6.0526995186790746e-05,
      "loss": 0.9037,
      "step": 6598
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.0024149417877197,
      "learning_rate": 6.049938037678344e-05,
      "loss": 0.898,
      "step": 6599
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.1063541173934937,
      "learning_rate": 6.0471769135433475e-05,
      "loss": 0.7537,
      "step": 6600
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9563291668891907,
      "learning_rate": 6.0444161465235346e-05,
      "loss": 0.8024,
      "step": 6601
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.108036994934082,
      "learning_rate": 6.0416557368683256e-05,
      "loss": 0.9593,
      "step": 6602
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.040736198425293,
      "learning_rate": 6.038895684827113e-05,
      "loss": 0.7056,
      "step": 6603
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.1083346605300903,
      "learning_rate": 6.036135990649243e-05,
      "loss": 0.9539,
      "step": 6604
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9028723239898682,
      "learning_rate": 6.033376654584049e-05,
      "loss": 0.8037,
      "step": 6605
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.1071391105651855,
      "learning_rate": 6.0306176768808075e-05,
      "loss": 0.8398,
      "step": 6606
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.0021347999572754,
      "learning_rate": 6.027859057788787e-05,
      "loss": 0.91,
      "step": 6607
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9143244624137878,
      "learning_rate": 6.0251007975572124e-05,
      "loss": 0.8246,
      "step": 6608
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9034011363983154,
      "learning_rate": 6.022342896435272e-05,
      "loss": 0.7811,
      "step": 6609
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.0634794235229492,
      "learning_rate": 6.019585354672135e-05,
      "loss": 0.8071,
      "step": 6610
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9506297707557678,
      "learning_rate": 6.016828172516917e-05,
      "loss": 0.668,
      "step": 6611
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9064805507659912,
      "learning_rate": 6.014071350218724e-05,
      "loss": 0.6629,
      "step": 6612
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.0146342515945435,
      "learning_rate": 6.01131488802662e-05,
      "loss": 0.9008,
      "step": 6613
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.0808639526367188,
      "learning_rate": 6.008558786189625e-05,
      "loss": 0.909,
      "step": 6614
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.979100227355957,
      "learning_rate": 6.0058030449567516e-05,
      "loss": 0.8102,
      "step": 6615
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.0906333923339844,
      "learning_rate": 6.003047664576953e-05,
      "loss": 0.8564,
      "step": 6616
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9375736713409424,
      "learning_rate": 6.000292645299166e-05,
      "loss": 0.7611,
      "step": 6617
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9777929782867432,
      "learning_rate": 5.9975379873722967e-05,
      "loss": 0.8396,
      "step": 6618
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9256249070167542,
      "learning_rate": 5.994783691045206e-05,
      "loss": 0.8613,
      "step": 6619
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9322515726089478,
      "learning_rate": 5.9920297565667326e-05,
      "loss": 0.7838,
      "step": 6620
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9981560111045837,
      "learning_rate": 5.989276184185676e-05,
      "loss": 0.7834,
      "step": 6621
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.0275251865386963,
      "learning_rate": 5.9865229741508075e-05,
      "loss": 0.8052,
      "step": 6622
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8906618356704712,
      "learning_rate": 5.9837701267108656e-05,
      "loss": 0.8151,
      "step": 6623
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8646613955497742,
      "learning_rate": 5.9810176421145506e-05,
      "loss": 0.8311,
      "step": 6624
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8780609369277954,
      "learning_rate": 5.97826552061054e-05,
      "loss": 0.7355,
      "step": 6625
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9192318320274353,
      "learning_rate": 5.975513762447464e-05,
      "loss": 0.8054,
      "step": 6626
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.941154956817627,
      "learning_rate": 5.972762367873934e-05,
      "loss": 0.8493,
      "step": 6627
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.036248803138733,
      "learning_rate": 5.970011337138526e-05,
      "loss": 0.8219,
      "step": 6628
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9037586450576782,
      "learning_rate": 5.9672606704897713e-05,
      "loss": 0.7278,
      "step": 6629
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8977223634719849,
      "learning_rate": 5.964510368176185e-05,
      "loss": 0.8595,
      "step": 6630
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9640159010887146,
      "learning_rate": 5.961760430446235e-05,
      "loss": 0.9585,
      "step": 6631
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9565543532371521,
      "learning_rate": 5.9590108575483663e-05,
      "loss": 0.8759,
      "step": 6632
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9012134075164795,
      "learning_rate": 5.95626164973099e-05,
      "loss": 0.7658,
      "step": 6633
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.874117910861969,
      "learning_rate": 5.953512807242474e-05,
      "loss": 0.7262,
      "step": 6634
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9759389758110046,
      "learning_rate": 5.950764330331167e-05,
      "loss": 0.8136,
      "step": 6635
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.065969467163086,
      "learning_rate": 5.948016219245375e-05,
      "loss": 0.7858,
      "step": 6636
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9532552361488342,
      "learning_rate": 5.945268474233374e-05,
      "loss": 0.8543,
      "step": 6637
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9501945972442627,
      "learning_rate": 5.942521095543412e-05,
      "loss": 0.7695,
      "step": 6638
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.001517653465271,
      "learning_rate": 5.939774083423693e-05,
      "loss": 0.8325,
      "step": 6639
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.999535858631134,
      "learning_rate": 5.9370274381223975e-05,
      "loss": 0.7181,
      "step": 6640
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9850019216537476,
      "learning_rate": 5.934281159887668e-05,
      "loss": 0.9851,
      "step": 6641
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.949331521987915,
      "learning_rate": 5.9315352489676144e-05,
      "loss": 0.8008,
      "step": 6642
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.0675513744354248,
      "learning_rate": 5.928789705610319e-05,
      "loss": 0.7361,
      "step": 6643
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9310736656188965,
      "learning_rate": 5.92604453006382e-05,
      "loss": 0.8125,
      "step": 6644
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.911928653717041,
      "learning_rate": 5.9232997225761343e-05,
      "loss": 0.8385,
      "step": 6645
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.0393922328948975,
      "learning_rate": 5.9205552833952316e-05,
      "loss": 0.8595,
      "step": 6646
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9656948447227478,
      "learning_rate": 5.9178112127690625e-05,
      "loss": 0.793,
      "step": 6647
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.086967945098877,
      "learning_rate": 5.915067510945542e-05,
      "loss": 0.9268,
      "step": 6648
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.8930530548095703,
      "learning_rate": 5.912324178172536e-05,
      "loss": 0.9024,
      "step": 6649
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.0900965929031372,
      "learning_rate": 5.9095812146979024e-05,
      "loss": 0.8494,
      "step": 6650
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.932288646697998,
      "learning_rate": 5.906838620769442e-05,
      "loss": 0.9706,
      "step": 6651
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9077619314193726,
      "learning_rate": 5.904096396634935e-05,
      "loss": 0.7886,
      "step": 6652
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.8794068694114685,
      "learning_rate": 5.901354542542133e-05,
      "loss": 0.7029,
      "step": 6653
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.8963539600372314,
      "learning_rate": 5.898613058738737e-05,
      "loss": 0.7728,
      "step": 6654
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.1612012386322021,
      "learning_rate": 5.8958719454724346e-05,
      "loss": 0.9314,
      "step": 6655
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.881064236164093,
      "learning_rate": 5.89313120299086e-05,
      "loss": 0.7476,
      "step": 6656
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9086899161338806,
      "learning_rate": 5.890390831541626e-05,
      "loss": 0.8062,
      "step": 6657
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.013569712638855,
      "learning_rate": 5.8876508313723175e-05,
      "loss": 0.8136,
      "step": 6658
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.1807419061660767,
      "learning_rate": 5.8849112027304695e-05,
      "loss": 0.8138,
      "step": 6659
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.0075334310531616,
      "learning_rate": 5.882171945863597e-05,
      "loss": 0.951,
      "step": 6660
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.961222767829895,
      "learning_rate": 5.87943306101917e-05,
      "loss": 0.8983,
      "step": 6661
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.0966322422027588,
      "learning_rate": 5.876694548444638e-05,
      "loss": 0.7196,
      "step": 6662
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9493511319160461,
      "learning_rate": 5.873956408387409e-05,
      "loss": 0.7949,
      "step": 6663
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9226420521736145,
      "learning_rate": 5.8712186410948536e-05,
      "loss": 0.874,
      "step": 6664
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9676820039749146,
      "learning_rate": 5.8684812468143194e-05,
      "loss": 0.8907,
      "step": 6665
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9125047922134399,
      "learning_rate": 5.8657442257931094e-05,
      "loss": 0.7284,
      "step": 6666
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8482916355133057,
      "learning_rate": 5.863007578278501e-05,
      "loss": 0.7075,
      "step": 6667
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.056573510169983,
      "learning_rate": 5.860271304517737e-05,
      "loss": 0.927,
      "step": 6668
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9677813649177551,
      "learning_rate": 5.857535404758019e-05,
      "loss": 0.8479,
      "step": 6669
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9716077446937561,
      "learning_rate": 5.854799879246524e-05,
      "loss": 0.8654,
      "step": 6670
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9621649384498596,
      "learning_rate": 5.852064728230389e-05,
      "loss": 0.7844,
      "step": 6671
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.0075055360794067,
      "learning_rate": 5.849329951956719e-05,
      "loss": 0.6371,
      "step": 6672
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.0559271574020386,
      "learning_rate": 5.846595550672589e-05,
      "loss": 1.0733,
      "step": 6673
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9799260497093201,
      "learning_rate": 5.843861524625034e-05,
      "loss": 0.758,
      "step": 6674
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9251304268836975,
      "learning_rate": 5.841127874061058e-05,
      "loss": 0.8044,
      "step": 6675
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.0604742765426636,
      "learning_rate": 5.838394599227632e-05,
      "loss": 0.8555,
      "step": 6676
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.991960346698761,
      "learning_rate": 5.8356617003716885e-05,
      "loss": 0.7523,
      "step": 6677
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8831610679626465,
      "learning_rate": 5.832929177740134e-05,
      "loss": 0.8091,
      "step": 6678
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.1847463846206665,
      "learning_rate": 5.830197031579836e-05,
      "loss": 0.9277,
      "step": 6679
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.0120302438735962,
      "learning_rate": 5.827465262137626e-05,
      "loss": 0.8833,
      "step": 6680
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.861012876033783,
      "learning_rate": 5.824733869660299e-05,
      "loss": 0.7181,
      "step": 6681
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9979028105735779,
      "learning_rate": 5.822002854394633e-05,
      "loss": 0.7781,
      "step": 6682
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8775619268417358,
      "learning_rate": 5.819272216587352e-05,
      "loss": 0.8617,
      "step": 6683
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.056386947631836,
      "learning_rate": 5.816541956485153e-05,
      "loss": 0.7084,
      "step": 6684
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.988563597202301,
      "learning_rate": 5.8138120743347034e-05,
      "loss": 0.8537,
      "step": 6685
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0198968648910522,
      "learning_rate": 5.811082570382632e-05,
      "loss": 0.7832,
      "step": 6686
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.1628360748291016,
      "learning_rate": 5.8083534448755284e-05,
      "loss": 0.8602,
      "step": 6687
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0326400995254517,
      "learning_rate": 5.8056246980599615e-05,
      "loss": 0.9324,
      "step": 6688
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9955177903175354,
      "learning_rate": 5.8028963301824546e-05,
      "loss": 0.8435,
      "step": 6689
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.8769357204437256,
      "learning_rate": 5.800168341489501e-05,
      "loss": 0.7892,
      "step": 6690
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9578089714050293,
      "learning_rate": 5.797440732227555e-05,
      "loss": 0.9745,
      "step": 6691
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.1331582069396973,
      "learning_rate": 5.794713502643048e-05,
      "loss": 0.8677,
      "step": 6692
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.919992208480835,
      "learning_rate": 5.791986652982364e-05,
      "loss": 0.8013,
      "step": 6693
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.028770089149475,
      "learning_rate": 5.789260183491857e-05,
      "loss": 0.7987,
      "step": 6694
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.001869797706604,
      "learning_rate": 5.7865340944178545e-05,
      "loss": 0.8242,
      "step": 6695
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0222487449645996,
      "learning_rate": 5.783808386006642e-05,
      "loss": 0.8577,
      "step": 6696
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9631983637809753,
      "learning_rate": 5.781083058504464e-05,
      "loss": 0.8824,
      "step": 6697
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0110154151916504,
      "learning_rate": 5.7783581121575494e-05,
      "loss": 0.7438,
      "step": 6698
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9673230051994324,
      "learning_rate": 5.775633547212076e-05,
      "loss": 0.948,
      "step": 6699
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9207396507263184,
      "learning_rate": 5.772909363914194e-05,
      "loss": 0.7733,
      "step": 6700
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.1381072998046875,
      "learning_rate": 5.770185562510012e-05,
      "loss": 0.8291,
      "step": 6701
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0670325756072998,
      "learning_rate": 5.76746214324562e-05,
      "loss": 1.0664,
      "step": 6702
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0272499322891235,
      "learning_rate": 5.764739106367059e-05,
      "loss": 0.917,
      "step": 6703
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0990335941314697,
      "learning_rate": 5.7620164521203356e-05,
      "loss": 0.8484,
      "step": 6704
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9762868881225586,
      "learning_rate": 5.7592941807514375e-05,
      "loss": 0.7325,
      "step": 6705
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0388116836547852,
      "learning_rate": 5.756572292506293e-05,
      "loss": 0.8576,
      "step": 6706
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0626600980758667,
      "learning_rate": 5.7538507876308145e-05,
      "loss": 0.8496,
      "step": 6707
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.904132068157196,
      "learning_rate": 5.7511296663708836e-05,
      "loss": 0.7153,
      "step": 6708
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.8037851452827454,
      "learning_rate": 5.748408928972322e-05,
      "loss": 0.7363,
      "step": 6709
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9641945362091064,
      "learning_rate": 5.7456885756809465e-05,
      "loss": 0.8954,
      "step": 6710
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8656486868858337,
      "learning_rate": 5.742968606742519e-05,
      "loss": 0.7797,
      "step": 6711
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9754137396812439,
      "learning_rate": 5.7402490224027716e-05,
      "loss": 0.9204,
      "step": 6712
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9762657284736633,
      "learning_rate": 5.7375298229074105e-05,
      "loss": 0.8431,
      "step": 6713
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.0026696920394897,
      "learning_rate": 5.734811008502095e-05,
      "loss": 0.9341,
      "step": 6714
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8647763133049011,
      "learning_rate": 5.732092579432455e-05,
      "loss": 0.8312,
      "step": 6715
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8759633898735046,
      "learning_rate": 5.729374535944083e-05,
      "loss": 0.7859,
      "step": 6716
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9404118657112122,
      "learning_rate": 5.7266568782825455e-05,
      "loss": 0.8811,
      "step": 6717
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.0193644762039185,
      "learning_rate": 5.723939606693362e-05,
      "loss": 0.8021,
      "step": 6718
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9990611672401428,
      "learning_rate": 5.721222721422022e-05,
      "loss": 0.881,
      "step": 6719
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.059485912322998,
      "learning_rate": 5.718506222713985e-05,
      "loss": 0.9278,
      "step": 6720
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8559978604316711,
      "learning_rate": 5.715790110814669e-05,
      "loss": 0.8284,
      "step": 6721
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9846999049186707,
      "learning_rate": 5.713074385969457e-05,
      "loss": 0.8801,
      "step": 6722
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.0460660457611084,
      "learning_rate": 5.710359048423706e-05,
      "loss": 0.9067,
      "step": 6723
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.0111370086669922,
      "learning_rate": 5.7076440984227244e-05,
      "loss": 0.8792,
      "step": 6724
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.025821328163147,
      "learning_rate": 5.704929536211797e-05,
      "loss": 0.7989,
      "step": 6725
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.0277032852172852,
      "learning_rate": 5.702215362036163e-05,
      "loss": 0.9084,
      "step": 6726
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8618652820587158,
      "learning_rate": 5.699501576141042e-05,
      "loss": 0.7553,
      "step": 6727
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.1251543760299683,
      "learning_rate": 5.6967881787716025e-05,
      "loss": 0.8687,
      "step": 6728
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.066477656364441,
      "learning_rate": 5.6940751701729834e-05,
      "loss": 0.7052,
      "step": 6729
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8859666585922241,
      "learning_rate": 5.691362550590297e-05,
      "loss": 0.7717,
      "step": 6730
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8953352570533752,
      "learning_rate": 5.688650320268607e-05,
      "loss": 0.6993,
      "step": 6731
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.0332285165786743,
      "learning_rate": 5.685938479452948e-05,
      "loss": 0.8573,
      "step": 6732
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.962002694606781,
      "learning_rate": 5.683227028388324e-05,
      "loss": 0.883,
      "step": 6733
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9662851691246033,
      "learning_rate": 5.680515967319696e-05,
      "loss": 0.8588,
      "step": 6734
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.0482203960418701,
      "learning_rate": 5.6778052964919937e-05,
      "loss": 0.9022,
      "step": 6735
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.003685474395752,
      "learning_rate": 5.675095016150107e-05,
      "loss": 0.7915,
      "step": 6736
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.002968192100525,
      "learning_rate": 5.672385126538903e-05,
      "loss": 0.7781,
      "step": 6737
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9445874094963074,
      "learning_rate": 5.6696756279031994e-05,
      "loss": 0.8009,
      "step": 6738
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.953427255153656,
      "learning_rate": 5.6669665204877796e-05,
      "loss": 0.7298,
      "step": 6739
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9817003607749939,
      "learning_rate": 5.66425780453741e-05,
      "loss": 0.9132,
      "step": 6740
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9142551422119141,
      "learning_rate": 5.6615494802967905e-05,
      "loss": 0.7375,
      "step": 6741
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.0406633615493774,
      "learning_rate": 5.658841548010612e-05,
      "loss": 0.8746,
      "step": 6742
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.935921847820282,
      "learning_rate": 5.6561340079235256e-05,
      "loss": 0.7827,
      "step": 6743
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9558234810829163,
      "learning_rate": 5.653426860280131e-05,
      "loss": 0.7314,
      "step": 6744
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2024295330047607,
      "learning_rate": 5.650720105325014e-05,
      "loss": 0.8042,
      "step": 6745
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.699968695640564,
      "learning_rate": 5.648013743302708e-05,
      "loss": 0.8794,
      "step": 6746
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.8990693092346191,
      "learning_rate": 5.645307774457716e-05,
      "loss": 0.7162,
      "step": 6747
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9553346037864685,
      "learning_rate": 5.6426021990345144e-05,
      "loss": 0.7503,
      "step": 6748
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9533629417419434,
      "learning_rate": 5.639897017277533e-05,
      "loss": 0.7091,
      "step": 6749
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9082868099212646,
      "learning_rate": 5.637192229431168e-05,
      "loss": 0.7923,
      "step": 6750
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.0246829986572266,
      "learning_rate": 5.634487835739781e-05,
      "loss": 0.9403,
      "step": 6751
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.8718262910842896,
      "learning_rate": 5.6317838364477026e-05,
      "loss": 0.7546,
      "step": 6752
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9176360368728638,
      "learning_rate": 5.6290802317992234e-05,
      "loss": 0.8194,
      "step": 6753
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.8438771963119507,
      "learning_rate": 5.626377022038592e-05,
      "loss": 0.7401,
      "step": 6754
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.0411665439605713,
      "learning_rate": 5.623674207410038e-05,
      "loss": 0.8231,
      "step": 6755
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9400526285171509,
      "learning_rate": 5.620971788157741e-05,
      "loss": 0.7444,
      "step": 6756
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9847108721733093,
      "learning_rate": 5.618269764525843e-05,
      "loss": 0.7339,
      "step": 6757
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.1326313018798828,
      "learning_rate": 5.6155681367584675e-05,
      "loss": 0.8089,
      "step": 6758
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.0429010391235352,
      "learning_rate": 5.612866905099687e-05,
      "loss": 0.8034,
      "step": 6759
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9895896911621094,
      "learning_rate": 5.610166069793541e-05,
      "loss": 0.9088,
      "step": 6760
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.892261803150177,
      "learning_rate": 5.607465631084032e-05,
      "loss": 0.8191,
      "step": 6761
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.952148973941803,
      "learning_rate": 5.604765589215136e-05,
      "loss": 0.9722,
      "step": 6762
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.1106581687927246,
      "learning_rate": 5.6020659444307844e-05,
      "loss": 0.7013,
      "step": 6763
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.944435715675354,
      "learning_rate": 5.599366696974868e-05,
      "loss": 0.8013,
      "step": 6764
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9799804091453552,
      "learning_rate": 5.5966678470912596e-05,
      "loss": 0.8255,
      "step": 6765
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.953001856803894,
      "learning_rate": 5.593969395023779e-05,
      "loss": 0.7889,
      "step": 6766
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.0873104333877563,
      "learning_rate": 5.591271341016213e-05,
      "loss": 0.7998,
      "step": 6767
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.0211257934570312,
      "learning_rate": 5.588573685312324e-05,
      "loss": 0.8481,
      "step": 6768
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9410220980644226,
      "learning_rate": 5.585876428155824e-05,
      "loss": 0.8061,
      "step": 6769
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9582226872444153,
      "learning_rate": 5.5831795697903957e-05,
      "loss": 0.8484,
      "step": 6770
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.0626345872879028,
      "learning_rate": 5.580483110459682e-05,
      "loss": 0.7536,
      "step": 6771
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.0079556703567505,
      "learning_rate": 5.577787050407299e-05,
      "loss": 0.8331,
      "step": 6772
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9008169174194336,
      "learning_rate": 5.5750913898768186e-05,
      "loss": 0.8268,
      "step": 6773
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.1124671697616577,
      "learning_rate": 5.572396129111774e-05,
      "loss": 0.9713,
      "step": 6774
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.94371098279953,
      "learning_rate": 5.569701268355677e-05,
      "loss": 0.9088,
      "step": 6775
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.8979617357254028,
      "learning_rate": 5.567006807851978e-05,
      "loss": 0.7951,
      "step": 6776
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9180106520652771,
      "learning_rate": 5.564312747844116e-05,
      "loss": 0.8821,
      "step": 6777
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.0199891328811646,
      "learning_rate": 5.561619088575488e-05,
      "loss": 0.7996,
      "step": 6778
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9957209825515747,
      "learning_rate": 5.5589258302894386e-05,
      "loss": 0.8374,
      "step": 6779
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9373449087142944,
      "learning_rate": 5.5562329732292984e-05,
      "loss": 0.8361,
      "step": 6780
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9723969101905823,
      "learning_rate": 5.5535405176383495e-05,
      "loss": 0.8756,
      "step": 6781
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9420078992843628,
      "learning_rate": 5.550848463759835e-05,
      "loss": 0.7588,
      "step": 6782
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9574902057647705,
      "learning_rate": 5.5481568118369744e-05,
      "loss": 0.8255,
      "step": 6783
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.0184106826782227,
      "learning_rate": 5.545465562112935e-05,
      "loss": 0.8957,
      "step": 6784
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9024528861045837,
      "learning_rate": 5.542774714830869e-05,
      "loss": 0.7603,
      "step": 6785
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9247129559516907,
      "learning_rate": 5.540084270233863e-05,
      "loss": 0.8698,
      "step": 6786
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9607877731323242,
      "learning_rate": 5.537394228564992e-05,
      "loss": 0.8561,
      "step": 6787
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.0621390342712402,
      "learning_rate": 5.534704590067292e-05,
      "loss": 0.8024,
      "step": 6788
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9471501708030701,
      "learning_rate": 5.5320153549837415e-05,
      "loss": 0.8535,
      "step": 6789
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.107270359992981,
      "learning_rate": 5.529326523557312e-05,
      "loss": 0.8063,
      "step": 6790
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9283701181411743,
      "learning_rate": 5.526638096030916e-05,
      "loss": 0.7854,
      "step": 6791
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.0086311101913452,
      "learning_rate": 5.523950072647437e-05,
      "loss": 0.8637,
      "step": 6792
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.8553213477134705,
      "learning_rate": 5.521262453649729e-05,
      "loss": 0.8246,
      "step": 6793
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.0585205554962158,
      "learning_rate": 5.518575239280601e-05,
      "loss": 0.8602,
      "step": 6794
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.0165480375289917,
      "learning_rate": 5.515888429782825e-05,
      "loss": 0.9377,
      "step": 6795
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9434266090393066,
      "learning_rate": 5.5132020253991376e-05,
      "loss": 0.7989,
      "step": 6796
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9420387744903564,
      "learning_rate": 5.5105160263722454e-05,
      "loss": 0.8159,
      "step": 6797
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9444668292999268,
      "learning_rate": 5.507830432944813e-05,
      "loss": 0.954,
      "step": 6798
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9778406023979187,
      "learning_rate": 5.505145245359461e-05,
      "loss": 0.8933,
      "step": 6799
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.0137708187103271,
      "learning_rate": 5.502460463858791e-05,
      "loss": 0.8036,
      "step": 6800
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.0338083505630493,
      "learning_rate": 5.499776088685352e-05,
      "loss": 0.9418,
      "step": 6801
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9403624534606934,
      "learning_rate": 5.4970921200816594e-05,
      "loss": 0.8651,
      "step": 6802
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9193729758262634,
      "learning_rate": 5.494408558290203e-05,
      "loss": 0.8052,
      "step": 6803
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9242038130760193,
      "learning_rate": 5.491725403553423e-05,
      "loss": 0.717,
      "step": 6804
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9462067484855652,
      "learning_rate": 5.489042656113727e-05,
      "loss": 0.8388,
      "step": 6805
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9006744623184204,
      "learning_rate": 5.4863603162134814e-05,
      "loss": 0.9258,
      "step": 6806
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9400488138198853,
      "learning_rate": 5.483678384095028e-05,
      "loss": 0.8968,
      "step": 6807
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.942817747592926,
      "learning_rate": 5.4809968600006635e-05,
      "loss": 0.7531,
      "step": 6808
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9494886994361877,
      "learning_rate": 5.4783157441726405e-05,
      "loss": 0.7241,
      "step": 6809
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9944396615028381,
      "learning_rate": 5.475635036853197e-05,
      "loss": 0.9379,
      "step": 6810
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.0850656032562256,
      "learning_rate": 5.4729547382845035e-05,
      "loss": 0.7806,
      "step": 6811
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.780096709728241,
      "learning_rate": 5.470274848708716e-05,
      "loss": 0.6456,
      "step": 6812
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.0622731447219849,
      "learning_rate": 5.467595368367955e-05,
      "loss": 0.8307,
      "step": 6813
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9203853011131287,
      "learning_rate": 5.4649162975042876e-05,
      "loss": 0.8609,
      "step": 6814
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9724249839782715,
      "learning_rate": 5.462237636359756e-05,
      "loss": 0.8423,
      "step": 6815
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9355921745300293,
      "learning_rate": 5.459559385176358e-05,
      "loss": 0.7992,
      "step": 6816
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9333254098892212,
      "learning_rate": 5.456881544196064e-05,
      "loss": 0.9283,
      "step": 6817
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9402013421058655,
      "learning_rate": 5.4542041136608016e-05,
      "loss": 0.9386,
      "step": 6818
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.922160804271698,
      "learning_rate": 5.451527093812454e-05,
      "loss": 0.8663,
      "step": 6819
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9373624324798584,
      "learning_rate": 5.448850484892887e-05,
      "loss": 0.7098,
      "step": 6820
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9394688010215759,
      "learning_rate": 5.446174287143904e-05,
      "loss": 0.8482,
      "step": 6821
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.001277208328247,
      "learning_rate": 5.443498500807289e-05,
      "loss": 0.8052,
      "step": 6822
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.0689027309417725,
      "learning_rate": 5.440823126124794e-05,
      "loss": 0.8094,
      "step": 6823
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9812199473381042,
      "learning_rate": 5.438148163338109e-05,
      "loss": 0.9841,
      "step": 6824
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9941795468330383,
      "learning_rate": 5.435473612688911e-05,
      "loss": 0.7297,
      "step": 6825
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9649080038070679,
      "learning_rate": 5.432799474418828e-05,
      "loss": 0.9047,
      "step": 6826
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.106789469718933,
      "learning_rate": 5.4301257487694504e-05,
      "loss": 0.8774,
      "step": 6827
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.1915044784545898,
      "learning_rate": 5.4274524359823386e-05,
      "loss": 0.7992,
      "step": 6828
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.8903177976608276,
      "learning_rate": 5.424779536299011e-05,
      "loss": 0.8283,
      "step": 6829
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9922155141830444,
      "learning_rate": 5.4221070499609475e-05,
      "loss": 0.8335,
      "step": 6830
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.0460870265960693,
      "learning_rate": 5.4194349772095866e-05,
      "loss": 0.9406,
      "step": 6831
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.8839626312255859,
      "learning_rate": 5.416763318286345e-05,
      "loss": 0.8007,
      "step": 6832
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.223207712173462,
      "learning_rate": 5.4140920734325886e-05,
      "loss": 0.8313,
      "step": 6833
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9138821363449097,
      "learning_rate": 5.4114212428896424e-05,
      "loss": 0.7649,
      "step": 6834
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.8357966542243958,
      "learning_rate": 5.408750826898812e-05,
      "loss": 0.7869,
      "step": 6835
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9664004445075989,
      "learning_rate": 5.406080825701347e-05,
      "loss": 0.7762,
      "step": 6836
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.0131416320800781,
      "learning_rate": 5.4034112395384664e-05,
      "loss": 0.8993,
      "step": 6837
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9176822900772095,
      "learning_rate": 5.400742068651358e-05,
      "loss": 0.8585,
      "step": 6838
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9635005593299866,
      "learning_rate": 5.398073313281162e-05,
      "loss": 0.8394,
      "step": 6839
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.1465282440185547,
      "learning_rate": 5.395404973668987e-05,
      "loss": 0.8786,
      "step": 6840
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9912209510803223,
      "learning_rate": 5.392737050055896e-05,
      "loss": 0.9657,
      "step": 6841
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8490775227546692,
      "learning_rate": 5.390069542682932e-05,
      "loss": 0.7544,
      "step": 6842
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.0732125043869019,
      "learning_rate": 5.387402451791084e-05,
      "loss": 0.8172,
      "step": 6843
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8101218938827515,
      "learning_rate": 5.3847357776213015e-05,
      "loss": 0.62,
      "step": 6844
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8600050210952759,
      "learning_rate": 5.382069520414516e-05,
      "loss": 0.82,
      "step": 6845
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.087180495262146,
      "learning_rate": 5.379403680411601e-05,
      "loss": 0.9679,
      "step": 6846
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9968470335006714,
      "learning_rate": 5.3767382578534e-05,
      "loss": 0.8311,
      "step": 6847
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.0656013488769531,
      "learning_rate": 5.374073252980722e-05,
      "loss": 0.8844,
      "step": 6848
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9682269096374512,
      "learning_rate": 5.3714086660343365e-05,
      "loss": 0.776,
      "step": 6849
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8912519812583923,
      "learning_rate": 5.3687444972549695e-05,
      "loss": 0.8103,
      "step": 6850
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9110443592071533,
      "learning_rate": 5.366080746883311e-05,
      "loss": 0.9421,
      "step": 6851
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.122183084487915,
      "learning_rate": 5.3634174151600233e-05,
      "loss": 0.9921,
      "step": 6852
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.0002410411834717,
      "learning_rate": 5.36075450232572e-05,
      "loss": 1.0811,
      "step": 6853
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9634022116661072,
      "learning_rate": 5.358092008620976e-05,
      "loss": 0.8842,
      "step": 6854
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8745232820510864,
      "learning_rate": 5.355429934286347e-05,
      "loss": 0.6739,
      "step": 6855
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9057954549789429,
      "learning_rate": 5.3527682795623146e-05,
      "loss": 0.7893,
      "step": 6856
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.0005015134811401,
      "learning_rate": 5.350107044689357e-05,
      "loss": 0.8161,
      "step": 6857
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.0776358842849731,
      "learning_rate": 5.347446229907909e-05,
      "loss": 0.978,
      "step": 6858
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.008244514465332,
      "learning_rate": 5.3447858354583434e-05,
      "loss": 0.8164,
      "step": 6859
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9960151314735413,
      "learning_rate": 5.342125861581022e-05,
      "loss": 0.7671,
      "step": 6860
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.002142071723938,
      "learning_rate": 5.339466308516259e-05,
      "loss": 0.7916,
      "step": 6861
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.1606249809265137,
      "learning_rate": 5.3368071765043216e-05,
      "loss": 0.7807,
      "step": 6862
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9500653743743896,
      "learning_rate": 5.3341484657854556e-05,
      "loss": 0.8476,
      "step": 6863
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9048900008201599,
      "learning_rate": 5.3314901765998596e-05,
      "loss": 0.7491,
      "step": 6864
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9902215600013733,
      "learning_rate": 5.3288323091876924e-05,
      "loss": 0.8751,
      "step": 6865
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.0511009693145752,
      "learning_rate": 5.326174863789074e-05,
      "loss": 0.9353,
      "step": 6866
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.074035882949829,
      "learning_rate": 5.323517840644097e-05,
      "loss": 0.8742,
      "step": 6867
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.0862635374069214,
      "learning_rate": 5.320861239992806e-05,
      "loss": 0.9625,
      "step": 6868
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.8584052920341492,
      "learning_rate": 5.318205062075204e-05,
      "loss": 0.7275,
      "step": 6869
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9199810028076172,
      "learning_rate": 5.315549307131271e-05,
      "loss": 0.8683,
      "step": 6870
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9773145318031311,
      "learning_rate": 5.312893975400934e-05,
      "loss": 0.8401,
      "step": 6871
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.8908659815788269,
      "learning_rate": 5.3102390671240855e-05,
      "loss": 0.7133,
      "step": 6872
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.8693983554840088,
      "learning_rate": 5.307584582540586e-05,
      "loss": 0.8739,
      "step": 6873
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.8846648931503296,
      "learning_rate": 5.304930521890252e-05,
      "loss": 0.8575,
      "step": 6874
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9141742587089539,
      "learning_rate": 5.3022768854128626e-05,
      "loss": 0.7828,
      "step": 6875
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.0899076461791992,
      "learning_rate": 5.299623673348153e-05,
      "loss": 0.8981,
      "step": 6876
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9468065500259399,
      "learning_rate": 5.296970885935836e-05,
      "loss": 0.9258,
      "step": 6877
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9790927767753601,
      "learning_rate": 5.29431852341557e-05,
      "loss": 0.9008,
      "step": 6878
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9928053617477417,
      "learning_rate": 5.291666586026979e-05,
      "loss": 0.8482,
      "step": 6879
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.0628911256790161,
      "learning_rate": 5.289015074009657e-05,
      "loss": 0.8192,
      "step": 6880
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.8648643493652344,
      "learning_rate": 5.286363987603149e-05,
      "loss": 0.8641,
      "step": 6881
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.8695665597915649,
      "learning_rate": 5.283713327046962e-05,
      "loss": 0.8806,
      "step": 6882
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9351721405982971,
      "learning_rate": 5.281063092580578e-05,
      "loss": 0.8487,
      "step": 6883
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.8205286264419556,
      "learning_rate": 5.278413284443424e-05,
      "loss": 0.7916,
      "step": 6884
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.953237771987915,
      "learning_rate": 5.275763902874895e-05,
      "loss": 0.8525,
      "step": 6885
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9503474235534668,
      "learning_rate": 5.273114948114346e-05,
      "loss": 0.961,
      "step": 6886
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9329468607902527,
      "learning_rate": 5.270466420401102e-05,
      "loss": 0.8279,
      "step": 6887
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.0341823101043701,
      "learning_rate": 5.267818319974438e-05,
      "loss": 0.8508,
      "step": 6888
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.8884573578834534,
      "learning_rate": 5.265170647073592e-05,
      "loss": 0.8582,
      "step": 6889
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9818419218063354,
      "learning_rate": 5.262523401937778e-05,
      "loss": 0.8337,
      "step": 6890
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.0287188291549683,
      "learning_rate": 5.2598765848061414e-05,
      "loss": 0.8217,
      "step": 6891
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9464033246040344,
      "learning_rate": 5.257230195917817e-05,
      "loss": 0.7938,
      "step": 6892
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9691477417945862,
      "learning_rate": 5.2545842355119e-05,
      "loss": 0.9423,
      "step": 6893
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.963091254234314,
      "learning_rate": 5.25193870382742e-05,
      "loss": 0.8896,
      "step": 6894
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9363139271736145,
      "learning_rate": 5.2492936011034e-05,
      "loss": 0.8423,
      "step": 6895
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9486441612243652,
      "learning_rate": 5.246648927578803e-05,
      "loss": 0.8229,
      "step": 6896
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9020103216171265,
      "learning_rate": 5.2440046834925586e-05,
      "loss": 0.8179,
      "step": 6897
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.1060616970062256,
      "learning_rate": 5.241360869083567e-05,
      "loss": 0.8072,
      "step": 6898
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.1523675918579102,
      "learning_rate": 5.2387174845906774e-05,
      "loss": 1.0646,
      "step": 6899
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.8641512393951416,
      "learning_rate": 5.236074530252704e-05,
      "loss": 0.7779,
      "step": 6900
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9608871340751648,
      "learning_rate": 5.233432006308421e-05,
      "loss": 0.8586,
      "step": 6901
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9036332964897156,
      "learning_rate": 5.23078991299657e-05,
      "loss": 0.9235,
      "step": 6902
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9491810202598572,
      "learning_rate": 5.228148250555849e-05,
      "loss": 0.8345,
      "step": 6903
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9753369092941284,
      "learning_rate": 5.225507019224911e-05,
      "loss": 0.8223,
      "step": 6904
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.8814740777015686,
      "learning_rate": 5.2228662192423835e-05,
      "loss": 0.7743,
      "step": 6905
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9747453331947327,
      "learning_rate": 5.220225850846847e-05,
      "loss": 0.929,
      "step": 6906
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9890178442001343,
      "learning_rate": 5.217585914276837e-05,
      "loss": 0.8082,
      "step": 6907
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9679883122444153,
      "learning_rate": 5.2149464097708646e-05,
      "loss": 0.8502,
      "step": 6908
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9747132658958435,
      "learning_rate": 5.212307337567393e-05,
      "loss": 0.7888,
      "step": 6909
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.1055711507797241,
      "learning_rate": 5.209668697904844e-05,
      "loss": 0.9157,
      "step": 6910
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9472057223320007,
      "learning_rate": 5.207030491021603e-05,
      "loss": 0.9064,
      "step": 6911
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.8460851311683655,
      "learning_rate": 5.2043927171560213e-05,
      "loss": 0.8386,
      "step": 6912
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.0734587907791138,
      "learning_rate": 5.201755376546407e-05,
      "loss": 0.8369,
      "step": 6913
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9430543184280396,
      "learning_rate": 5.199118469431021e-05,
      "loss": 0.942,
      "step": 6914
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.943413257598877,
      "learning_rate": 5.1964819960481036e-05,
      "loss": 0.9508,
      "step": 6915
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.047995924949646,
      "learning_rate": 5.193845956635841e-05,
      "loss": 0.9373,
      "step": 6916
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9492095708847046,
      "learning_rate": 5.1912103514323795e-05,
      "loss": 1.0152,
      "step": 6917
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.0304605960845947,
      "learning_rate": 5.1885751806758384e-05,
      "loss": 0.9368,
      "step": 6918
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.042935848236084,
      "learning_rate": 5.185940444604289e-05,
      "loss": 0.8534,
      "step": 6919
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.8989212512969971,
      "learning_rate": 5.183306143455764e-05,
      "loss": 0.7906,
      "step": 6920
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9429810643196106,
      "learning_rate": 5.18067227746825e-05,
      "loss": 0.8224,
      "step": 6921
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.0301026105880737,
      "learning_rate": 5.178038846879716e-05,
      "loss": 0.8341,
      "step": 6922
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.8430018424987793,
      "learning_rate": 5.175405851928068e-05,
      "loss": 0.7489,
      "step": 6923
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.05275559425354,
      "learning_rate": 5.172773292851181e-05,
      "loss": 0.8483,
      "step": 6924
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9597530364990234,
      "learning_rate": 5.170141169886904e-05,
      "loss": 0.7273,
      "step": 6925
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9168379306793213,
      "learning_rate": 5.1675094832730165e-05,
      "loss": 0.7783,
      "step": 6926
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.095482587814331,
      "learning_rate": 5.1648782332472876e-05,
      "loss": 0.8039,
      "step": 6927
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.8899198770523071,
      "learning_rate": 5.16224742004744e-05,
      "loss": 0.821,
      "step": 6928
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9685090184211731,
      "learning_rate": 5.1596170439111404e-05,
      "loss": 0.8597,
      "step": 6929
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9671900868415833,
      "learning_rate": 5.156987105076036e-05,
      "loss": 0.8965,
      "step": 6930
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9450376033782959,
      "learning_rate": 5.154357603779728e-05,
      "loss": 0.8372,
      "step": 6931
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9386870265007019,
      "learning_rate": 5.15172854025977e-05,
      "loss": 0.7356,
      "step": 6932
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9371140599250793,
      "learning_rate": 5.149099914753692e-05,
      "loss": 1.013,
      "step": 6933
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9186777472496033,
      "learning_rate": 5.14647172749897e-05,
      "loss": 0.7554,
      "step": 6934
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.894643247127533,
      "learning_rate": 5.143843978733047e-05,
      "loss": 0.8664,
      "step": 6935
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9144344329833984,
      "learning_rate": 5.141216668693322e-05,
      "loss": 0.8347,
      "step": 6936
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9122389554977417,
      "learning_rate": 5.1385897976171594e-05,
      "loss": 0.7896,
      "step": 6937
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9139106273651123,
      "learning_rate": 5.1359633657418915e-05,
      "loss": 0.9626,
      "step": 6938
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.8711784482002258,
      "learning_rate": 5.133337373304787e-05,
      "loss": 0.6826,
      "step": 6939
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9171251654624939,
      "learning_rate": 5.130711820543098e-05,
      "loss": 0.6723,
      "step": 6940
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.0609540939331055,
      "learning_rate": 5.128086707694027e-05,
      "loss": 0.7173,
      "step": 6941
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9223871827125549,
      "learning_rate": 5.125462034994732e-05,
      "loss": 0.7379,
      "step": 6942
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.0631743669509888,
      "learning_rate": 5.1228378026823486e-05,
      "loss": 0.902,
      "step": 6943
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9678778052330017,
      "learning_rate": 5.120214010993954e-05,
      "loss": 0.7699,
      "step": 6944
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9355421662330627,
      "learning_rate": 5.1175906601665956e-05,
      "loss": 0.9389,
      "step": 6945
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.891808271408081,
      "learning_rate": 5.114967750437272e-05,
      "loss": 0.8008,
      "step": 6946
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0875853300094604,
      "learning_rate": 5.112345282042956e-05,
      "loss": 0.7034,
      "step": 6947
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0659302473068237,
      "learning_rate": 5.109723255220571e-05,
      "loss": 0.8083,
      "step": 6948
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9340061545372009,
      "learning_rate": 5.107101670206998e-05,
      "loss": 0.8931,
      "step": 6949
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0546252727508545,
      "learning_rate": 5.1044805272390885e-05,
      "loss": 0.752,
      "step": 6950
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0769869089126587,
      "learning_rate": 5.1018598265536454e-05,
      "loss": 0.7231,
      "step": 6951
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9446578621864319,
      "learning_rate": 5.0992395683874306e-05,
      "loss": 0.8378,
      "step": 6952
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.8647546172142029,
      "learning_rate": 5.0966197529771754e-05,
      "loss": 0.8491,
      "step": 6953
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.8613917231559753,
      "learning_rate": 5.0940003805595626e-05,
      "loss": 0.8099,
      "step": 6954
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.8530964255332947,
      "learning_rate": 5.0913814513712376e-05,
      "loss": 0.908,
      "step": 6955
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.8596360683441162,
      "learning_rate": 5.088762965648804e-05,
      "loss": 0.8732,
      "step": 6956
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9816997647285461,
      "learning_rate": 5.0861449236288306e-05,
      "loss": 0.9049,
      "step": 6957
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9223150610923767,
      "learning_rate": 5.0835273255478436e-05,
      "loss": 0.8334,
      "step": 6958
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0287200212478638,
      "learning_rate": 5.080910171642319e-05,
      "loss": 0.9553,
      "step": 6959
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.1011673212051392,
      "learning_rate": 5.078293462148718e-05,
      "loss": 0.9335,
      "step": 6960
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9374191164970398,
      "learning_rate": 5.075677197303428e-05,
      "loss": 0.882,
      "step": 6961
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9520727396011353,
      "learning_rate": 5.0730613773428224e-05,
      "loss": 0.8506,
      "step": 6962
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0065456628799438,
      "learning_rate": 5.070446002503233e-05,
      "loss": 0.8181,
      "step": 6963
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.8654084801673889,
      "learning_rate": 5.0678310730209275e-05,
      "loss": 0.7959,
      "step": 6964
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.986094057559967,
      "learning_rate": 5.0652165891321624e-05,
      "loss": 0.9663,
      "step": 6965
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0231149196624756,
      "learning_rate": 5.0626025510731344e-05,
      "loss": 0.8941,
      "step": 6966
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9566606283187866,
      "learning_rate": 5.0599889590800154e-05,
      "loss": 0.894,
      "step": 6967
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0626215934753418,
      "learning_rate": 5.057375813388923e-05,
      "loss": 0.7947,
      "step": 6968
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9994633197784424,
      "learning_rate": 5.054763114235935e-05,
      "loss": 0.8555,
      "step": 6969
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0460691452026367,
      "learning_rate": 5.05215086185711e-05,
      "loss": 0.7892,
      "step": 6970
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9017778038978577,
      "learning_rate": 5.0495390564884304e-05,
      "loss": 0.7728,
      "step": 6971
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0599709749221802,
      "learning_rate": 5.0469276983658686e-05,
      "loss": 0.7596,
      "step": 6972
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8913029432296753,
      "learning_rate": 5.0443167877253526e-05,
      "loss": 0.7783,
      "step": 6973
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9339177012443542,
      "learning_rate": 5.0417063248027484e-05,
      "loss": 0.8226,
      "step": 6974
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9519619345664978,
      "learning_rate": 5.0390963098339087e-05,
      "loss": 0.9309,
      "step": 6975
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.1157145500183105,
      "learning_rate": 5.036486743054627e-05,
      "loss": 0.826,
      "step": 6976
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0372145175933838,
      "learning_rate": 5.033877624700661e-05,
      "loss": 0.7927,
      "step": 6977
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9890812039375305,
      "learning_rate": 5.0312689550077386e-05,
      "loss": 0.8622,
      "step": 6978
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8893467783927917,
      "learning_rate": 5.028660734211532e-05,
      "loss": 0.7823,
      "step": 6979
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9978625774383545,
      "learning_rate": 5.026052962547681e-05,
      "loss": 0.8506,
      "step": 6980
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.1442023515701294,
      "learning_rate": 5.0234456402517796e-05,
      "loss": 0.9455,
      "step": 6981
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9040158987045288,
      "learning_rate": 5.02083876755939e-05,
      "loss": 0.8085,
      "step": 6982
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8826153874397278,
      "learning_rate": 5.018232344706026e-05,
      "loss": 0.6778,
      "step": 6983
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0761096477508545,
      "learning_rate": 5.015626371927159e-05,
      "loss": 0.8416,
      "step": 6984
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8360695242881775,
      "learning_rate": 5.013020849458232e-05,
      "loss": 0.7159,
      "step": 6985
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9459480047225952,
      "learning_rate": 5.010415777534635e-05,
      "loss": 0.8532,
      "step": 6986
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9853194952011108,
      "learning_rate": 5.007811156391717e-05,
      "loss": 1.0038,
      "step": 6987
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0642902851104736,
      "learning_rate": 5.005206986264801e-05,
      "loss": 0.7724,
      "step": 6988
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0813014507293701,
      "learning_rate": 5.002603267389152e-05,
      "loss": 0.7504,
      "step": 6989
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8983114361763,
      "learning_rate": 5.000000000000002e-05,
      "loss": 0.7516,
      "step": 6990
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9147076606750488,
      "learning_rate": 4.9973971843325396e-05,
      "loss": 0.8647,
      "step": 6991
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9630460143089294,
      "learning_rate": 4.99479482062192e-05,
      "loss": 0.795,
      "step": 6992
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0792957544326782,
      "learning_rate": 4.9921929091032494e-05,
      "loss": 0.7978,
      "step": 6993
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9525182843208313,
      "learning_rate": 4.98959145001159e-05,
      "loss": 0.7397,
      "step": 6994
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.101427435874939,
      "learning_rate": 4.98699044358198e-05,
      "loss": 0.979,
      "step": 6995
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0763685703277588,
      "learning_rate": 4.984389890049399e-05,
      "loss": 0.862,
      "step": 6996
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.1603392362594604,
      "learning_rate": 4.9817897896487896e-05,
      "loss": 0.8487,
      "step": 6997
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0379005670547485,
      "learning_rate": 4.9791901426150634e-05,
      "loss": 0.9381,
      "step": 6998
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.013866662979126,
      "learning_rate": 4.97659094918308e-05,
      "loss": 0.9473,
      "step": 6999
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.0111335515975952,
      "learning_rate": 4.973992209587662e-05,
      "loss": 1.0283,
      "step": 7000
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.8699116110801697,
      "learning_rate": 4.9713939240635865e-05,
      "loss": 0.7466,
      "step": 7001
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9282172918319702,
      "learning_rate": 4.968796092845602e-05,
      "loss": 0.8761,
      "step": 7002
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.8358901739120483,
      "learning_rate": 4.9661987161684045e-05,
      "loss": 0.7657,
      "step": 7003
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.0280259847640991,
      "learning_rate": 4.963601794266647e-05,
      "loss": 0.898,
      "step": 7004
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9688712954521179,
      "learning_rate": 4.961005327374959e-05,
      "loss": 0.9126,
      "step": 7005
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9416438341140747,
      "learning_rate": 4.958409315727902e-05,
      "loss": 0.7019,
      "step": 7006
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9312397837638855,
      "learning_rate": 4.955813759560017e-05,
      "loss": 0.7853,
      "step": 7007
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.0103497505187988,
      "learning_rate": 4.953218659105808e-05,
      "loss": 0.8615,
      "step": 7008
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9424223899841309,
      "learning_rate": 4.95062401459971e-05,
      "loss": 0.844,
      "step": 7009
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9779973030090332,
      "learning_rate": 4.948029826276146e-05,
      "loss": 0.8157,
      "step": 7010
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9257150888442993,
      "learning_rate": 4.945436094369483e-05,
      "loss": 0.8296,
      "step": 7011
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.0849761962890625,
      "learning_rate": 4.9428428191140465e-05,
      "loss": 0.9531,
      "step": 7012
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.8950340151786804,
      "learning_rate": 4.940250000744132e-05,
      "loss": 0.7752,
      "step": 7013
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9071115851402283,
      "learning_rate": 4.937657639493981e-05,
      "loss": 0.6871,
      "step": 7014
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9212459921836853,
      "learning_rate": 4.9350657355978e-05,
      "loss": 0.8703,
      "step": 7015
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.0368958711624146,
      "learning_rate": 4.932474289289748e-05,
      "loss": 0.8566,
      "step": 7016
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.0109087228775024,
      "learning_rate": 4.9298833008039556e-05,
      "loss": 0.7926,
      "step": 7017
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.0654046535491943,
      "learning_rate": 4.9272927703745e-05,
      "loss": 0.8974,
      "step": 7018
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9468387365341187,
      "learning_rate": 4.9247026982354174e-05,
      "loss": 0.8999,
      "step": 7019
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9901147484779358,
      "learning_rate": 4.922113084620713e-05,
      "loss": 0.6839,
      "step": 7020
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.891943097114563,
      "learning_rate": 4.91952392976434e-05,
      "loss": 0.8097,
      "step": 7021
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.8627398610115051,
      "learning_rate": 4.9169352339002114e-05,
      "loss": 0.6818,
      "step": 7022
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.8250307440757751,
      "learning_rate": 4.914346997262209e-05,
      "loss": 0.7183,
      "step": 7023
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.0720813274383545,
      "learning_rate": 4.911759220084159e-05,
      "loss": 0.806,
      "step": 7024
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8944665193557739,
      "learning_rate": 4.909171902599855e-05,
      "loss": 0.981,
      "step": 7025
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.0123556852340698,
      "learning_rate": 4.906585045043041e-05,
      "loss": 0.9313,
      "step": 7026
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.806859016418457,
      "learning_rate": 4.903998647647433e-05,
      "loss": 0.7039,
      "step": 7027
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9393472075462341,
      "learning_rate": 4.9014127106466935e-05,
      "loss": 0.8605,
      "step": 7028
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9421588778495789,
      "learning_rate": 4.8988272342744445e-05,
      "loss": 0.7877,
      "step": 7029
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.1050317287445068,
      "learning_rate": 4.896242218764275e-05,
      "loss": 0.9252,
      "step": 7030
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9901186227798462,
      "learning_rate": 4.893657664349725e-05,
      "loss": 0.882,
      "step": 7031
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9524365067481995,
      "learning_rate": 4.891073571264288e-05,
      "loss": 0.9751,
      "step": 7032
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9938132166862488,
      "learning_rate": 4.888489939741431e-05,
      "loss": 0.7933,
      "step": 7033
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.040974497795105,
      "learning_rate": 4.885906770014567e-05,
      "loss": 0.8743,
      "step": 7034
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9026747345924377,
      "learning_rate": 4.8833240623170706e-05,
      "loss": 0.9481,
      "step": 7035
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9507520198822021,
      "learning_rate": 4.8807418168822705e-05,
      "loss": 0.6265,
      "step": 7036
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.0344043970108032,
      "learning_rate": 4.878160033943465e-05,
      "loss": 0.8159,
      "step": 7037
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.0946036577224731,
      "learning_rate": 4.875578713733903e-05,
      "loss": 0.744,
      "step": 7038
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.1028919219970703,
      "learning_rate": 4.872997856486784e-05,
      "loss": 0.9262,
      "step": 7039
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.034663438796997,
      "learning_rate": 4.870417462435288e-05,
      "loss": 0.8584,
      "step": 7040
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.0321526527404785,
      "learning_rate": 4.867837531812522e-05,
      "loss": 0.8321,
      "step": 7041
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9382997751235962,
      "learning_rate": 4.865258064851579e-05,
      "loss": 0.9163,
      "step": 7042
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8348347544670105,
      "learning_rate": 4.862679061785502e-05,
      "loss": 0.609,
      "step": 7043
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9279749393463135,
      "learning_rate": 4.8601005228472783e-05,
      "loss": 0.8091,
      "step": 7044
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9362346529960632,
      "learning_rate": 4.857522448269874e-05,
      "loss": 0.6418,
      "step": 7045
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.000012993812561,
      "learning_rate": 4.854944838286199e-05,
      "loss": 0.819,
      "step": 7046
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9438902139663696,
      "learning_rate": 4.852367693129125e-05,
      "loss": 0.8875,
      "step": 7047
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.0057377815246582,
      "learning_rate": 4.849791013031486e-05,
      "loss": 0.8715,
      "step": 7048
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.0180364847183228,
      "learning_rate": 4.8472147982260695e-05,
      "loss": 0.9626,
      "step": 7049
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8775011897087097,
      "learning_rate": 4.84463904894562e-05,
      "loss": 0.7978,
      "step": 7050
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9489887952804565,
      "learning_rate": 4.84206376542284e-05,
      "loss": 0.915,
      "step": 7051
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.0759973526000977,
      "learning_rate": 4.839488947890398e-05,
      "loss": 0.7829,
      "step": 7052
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.952645480632782,
      "learning_rate": 4.8369145965809105e-05,
      "loss": 0.8563,
      "step": 7053
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.026216983795166,
      "learning_rate": 4.8343407117269524e-05,
      "loss": 0.9297,
      "step": 7054
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9259018898010254,
      "learning_rate": 4.831767293561067e-05,
      "loss": 0.7027,
      "step": 7055
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.03157377243042,
      "learning_rate": 4.8291943423157435e-05,
      "loss": 0.872,
      "step": 7056
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.8271607160568237,
      "learning_rate": 4.826621858223431e-05,
      "loss": 0.619,
      "step": 7057
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.0726627111434937,
      "learning_rate": 4.8240498415165455e-05,
      "loss": 0.8177,
      "step": 7058
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9494621753692627,
      "learning_rate": 4.8214782924274514e-05,
      "loss": 0.8342,
      "step": 7059
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.1226849555969238,
      "learning_rate": 4.8189072111884723e-05,
      "loss": 0.8907,
      "step": 7060
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.039818525314331,
      "learning_rate": 4.816336598031888e-05,
      "loss": 0.9034,
      "step": 7061
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9842699766159058,
      "learning_rate": 4.813766453189946e-05,
      "loss": 0.7722,
      "step": 7062
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9690228700637817,
      "learning_rate": 4.8111967768948394e-05,
      "loss": 0.9774,
      "step": 7063
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.8920415639877319,
      "learning_rate": 4.808627569378723e-05,
      "loss": 0.8333,
      "step": 7064
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9937723875045776,
      "learning_rate": 4.806058830873715e-05,
      "loss": 0.9051,
      "step": 7065
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.8087983727455139,
      "learning_rate": 4.803490561611884e-05,
      "loss": 0.773,
      "step": 7066
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.928328812122345,
      "learning_rate": 4.800922761825254e-05,
      "loss": 0.8242,
      "step": 7067
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9896306991577148,
      "learning_rate": 4.7983554317458204e-05,
      "loss": 0.9074,
      "step": 7068
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9831765294075012,
      "learning_rate": 4.795788571605521e-05,
      "loss": 0.9084,
      "step": 7069
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.8851191401481628,
      "learning_rate": 4.793222181636257e-05,
      "loss": 0.8524,
      "step": 7070
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9489747881889343,
      "learning_rate": 4.790656262069886e-05,
      "loss": 0.6502,
      "step": 7071
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.081588625907898,
      "learning_rate": 4.7880908131382284e-05,
      "loss": 0.9203,
      "step": 7072
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9752261638641357,
      "learning_rate": 4.785525835073058e-05,
      "loss": 0.8661,
      "step": 7073
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.8664730191230774,
      "learning_rate": 4.782961328106099e-05,
      "loss": 0.7679,
      "step": 7074
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.0531363487243652,
      "learning_rate": 4.780397292469054e-05,
      "loss": 0.8977,
      "step": 7075
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.06095552444458,
      "learning_rate": 4.777833728393552e-05,
      "loss": 0.8605,
      "step": 7076
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9714718461036682,
      "learning_rate": 4.775270636111206e-05,
      "loss": 0.8294,
      "step": 7077
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9399491548538208,
      "learning_rate": 4.772708015853581e-05,
      "loss": 0.7834,
      "step": 7078
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.8987073302268982,
      "learning_rate": 4.770145867852184e-05,
      "loss": 0.8535,
      "step": 7079
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.8827242255210876,
      "learning_rate": 4.7675841923385e-05,
      "loss": 0.6656,
      "step": 7080
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9393911957740784,
      "learning_rate": 4.765022989543958e-05,
      "loss": 0.9842,
      "step": 7081
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9578714966773987,
      "learning_rate": 4.7624622596999445e-05,
      "loss": 0.8467,
      "step": 7082
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.8853945136070251,
      "learning_rate": 4.759902003037815e-05,
      "loss": 0.7705,
      "step": 7083
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.0448302030563354,
      "learning_rate": 4.7573422197888704e-05,
      "loss": 0.8896,
      "step": 7084
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9766141176223755,
      "learning_rate": 4.754782910184373e-05,
      "loss": 0.9162,
      "step": 7085
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.0868399143218994,
      "learning_rate": 4.752224074455537e-05,
      "loss": 0.8805,
      "step": 7086
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.813254714012146,
      "learning_rate": 4.7496657128335464e-05,
      "loss": 0.8746,
      "step": 7087
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9296266436576843,
      "learning_rate": 4.747107825549532e-05,
      "loss": 0.8671,
      "step": 7088
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.890579342842102,
      "learning_rate": 4.7445504128345796e-05,
      "loss": 0.7531,
      "step": 7089
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9760721325874329,
      "learning_rate": 4.7419934749197446e-05,
      "loss": 0.7774,
      "step": 7090
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.910007894039154,
      "learning_rate": 4.7394370120360284e-05,
      "loss": 0.6781,
      "step": 7091
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.0647732019424438,
      "learning_rate": 4.7368810244143905e-05,
      "loss": 0.7831,
      "step": 7092
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9459065198898315,
      "learning_rate": 4.734325512285754e-05,
      "loss": 0.8581,
      "step": 7093
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9992945194244385,
      "learning_rate": 4.7317704758809946e-05,
      "loss": 0.8799,
      "step": 7094
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.8945961594581604,
      "learning_rate": 4.729215915430944e-05,
      "loss": 0.861,
      "step": 7095
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.0408872365951538,
      "learning_rate": 4.726661831166389e-05,
      "loss": 0.923,
      "step": 7096
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.0040481090545654,
      "learning_rate": 4.724108223318082e-05,
      "loss": 0.8587,
      "step": 7097
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9879751205444336,
      "learning_rate": 4.7215550921167275e-05,
      "loss": 0.6756,
      "step": 7098
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.0201168060302734,
      "learning_rate": 4.719002437792979e-05,
      "loss": 1.0533,
      "step": 7099
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9055292010307312,
      "learning_rate": 4.716450260577462e-05,
      "loss": 0.8501,
      "step": 7100
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.00068199634552,
      "learning_rate": 4.71389856070075e-05,
      "loss": 0.9672,
      "step": 7101
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9292241334915161,
      "learning_rate": 4.711347338393369e-05,
      "loss": 0.6878,
      "step": 7102
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.0443757772445679,
      "learning_rate": 4.708796593885816e-05,
      "loss": 0.8234,
      "step": 7103
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.0028619766235352,
      "learning_rate": 4.706246327408531e-05,
      "loss": 0.7903,
      "step": 7104
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.8576123118400574,
      "learning_rate": 4.703696539191917e-05,
      "loss": 0.6617,
      "step": 7105
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9056089520454407,
      "learning_rate": 4.7011472294663307e-05,
      "loss": 0.7582,
      "step": 7106
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.0198500156402588,
      "learning_rate": 4.698598398462093e-05,
      "loss": 0.8763,
      "step": 7107
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.89089035987854,
      "learning_rate": 4.696050046409474e-05,
      "loss": 0.9343,
      "step": 7108
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.060715913772583,
      "learning_rate": 4.6935021735386986e-05,
      "loss": 0.8344,
      "step": 7109
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.0227528810501099,
      "learning_rate": 4.6909547800799637e-05,
      "loss": 0.8058,
      "step": 7110
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.8945808410644531,
      "learning_rate": 4.688407866263397e-05,
      "loss": 0.6519,
      "step": 7111
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9560437202453613,
      "learning_rate": 4.685861432319106e-05,
      "loss": 0.7787,
      "step": 7112
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9864040613174438,
      "learning_rate": 4.6833154784771524e-05,
      "loss": 0.76,
      "step": 7113
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9734664559364319,
      "learning_rate": 4.680770004967536e-05,
      "loss": 0.8526,
      "step": 7114
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.01016104221344,
      "learning_rate": 4.678225012020236e-05,
      "loss": 0.8864,
      "step": 7115
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9251781702041626,
      "learning_rate": 4.675680499865175e-05,
      "loss": 1.0329,
      "step": 7116
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.1172655820846558,
      "learning_rate": 4.673136468732229e-05,
      "loss": 1.0584,
      "step": 7117
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9011834263801575,
      "learning_rate": 4.670592918851248e-05,
      "loss": 0.9043,
      "step": 7118
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.0175445079803467,
      "learning_rate": 4.668049850452016e-05,
      "loss": 0.9005,
      "step": 7119
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.849280059337616,
      "learning_rate": 4.665507263764299e-05,
      "loss": 0.7646,
      "step": 7120
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.8947120904922485,
      "learning_rate": 4.66296515901779e-05,
      "loss": 0.8824,
      "step": 7121
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.8137638568878174,
      "learning_rate": 4.660423536442161e-05,
      "loss": 0.7603,
      "step": 7122
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9065663814544678,
      "learning_rate": 4.657882396267039e-05,
      "loss": 0.8399,
      "step": 7123
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9731753468513489,
      "learning_rate": 4.6553417387219886e-05,
      "loss": 0.6504,
      "step": 7124
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.0348753929138184,
      "learning_rate": 4.6528015640365555e-05,
      "loss": 0.9482,
      "step": 7125
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.8973771333694458,
      "learning_rate": 4.6502618724402235e-05,
      "loss": 0.74,
      "step": 7126
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9787579774856567,
      "learning_rate": 4.647722664162438e-05,
      "loss": 0.9321,
      "step": 7127
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.0852323770523071,
      "learning_rate": 4.64518393943261e-05,
      "loss": 0.8421,
      "step": 7128
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.910737931728363,
      "learning_rate": 4.642645698480094e-05,
      "loss": 0.8428,
      "step": 7129
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9322640895843506,
      "learning_rate": 4.640107941534205e-05,
      "loss": 0.7745,
      "step": 7130
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9430921077728271,
      "learning_rate": 4.637570668824213e-05,
      "loss": 0.8232,
      "step": 7131
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9245215654373169,
      "learning_rate": 4.635033880579353e-05,
      "loss": 0.8706,
      "step": 7132
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9775652289390564,
      "learning_rate": 4.6324975770288035e-05,
      "loss": 0.8884,
      "step": 7133
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9350488185882568,
      "learning_rate": 4.629961758401705e-05,
      "loss": 0.68,
      "step": 7134
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9310065507888794,
      "learning_rate": 4.6274264249271605e-05,
      "loss": 0.8696,
      "step": 7135
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9363387823104858,
      "learning_rate": 4.624891576834218e-05,
      "loss": 0.8119,
      "step": 7136
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9280242919921875,
      "learning_rate": 4.6223572143518856e-05,
      "loss": 0.8479,
      "step": 7137
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9066422581672668,
      "learning_rate": 4.619823337709134e-05,
      "loss": 0.7334,
      "step": 7138
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.8370515704154968,
      "learning_rate": 4.617289947134881e-05,
      "loss": 0.7283,
      "step": 7139
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.933133065700531,
      "learning_rate": 4.614757042858003e-05,
      "loss": 0.892,
      "step": 7140
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.978530764579773,
      "learning_rate": 4.612224625107333e-05,
      "loss": 0.7472,
      "step": 7141
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.0474138259887695,
      "learning_rate": 4.609692694111667e-05,
      "loss": 0.7712,
      "step": 7142
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9915986061096191,
      "learning_rate": 4.607161250099745e-05,
      "loss": 0.9486,
      "step": 7143
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.0672756433486938,
      "learning_rate": 4.604630293300266e-05,
      "loss": 0.8289,
      "step": 7144
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9925188422203064,
      "learning_rate": 4.6020998239419e-05,
      "loss": 0.7584,
      "step": 7145
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9255876541137695,
      "learning_rate": 4.5995698422532444e-05,
      "loss": 0.7263,
      "step": 7146
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9482066631317139,
      "learning_rate": 4.597040348462875e-05,
      "loss": 0.6086,
      "step": 7147
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9802579283714294,
      "learning_rate": 4.5945113427993236e-05,
      "loss": 0.8329,
      "step": 7148
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.247625708580017,
      "learning_rate": 4.5919828254910666e-05,
      "loss": 1.0188,
      "step": 7149
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9492368102073669,
      "learning_rate": 4.5894547967665416e-05,
      "loss": 0.716,
      "step": 7150
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9327245354652405,
      "learning_rate": 4.5869272568541364e-05,
      "loss": 0.9275,
      "step": 7151
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9280490875244141,
      "learning_rate": 4.584400205982209e-05,
      "loss": 0.9771,
      "step": 7152
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.978701114654541,
      "learning_rate": 4.58187364437906e-05,
      "loss": 0.836,
      "step": 7153
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.0173463821411133,
      "learning_rate": 4.579347572272946e-05,
      "loss": 1.03,
      "step": 7154
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.1032975912094116,
      "learning_rate": 4.576821989892095e-05,
      "loss": 0.981,
      "step": 7155
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.0038541555404663,
      "learning_rate": 4.574296897464664e-05,
      "loss": 0.8393,
      "step": 7156
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9559672474861145,
      "learning_rate": 4.571772295218787e-05,
      "loss": 0.959,
      "step": 7157
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.924343466758728,
      "learning_rate": 4.5692481833825574e-05,
      "loss": 0.7731,
      "step": 7158
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.0870634317398071,
      "learning_rate": 4.5667245621839974e-05,
      "loss": 0.9086,
      "step": 7159
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8793196082115173,
      "learning_rate": 4.564201431851113e-05,
      "loss": 0.6783,
      "step": 7160
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8712670803070068,
      "learning_rate": 4.5616787926118534e-05,
      "loss": 0.8089,
      "step": 7161
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.0154775381088257,
      "learning_rate": 4.559156644694118e-05,
      "loss": 0.9986,
      "step": 7162
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.0500681400299072,
      "learning_rate": 4.556634988325779e-05,
      "loss": 0.8549,
      "step": 7163
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9806252717971802,
      "learning_rate": 4.554113823734647e-05,
      "loss": 0.9627,
      "step": 7164
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.088459849357605,
      "learning_rate": 4.551593151148498e-05,
      "loss": 0.8966,
      "step": 7165
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.0001275539398193,
      "learning_rate": 4.5490729707950545e-05,
      "loss": 0.9417,
      "step": 7166
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.925566554069519,
      "learning_rate": 4.546553282902009e-05,
      "loss": 0.8322,
      "step": 7167
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9479320049285889,
      "learning_rate": 4.544034087696997e-05,
      "loss": 0.7325,
      "step": 7168
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9212580919265747,
      "learning_rate": 4.54151538540761e-05,
      "loss": 0.7753,
      "step": 7169
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.0211594104766846,
      "learning_rate": 4.538997176261405e-05,
      "loss": 0.7824,
      "step": 7170
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8724222183227539,
      "learning_rate": 4.536479460485885e-05,
      "loss": 0.8371,
      "step": 7171
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9361341595649719,
      "learning_rate": 4.533962238308509e-05,
      "loss": 0.8301,
      "step": 7172
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9695194959640503,
      "learning_rate": 4.5314455099566986e-05,
      "loss": 0.6768,
      "step": 7173
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9724425077438354,
      "learning_rate": 4.5289292756578236e-05,
      "loss": 0.8218,
      "step": 7174
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.0002604722976685,
      "learning_rate": 4.526413535639212e-05,
      "loss": 0.8484,
      "step": 7175
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8950664401054382,
      "learning_rate": 4.5238982901281426e-05,
      "loss": 0.8444,
      "step": 7176
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.1057254076004028,
      "learning_rate": 4.52138353935186e-05,
      "loss": 0.8149,
      "step": 7177
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8854233026504517,
      "learning_rate": 4.518869283537556e-05,
      "loss": 0.7602,
      "step": 7178
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.991965651512146,
      "learning_rate": 4.516355522912373e-05,
      "loss": 0.7852,
      "step": 7179
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9530578851699829,
      "learning_rate": 4.5138422577034244e-05,
      "loss": 0.7967,
      "step": 7180
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9978302121162415,
      "learning_rate": 4.511329488137766e-05,
      "loss": 0.8893,
      "step": 7181
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.013171911239624,
      "learning_rate": 4.508817214442407e-05,
      "loss": 0.9723,
      "step": 7182
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.0836330652236938,
      "learning_rate": 4.5063054368443245e-05,
      "loss": 0.6956,
      "step": 7183
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.0929409265518188,
      "learning_rate": 4.503794155570442e-05,
      "loss": 0.874,
      "step": 7184
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.8272426128387451,
      "learning_rate": 4.5012833708476376e-05,
      "loss": 0.7856,
      "step": 7185
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.0220280885696411,
      "learning_rate": 4.4987730829027444e-05,
      "loss": 0.7893,
      "step": 7186
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.0553988218307495,
      "learning_rate": 4.496263291962558e-05,
      "loss": 0.9688,
      "step": 7187
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.049229621887207,
      "learning_rate": 4.493753998253821e-05,
      "loss": 0.9525,
      "step": 7188
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.934158205986023,
      "learning_rate": 4.4912452020032305e-05,
      "loss": 0.9036,
      "step": 7189
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.8396526575088501,
      "learning_rate": 4.488736903437453e-05,
      "loss": 0.8917,
      "step": 7190
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9803813099861145,
      "learning_rate": 4.486229102783084e-05,
      "loss": 0.8718,
      "step": 7191
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.0434772968292236,
      "learning_rate": 4.4837218002666955e-05,
      "loss": 0.8842,
      "step": 7192
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.029104232788086,
      "learning_rate": 4.4812149961148185e-05,
      "loss": 0.9453,
      "step": 7193
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9319555759429932,
      "learning_rate": 4.478708690553911e-05,
      "loss": 0.8258,
      "step": 7194
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.8635320067405701,
      "learning_rate": 4.476202883810416e-05,
      "loss": 0.7196,
      "step": 7195
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9404241442680359,
      "learning_rate": 4.473697576110715e-05,
      "loss": 0.9073,
      "step": 7196
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.0130743980407715,
      "learning_rate": 4.471192767681144e-05,
      "loss": 0.7698,
      "step": 7197
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.0991601943969727,
      "learning_rate": 4.468688458748006e-05,
      "loss": 0.7257,
      "step": 7198
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9283801317214966,
      "learning_rate": 4.466184649537548e-05,
      "loss": 0.7853,
      "step": 7199
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.0164464712142944,
      "learning_rate": 4.463681340275975e-05,
      "loss": 0.8288,
      "step": 7200
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.000862717628479,
      "learning_rate": 4.4611785311894415e-05,
      "loss": 0.8925,
      "step": 7201
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9698413610458374,
      "learning_rate": 4.458676222504071e-05,
      "loss": 0.7665,
      "step": 7202
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9689275026321411,
      "learning_rate": 4.456174414445929e-05,
      "loss": 0.7414,
      "step": 7203
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.96601402759552,
      "learning_rate": 4.4536731072410354e-05,
      "loss": 0.7313,
      "step": 7204
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9701598882675171,
      "learning_rate": 4.451172301115378e-05,
      "loss": 1.0088,
      "step": 7205
    },
    {
      "epoch": 2.71,
      "eval_loss": 0.9850345849990845,
      "eval_runtime": 17.7905,
      "eval_samples_per_second": 153.397,
      "eval_steps_per_second": 38.391,
      "step": 7205
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9065946936607361,
      "learning_rate": 4.448671996294885e-05,
      "loss": 0.9334,
      "step": 7206
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9392824769020081,
      "learning_rate": 4.446172193005442e-05,
      "loss": 1.0155,
      "step": 7207
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.017695665359497,
      "learning_rate": 4.4436728914729e-05,
      "loss": 0.6905,
      "step": 7208
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.946652352809906,
      "learning_rate": 4.441174091923053e-05,
      "loss": 0.6967,
      "step": 7209
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9378321170806885,
      "learning_rate": 4.4386757945816514e-05,
      "loss": 0.8453,
      "step": 7210
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.8811726570129395,
      "learning_rate": 4.4361779996744e-05,
      "loss": 0.7867,
      "step": 7211
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.0655052661895752,
      "learning_rate": 4.433680707426968e-05,
      "loss": 0.878,
      "step": 7212
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.8657563328742981,
      "learning_rate": 4.431183918064968e-05,
      "loss": 0.6503,
      "step": 7213
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9314761161804199,
      "learning_rate": 4.428687631813967e-05,
      "loss": 0.8032,
      "step": 7214
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.070124626159668,
      "learning_rate": 4.4261918488994966e-05,
      "loss": 0.958,
      "step": 7215
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.0869014263153076,
      "learning_rate": 4.423696569547035e-05,
      "loss": 0.91,
      "step": 7216
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.0924885272979736,
      "learning_rate": 4.4212017939820104e-05,
      "loss": 0.8737,
      "step": 7217
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.0023492574691772,
      "learning_rate": 4.4187075224298215e-05,
      "loss": 0.7669,
      "step": 7218
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9483810663223267,
      "learning_rate": 4.4162137551158054e-05,
      "loss": 0.8305,
      "step": 7219
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.8692176938056946,
      "learning_rate": 4.413720492265262e-05,
      "loss": 0.8534,
      "step": 7220
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9081084728240967,
      "learning_rate": 4.411227734103439e-05,
      "loss": 0.8995,
      "step": 7221
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9187408685684204,
      "learning_rate": 4.40873548085555e-05,
      "loss": 0.9952,
      "step": 7222
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9795107245445251,
      "learning_rate": 4.406243732746753e-05,
      "loss": 0.7968,
      "step": 7223
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9245018362998962,
      "learning_rate": 4.40375249000216e-05,
      "loss": 0.6689,
      "step": 7224
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9088798761367798,
      "learning_rate": 4.4012617528468504e-05,
      "loss": 0.9365,
      "step": 7225
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.932998538017273,
      "learning_rate": 4.398771521505833e-05,
      "loss": 0.9203,
      "step": 7226
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9866093993186951,
      "learning_rate": 4.396281796204096e-05,
      "loss": 0.7832,
      "step": 7227
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.99666428565979,
      "learning_rate": 4.393792577166578e-05,
      "loss": 0.821,
      "step": 7228
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9267418384552002,
      "learning_rate": 4.39130386461815e-05,
      "loss": 0.7672,
      "step": 7229
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.1924302577972412,
      "learning_rate": 4.388815658783666e-05,
      "loss": 0.9809,
      "step": 7230
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.986389696598053,
      "learning_rate": 4.386327959887916e-05,
      "loss": 0.9706,
      "step": 7231
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9177179932594299,
      "learning_rate": 4.383840768155647e-05,
      "loss": 0.5446,
      "step": 7232
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.0766843557357788,
      "learning_rate": 4.3813540838115695e-05,
      "loss": 0.717,
      "step": 7233
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.930751621723175,
      "learning_rate": 4.378867907080338e-05,
      "loss": 0.7828,
      "step": 7234
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8974274396896362,
      "learning_rate": 4.376382238186565e-05,
      "loss": 0.8374,
      "step": 7235
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.0809255838394165,
      "learning_rate": 4.3738970773548116e-05,
      "loss": 0.8765,
      "step": 7236
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.1457940340042114,
      "learning_rate": 4.3714124248096067e-05,
      "loss": 0.8507,
      "step": 7237
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9598510265350342,
      "learning_rate": 4.368928280775419e-05,
      "loss": 0.8343,
      "step": 7238
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9591374397277832,
      "learning_rate": 4.366444645476675e-05,
      "loss": 0.8238,
      "step": 7239
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9190354943275452,
      "learning_rate": 4.363961519137764e-05,
      "loss": 0.9266,
      "step": 7240
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9613643288612366,
      "learning_rate": 4.3614789019830184e-05,
      "loss": 0.7677,
      "step": 7241
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9642369747161865,
      "learning_rate": 4.358996794236725e-05,
      "loss": 0.8393,
      "step": 7242
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8974925875663757,
      "learning_rate": 4.3565151961231366e-05,
      "loss": 0.7298,
      "step": 7243
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4597151279449463,
      "learning_rate": 4.354034107866446e-05,
      "loss": 0.8783,
      "step": 7244
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8994614481925964,
      "learning_rate": 4.351553529690806e-05,
      "loss": 0.8828,
      "step": 7245
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.1431841850280762,
      "learning_rate": 4.34907346182032e-05,
      "loss": 0.8983,
      "step": 7246
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9401236772537231,
      "learning_rate": 4.346593904479055e-05,
      "loss": 0.8528,
      "step": 7247
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.019606113433838,
      "learning_rate": 4.3441148578910215e-05,
      "loss": 0.8521,
      "step": 7248
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.958446204662323,
      "learning_rate": 4.3416363222801824e-05,
      "loss": 0.8572,
      "step": 7249
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.911415159702301,
      "learning_rate": 4.339158297870469e-05,
      "loss": 0.7971,
      "step": 7250
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.92835932970047,
      "learning_rate": 4.33668078488575e-05,
      "loss": 0.94,
      "step": 7251
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9959993958473206,
      "learning_rate": 4.334203783549855e-05,
      "loss": 0.7044,
      "step": 7252
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8898856043815613,
      "learning_rate": 4.331727294086571e-05,
      "loss": 0.8639,
      "step": 7253
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.0366259813308716,
      "learning_rate": 4.329251316719632e-05,
      "loss": 1.0,
      "step": 7254
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9531504511833191,
      "learning_rate": 4.32677585167273e-05,
      "loss": 0.7704,
      "step": 7255
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9575711488723755,
      "learning_rate": 4.3243008991695036e-05,
      "loss": 0.7923,
      "step": 7256
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.025300145149231,
      "learning_rate": 4.321826459433559e-05,
      "loss": 0.8113,
      "step": 7257
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9057744145393372,
      "learning_rate": 4.3193525326884435e-05,
      "loss": 0.7566,
      "step": 7258
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.0421050786972046,
      "learning_rate": 4.3168791191576594e-05,
      "loss": 0.8432,
      "step": 7259
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.0239280462265015,
      "learning_rate": 4.314406219064677e-05,
      "loss": 0.901,
      "step": 7260
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.0207209587097168,
      "learning_rate": 4.311933832632894e-05,
      "loss": 0.854,
      "step": 7261
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9745489358901978,
      "learning_rate": 4.3094619600856825e-05,
      "loss": 0.7285,
      "step": 7262
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.0606648921966553,
      "learning_rate": 4.3069906016463714e-05,
      "loss": 0.9441,
      "step": 7263
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9959872364997864,
      "learning_rate": 4.304519757538218e-05,
      "loss": 0.9481,
      "step": 7264
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9500430226325989,
      "learning_rate": 4.3020494279844605e-05,
      "loss": 0.9082,
      "step": 7265
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9639178514480591,
      "learning_rate": 4.299579613208274e-05,
      "loss": 0.7819,
      "step": 7266
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.1266287565231323,
      "learning_rate": 4.297110313432792e-05,
      "loss": 0.7925,
      "step": 7267
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9916796088218689,
      "learning_rate": 4.294641528881107e-05,
      "loss": 0.8896,
      "step": 7268
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9956557154655457,
      "learning_rate": 4.2921732597762556e-05,
      "loss": 0.8826,
      "step": 7269
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8546841144561768,
      "learning_rate": 4.2897055063412325e-05,
      "loss": 0.6859,
      "step": 7270
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.880997896194458,
      "learning_rate": 4.287238268798982e-05,
      "loss": 0.768,
      "step": 7271
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8876153826713562,
      "learning_rate": 4.284771547372408e-05,
      "loss": 0.7929,
      "step": 7272
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.0851702690124512,
      "learning_rate": 4.282305342284373e-05,
      "loss": 0.8387,
      "step": 7273
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.268347978591919,
      "learning_rate": 4.279839653757669e-05,
      "loss": 0.7774,
      "step": 7274
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.199522852897644,
      "learning_rate": 4.2773744820150683e-05,
      "loss": 0.7802,
      "step": 7275
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.06954824924469,
      "learning_rate": 4.274909827279283e-05,
      "loss": 0.7562,
      "step": 7276
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.1091365814208984,
      "learning_rate": 4.2724456897729746e-05,
      "loss": 0.7867,
      "step": 7277
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8760621547698975,
      "learning_rate": 4.269982069718773e-05,
      "loss": 0.7952,
      "step": 7278
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.963731586933136,
      "learning_rate": 4.267518967339248e-05,
      "loss": 0.88,
      "step": 7279
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8323071002960205,
      "learning_rate": 4.265056382856927e-05,
      "loss": 0.7231,
      "step": 7280
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8988060355186462,
      "learning_rate": 4.262594316494286e-05,
      "loss": 0.6817,
      "step": 7281
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9780885577201843,
      "learning_rate": 4.260132768473769e-05,
      "loss": 0.8641,
      "step": 7282
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8766704201698303,
      "learning_rate": 4.257671739017756e-05,
      "loss": 0.8008,
      "step": 7283
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8053950667381287,
      "learning_rate": 4.255211228348584e-05,
      "loss": 0.7223,
      "step": 7284
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.0516893863677979,
      "learning_rate": 4.252751236688556e-05,
      "loss": 0.751,
      "step": 7285
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9704694747924805,
      "learning_rate": 4.250291764259913e-05,
      "loss": 0.7551,
      "step": 7286
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9660972356796265,
      "learning_rate": 4.24783281128485e-05,
      "loss": 0.8246,
      "step": 7287
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9879716038703918,
      "learning_rate": 4.245374377985527e-05,
      "loss": 0.7224,
      "step": 7288
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9065752029418945,
      "learning_rate": 4.242916464584047e-05,
      "loss": 0.6902,
      "step": 7289
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8731761574745178,
      "learning_rate": 4.2404590713024674e-05,
      "loss": 0.705,
      "step": 7290
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8619717359542847,
      "learning_rate": 4.238002198362797e-05,
      "loss": 0.7319,
      "step": 7291
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8989507555961609,
      "learning_rate": 4.235545845987007e-05,
      "loss": 0.7846,
      "step": 7292
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9171220064163208,
      "learning_rate": 4.233090014397012e-05,
      "loss": 0.8149,
      "step": 7293
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8818581700325012,
      "learning_rate": 4.230634703814678e-05,
      "loss": 0.7605,
      "step": 7294
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.1395118236541748,
      "learning_rate": 4.22817991446184e-05,
      "loss": 0.9382,
      "step": 7295
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9644754528999329,
      "learning_rate": 4.225725646560259e-05,
      "loss": 0.7073,
      "step": 7296
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.0938143730163574,
      "learning_rate": 4.223271900331672e-05,
      "loss": 0.767,
      "step": 7297
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.037112832069397,
      "learning_rate": 4.2208186759977695e-05,
      "loss": 0.8437,
      "step": 7298
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9030419588088989,
      "learning_rate": 4.2183659737801694e-05,
      "loss": 0.8218,
      "step": 7299
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.0647999048233032,
      "learning_rate": 4.215913793900472e-05,
      "loss": 0.8803,
      "step": 7300
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9412992000579834,
      "learning_rate": 4.213462136580211e-05,
      "loss": 0.828,
      "step": 7301
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.0285842418670654,
      "learning_rate": 4.2110110020408855e-05,
      "loss": 0.8852,
      "step": 7302
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9961701035499573,
      "learning_rate": 4.2085603905039394e-05,
      "loss": 0.8578,
      "step": 7303
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9660602807998657,
      "learning_rate": 4.2061103021907665e-05,
      "loss": 0.8541,
      "step": 7304
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9902793765068054,
      "learning_rate": 4.203660737322731e-05,
      "loss": 0.8089,
      "step": 7305
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9722169637680054,
      "learning_rate": 4.2012116961211214e-05,
      "loss": 0.8321,
      "step": 7306
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.0839567184448242,
      "learning_rate": 4.198763178807202e-05,
      "loss": 0.8867,
      "step": 7307
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8626823425292969,
      "learning_rate": 4.196315185602191e-05,
      "loss": 0.8565,
      "step": 7308
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8883689045906067,
      "learning_rate": 4.193867716727235e-05,
      "loss": 0.9333,
      "step": 7309
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.0600382089614868,
      "learning_rate": 4.191420772403461e-05,
      "loss": 0.9208,
      "step": 7310
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.0022799968719482,
      "learning_rate": 4.1889743528519297e-05,
      "loss": 0.7623,
      "step": 7311
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9760846495628357,
      "learning_rate": 4.186528458293662e-05,
      "loss": 0.7926,
      "step": 7312
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9790765643119812,
      "learning_rate": 4.184083088949635e-05,
      "loss": 0.9401,
      "step": 7313
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9969648122787476,
      "learning_rate": 4.181638245040772e-05,
      "loss": 0.8996,
      "step": 7314
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9659683704376221,
      "learning_rate": 4.17919392678795e-05,
      "loss": 0.8423,
      "step": 7315
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.0410698652267456,
      "learning_rate": 4.176750134411996e-05,
      "loss": 0.861,
      "step": 7316
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.894158661365509,
      "learning_rate": 4.1743068681336995e-05,
      "loss": 0.7207,
      "step": 7317
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.8676133155822754,
      "learning_rate": 4.1718641281737927e-05,
      "loss": 0.6853,
      "step": 7318
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9624258279800415,
      "learning_rate": 4.169421914752959e-05,
      "loss": 0.8348,
      "step": 7319
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.8890660405158997,
      "learning_rate": 4.166980228091847e-05,
      "loss": 0.6789,
      "step": 7320
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.0068612098693848,
      "learning_rate": 4.164539068411044e-05,
      "loss": 0.7947,
      "step": 7321
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9628048539161682,
      "learning_rate": 4.162098435931092e-05,
      "loss": 1.0338,
      "step": 7322
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.1374199390411377,
      "learning_rate": 4.159658330872497e-05,
      "loss": 0.7515,
      "step": 7323
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9908962845802307,
      "learning_rate": 4.157218753455704e-05,
      "loss": 0.7079,
      "step": 7324
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9434610605239868,
      "learning_rate": 4.154779703901114e-05,
      "loss": 0.8083,
      "step": 7325
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.172394037246704,
      "learning_rate": 4.152341182429079e-05,
      "loss": 0.8263,
      "step": 7326
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.8510091304779053,
      "learning_rate": 4.1499031892599114e-05,
      "loss": 0.6817,
      "step": 7327
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.012534499168396,
      "learning_rate": 4.1474657246138695e-05,
      "loss": 0.7657,
      "step": 7328
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.8786703944206238,
      "learning_rate": 4.1450287887111575e-05,
      "loss": 0.5912,
      "step": 7329
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.009730339050293,
      "learning_rate": 4.142592381771947e-05,
      "loss": 0.9187,
      "step": 7330
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.0721036195755005,
      "learning_rate": 4.1401565040163515e-05,
      "loss": 0.832,
      "step": 7331
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9200707077980042,
      "learning_rate": 4.137721155664432e-05,
      "loss": 0.8043,
      "step": 7332
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.0563298463821411,
      "learning_rate": 4.1352863369362195e-05,
      "loss": 0.865,
      "step": 7333
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.176384449005127,
      "learning_rate": 4.1328520480516794e-05,
      "loss": 0.9401,
      "step": 7334
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.1089730262756348,
      "learning_rate": 4.1304182892307374e-05,
      "loss": 0.7542,
      "step": 7335
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.8969682455062866,
      "learning_rate": 4.127985060693266e-05,
      "loss": 0.8183,
      "step": 7336
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.05394446849823,
      "learning_rate": 4.125552362659101e-05,
      "loss": 0.9249,
      "step": 7337
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9255580306053162,
      "learning_rate": 4.123120195348018e-05,
      "loss": 0.7037,
      "step": 7338
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.064882755279541,
      "learning_rate": 4.1206885589797483e-05,
      "loss": 0.7679,
      "step": 7339
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9762742519378662,
      "learning_rate": 4.1182574537739846e-05,
      "loss": 0.6998,
      "step": 7340
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.120669960975647,
      "learning_rate": 4.115826879950351e-05,
      "loss": 0.7858,
      "step": 7341
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0707799196243286,
      "learning_rate": 4.113396837728443e-05,
      "loss": 0.9495,
      "step": 7342
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9362640976905823,
      "learning_rate": 4.1109673273278074e-05,
      "loss": 0.7596,
      "step": 7343
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9307307600975037,
      "learning_rate": 4.108538348967923e-05,
      "loss": 0.7302,
      "step": 7344
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9993686079978943,
      "learning_rate": 4.1061099028682456e-05,
      "loss": 0.9192,
      "step": 7345
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9033443331718445,
      "learning_rate": 4.103681989248167e-05,
      "loss": 0.8158,
      "step": 7346
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.8738240003585815,
      "learning_rate": 4.101254608327032e-05,
      "loss": 0.6937,
      "step": 7347
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.943271815776825,
      "learning_rate": 4.098827760324149e-05,
      "loss": 0.7507,
      "step": 7348
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9092500805854797,
      "learning_rate": 4.096401445458765e-05,
      "loss": 0.7216,
      "step": 7349
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9304053783416748,
      "learning_rate": 4.093975663950084e-05,
      "loss": 0.7273,
      "step": 7350
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0016545057296753,
      "learning_rate": 4.091550416017258e-05,
      "loss": 0.8689,
      "step": 7351
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9293085336685181,
      "learning_rate": 4.089125701879403e-05,
      "loss": 0.9104,
      "step": 7352
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0418977737426758,
      "learning_rate": 4.0867015217555725e-05,
      "loss": 0.8601,
      "step": 7353
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9580303430557251,
      "learning_rate": 4.084277875864776e-05,
      "loss": 0.9839,
      "step": 7354
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0078938007354736,
      "learning_rate": 4.0818547644259817e-05,
      "loss": 0.8539,
      "step": 7355
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9622752070426941,
      "learning_rate": 4.079432187658101e-05,
      "loss": 0.8285,
      "step": 7356
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0820382833480835,
      "learning_rate": 4.0770101457799957e-05,
      "loss": 0.8548,
      "step": 7357
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9905309081077576,
      "learning_rate": 4.074588639010493e-05,
      "loss": 1.1064,
      "step": 7358
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9273553490638733,
      "learning_rate": 4.0721676675683564e-05,
      "loss": 0.8549,
      "step": 7359
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9986502528190613,
      "learning_rate": 4.069747231672307e-05,
      "loss": 0.8457,
      "step": 7360
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.039328694343567,
      "learning_rate": 4.0673273315410155e-05,
      "loss": 0.8827,
      "step": 7361
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.1118755340576172,
      "learning_rate": 4.0649079673931125e-05,
      "loss": 0.8438,
      "step": 7362
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9130277633666992,
      "learning_rate": 4.06248913944717e-05,
      "loss": 0.742,
      "step": 7363
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9434802532196045,
      "learning_rate": 4.060070847921712e-05,
      "loss": 0.9579,
      "step": 7364
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.0194100141525269,
      "learning_rate": 4.0576530930352253e-05,
      "loss": 0.8477,
      "step": 7365
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9911008477210999,
      "learning_rate": 4.055235875006135e-05,
      "loss": 0.9051,
      "step": 7366
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9497823119163513,
      "learning_rate": 4.052819194052822e-05,
      "loss": 0.8551,
      "step": 7367
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9702839255332947,
      "learning_rate": 4.050403050393624e-05,
      "loss": 0.8153,
      "step": 7368
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.359513282775879,
      "learning_rate": 4.047987444246825e-05,
      "loss": 0.9259,
      "step": 7369
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9114631414413452,
      "learning_rate": 4.0455723758306594e-05,
      "loss": 0.7552,
      "step": 7370
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9315997958183289,
      "learning_rate": 4.043157845363314e-05,
      "loss": 0.8662,
      "step": 7371
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.0162763595581055,
      "learning_rate": 4.040743853062932e-05,
      "loss": 0.7385,
      "step": 7372
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9019334316253662,
      "learning_rate": 4.0383303991476004e-05,
      "loss": 0.8926,
      "step": 7373
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.0113277435302734,
      "learning_rate": 4.035917483835361e-05,
      "loss": 0.8331,
      "step": 7374
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.0835962295532227,
      "learning_rate": 4.033505107344214e-05,
      "loss": 0.8599,
      "step": 7375
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.8414437174797058,
      "learning_rate": 4.031093269892091e-05,
      "loss": 0.6151,
      "step": 7376
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9826193451881409,
      "learning_rate": 4.028681971696897e-05,
      "loss": 0.8845,
      "step": 7377
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9371057152748108,
      "learning_rate": 4.0262712129764834e-05,
      "loss": 0.7736,
      "step": 7378
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9073285460472107,
      "learning_rate": 4.023860993948635e-05,
      "loss": 0.6118,
      "step": 7379
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9783820509910583,
      "learning_rate": 4.021451314831113e-05,
      "loss": 0.8459,
      "step": 7380
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.912179708480835,
      "learning_rate": 4.019042175841614e-05,
      "loss": 0.7883,
      "step": 7381
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9421423077583313,
      "learning_rate": 4.016633577197787e-05,
      "loss": 0.8628,
      "step": 7382
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9135862588882446,
      "learning_rate": 4.014225519117243e-05,
      "loss": 0.7279,
      "step": 7383
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.842886209487915,
      "learning_rate": 4.011818001817531e-05,
      "loss": 0.6586,
      "step": 7384
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9631887674331665,
      "learning_rate": 4.0094110255161574e-05,
      "loss": 0.9358,
      "step": 7385
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.8843592405319214,
      "learning_rate": 4.0070045904305755e-05,
      "loss": 0.8878,
      "step": 7386
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.8904747366905212,
      "learning_rate": 4.0045986967782004e-05,
      "loss": 0.7899,
      "step": 7387
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9461697340011597,
      "learning_rate": 4.002193344776387e-05,
      "loss": 0.8929,
      "step": 7388
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.0168232917785645,
      "learning_rate": 3.9997885346424425e-05,
      "loss": 0.8377,
      "step": 7389
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9717044830322266,
      "learning_rate": 3.9973842665936336e-05,
      "loss": 0.8453,
      "step": 7390
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9901056885719299,
      "learning_rate": 3.99498054084717e-05,
      "loss": 0.783,
      "step": 7391
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0056730508804321,
      "learning_rate": 3.99257735762021e-05,
      "loss": 0.7241,
      "step": 7392
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0124402046203613,
      "learning_rate": 3.9901747171298766e-05,
      "loss": 0.7749,
      "step": 7393
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.8901179432868958,
      "learning_rate": 3.9877726195932284e-05,
      "loss": 0.7961,
      "step": 7394
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.1125584840774536,
      "learning_rate": 3.9853710652272834e-05,
      "loss": 0.9293,
      "step": 7395
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.8503389358520508,
      "learning_rate": 3.9829700542490046e-05,
      "loss": 0.814,
      "step": 7396
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9520916938781738,
      "learning_rate": 3.9805695868753166e-05,
      "loss": 0.7718,
      "step": 7397
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.957912027835846,
      "learning_rate": 3.978169663323085e-05,
      "loss": 0.8121,
      "step": 7398
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.8091579079627991,
      "learning_rate": 3.975770283809125e-05,
      "loss": 0.6461,
      "step": 7399
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.8797688484191895,
      "learning_rate": 3.9733714485502135e-05,
      "loss": 0.6825,
      "step": 7400
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.1208490133285522,
      "learning_rate": 3.970973157763069e-05,
      "loss": 0.8047,
      "step": 7401
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0606164932250977,
      "learning_rate": 3.9685754116643606e-05,
      "loss": 0.8813,
      "step": 7402
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.1251953840255737,
      "learning_rate": 3.966178210470718e-05,
      "loss": 1.2012,
      "step": 7403
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0652228593826294,
      "learning_rate": 3.9637815543987086e-05,
      "loss": 0.9501,
      "step": 7404
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.1458171606063843,
      "learning_rate": 3.961385443664861e-05,
      "loss": 0.8727,
      "step": 7405
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0286813974380493,
      "learning_rate": 3.958989878485644e-05,
      "loss": 0.8459,
      "step": 7406
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9711876511573792,
      "learning_rate": 3.9565948590774904e-05,
      "loss": 0.8425,
      "step": 7407
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9978609681129456,
      "learning_rate": 3.9542003856567736e-05,
      "loss": 0.666,
      "step": 7408
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0384318828582764,
      "learning_rate": 3.9518064584398176e-05,
      "loss": 1.0148,
      "step": 7409
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.2954719066619873,
      "learning_rate": 3.949413077642909e-05,
      "loss": 0.9136,
      "step": 7410
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9489061832427979,
      "learning_rate": 3.947020243482265e-05,
      "loss": 0.8507,
      "step": 7411
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9709159731864929,
      "learning_rate": 3.9446279561740695e-05,
      "loss": 0.7461,
      "step": 7412
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0093650817871094,
      "learning_rate": 3.94223621593446e-05,
      "loss": 0.8508,
      "step": 7413
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.214653491973877,
      "learning_rate": 3.9398450229795026e-05,
      "loss": 0.7463,
      "step": 7414
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9259235858917236,
      "learning_rate": 3.937454377525238e-05,
      "loss": 0.8495,
      "step": 7415
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.8830484747886658,
      "learning_rate": 3.935064279787644e-05,
      "loss": 0.667,
      "step": 7416
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0812674760818481,
      "learning_rate": 3.9326747299826495e-05,
      "loss": 0.9125,
      "step": 7417
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9452555775642395,
      "learning_rate": 3.930285728326144e-05,
      "loss": 0.7865,
      "step": 7418
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.0270934104919434,
      "learning_rate": 3.927897275033956e-05,
      "loss": 0.8626,
      "step": 7419
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9184280037879944,
      "learning_rate": 3.925509370321869e-05,
      "loss": 0.6856,
      "step": 7420
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9670143723487854,
      "learning_rate": 3.923122014405613e-05,
      "loss": 0.7569,
      "step": 7421
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9759781360626221,
      "learning_rate": 3.920735207500876e-05,
      "loss": 0.8584,
      "step": 7422
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9510020017623901,
      "learning_rate": 3.9183489498233e-05,
      "loss": 0.7545,
      "step": 7423
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9124219417572021,
      "learning_rate": 3.915963241588454e-05,
      "loss": 0.7173,
      "step": 7424
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.959901750087738,
      "learning_rate": 3.913578083011885e-05,
      "loss": 0.9627,
      "step": 7425
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9635375142097473,
      "learning_rate": 3.911193474309076e-05,
      "loss": 0.9591,
      "step": 7426
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9736372828483582,
      "learning_rate": 3.908809415695458e-05,
      "loss": 0.8499,
      "step": 7427
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.097109079360962,
      "learning_rate": 3.906425907386424e-05,
      "loss": 0.8765,
      "step": 7428
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.1514911651611328,
      "learning_rate": 3.904042949597309e-05,
      "loss": 0.8277,
      "step": 7429
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9284905195236206,
      "learning_rate": 3.901660542543396e-05,
      "loss": 0.8331,
      "step": 7430
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.085813045501709,
      "learning_rate": 3.899278686439922e-05,
      "loss": 0.8728,
      "step": 7431
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.1249289512634277,
      "learning_rate": 3.8968973815020806e-05,
      "loss": 1.0557,
      "step": 7432
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9731318950653076,
      "learning_rate": 3.894516627945004e-05,
      "loss": 0.8294,
      "step": 7433
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9025543332099915,
      "learning_rate": 3.8921364259837776e-05,
      "loss": 0.7009,
      "step": 7434
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.8979260921478271,
      "learning_rate": 3.8897567758334464e-05,
      "loss": 0.7896,
      "step": 7435
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.0135040283203125,
      "learning_rate": 3.887377677708994e-05,
      "loss": 0.8734,
      "step": 7436
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9246363639831543,
      "learning_rate": 3.884999131825355e-05,
      "loss": 0.9427,
      "step": 7437
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.059679627418518,
      "learning_rate": 3.8826211383974266e-05,
      "loss": 0.8051,
      "step": 7438
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9498894214630127,
      "learning_rate": 3.88024369764004e-05,
      "loss": 0.8145,
      "step": 7439
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.006415843963623,
      "learning_rate": 3.877866809767986e-05,
      "loss": 0.8104,
      "step": 7440
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.0181704759597778,
      "learning_rate": 3.875490474996e-05,
      "loss": 1.0395,
      "step": 7441
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.1583374738693237,
      "learning_rate": 3.873114693538774e-05,
      "loss": 0.8152,
      "step": 7442
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.061574935913086,
      "learning_rate": 3.870739465610946e-05,
      "loss": 0.8941,
      "step": 7443
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9605360627174377,
      "learning_rate": 3.8683647914271004e-05,
      "loss": 0.9128,
      "step": 7444
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.897330105304718,
      "learning_rate": 3.8659906712017856e-05,
      "loss": 0.8667,
      "step": 7445
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.0486851930618286,
      "learning_rate": 3.863617105149476e-05,
      "loss": 0.9002,
      "step": 7446
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.8256691694259644,
      "learning_rate": 3.861244093484616e-05,
      "loss": 0.6867,
      "step": 7447
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9991077184677124,
      "learning_rate": 3.8588716364216025e-05,
      "loss": 0.863,
      "step": 7448
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9601210951805115,
      "learning_rate": 3.856499734174759e-05,
      "loss": 0.9451,
      "step": 7449
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9319004416465759,
      "learning_rate": 3.854128386958382e-05,
      "loss": 0.8907,
      "step": 7450
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9763748645782471,
      "learning_rate": 3.851757594986706e-05,
      "loss": 0.6948,
      "step": 7451
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9684788584709167,
      "learning_rate": 3.849387358473923e-05,
      "loss": 0.8829,
      "step": 7452
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.0739248991012573,
      "learning_rate": 3.847017677634166e-05,
      "loss": 0.8864,
      "step": 7453
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9716264605522156,
      "learning_rate": 3.8446485526815215e-05,
      "loss": 0.7347,
      "step": 7454
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.1437360048294067,
      "learning_rate": 3.842279983830035e-05,
      "loss": 0.8667,
      "step": 7455
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9826272130012512,
      "learning_rate": 3.8399119712936805e-05,
      "loss": 0.8979,
      "step": 7456
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9834003448486328,
      "learning_rate": 3.837544515286401e-05,
      "loss": 0.7194,
      "step": 7457
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9424280524253845,
      "learning_rate": 3.83517761602209e-05,
      "loss": 0.7976,
      "step": 7458
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9623442888259888,
      "learning_rate": 3.832811273714569e-05,
      "loss": 0.7613,
      "step": 7459
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9138621091842651,
      "learning_rate": 3.8304454885776356e-05,
      "loss": 0.8326,
      "step": 7460
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.0111273527145386,
      "learning_rate": 3.8280802608250185e-05,
      "loss": 0.9576,
      "step": 7461
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.8794863820075989,
      "learning_rate": 3.825715590670402e-05,
      "loss": 0.8351,
      "step": 7462
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9375495910644531,
      "learning_rate": 3.8233514783274274e-05,
      "loss": 0.7657,
      "step": 7463
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.0412698984146118,
      "learning_rate": 3.8209879240096724e-05,
      "loss": 0.9017,
      "step": 7464
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.0985389947891235,
      "learning_rate": 3.8186249279306754e-05,
      "loss": 0.786,
      "step": 7465
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9962413311004639,
      "learning_rate": 3.816262490303912e-05,
      "loss": 0.9082,
      "step": 7466
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.990900456905365,
      "learning_rate": 3.813900611342825e-05,
      "loss": 0.8882,
      "step": 7467
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.911271870136261,
      "learning_rate": 3.8115392912607916e-05,
      "loss": 0.7605,
      "step": 7468
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.0053874254226685,
      "learning_rate": 3.8091785302711414e-05,
      "loss": 0.841,
      "step": 7469
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9500042796134949,
      "learning_rate": 3.806818328587162e-05,
      "loss": 0.8396,
      "step": 7470
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9464622735977173,
      "learning_rate": 3.80445868642208e-05,
      "loss": 0.8365,
      "step": 7471
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.0151314735412598,
      "learning_rate": 3.8020996039890735e-05,
      "loss": 0.9998,
      "step": 7472
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9761706590652466,
      "learning_rate": 3.799741081501279e-05,
      "loss": 0.8283,
      "step": 7473
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.0961986780166626,
      "learning_rate": 3.7973831191717726e-05,
      "loss": 0.746,
      "step": 7474
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9782766699790955,
      "learning_rate": 3.7950257172135815e-05,
      "loss": 0.844,
      "step": 7475
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.1434317827224731,
      "learning_rate": 3.792668875839681e-05,
      "loss": 0.886,
      "step": 7476
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9561150074005127,
      "learning_rate": 3.790312595263006e-05,
      "loss": 0.8252,
      "step": 7477
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9355947375297546,
      "learning_rate": 3.7879568756964276e-05,
      "loss": 0.8631,
      "step": 7478
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.109441876411438,
      "learning_rate": 3.78560171735277e-05,
      "loss": 0.9409,
      "step": 7479
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9464274048805237,
      "learning_rate": 3.7832471204448195e-05,
      "loss": 0.7677,
      "step": 7480
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.8974305987358093,
      "learning_rate": 3.780893085185284e-05,
      "loss": 0.8906,
      "step": 7481
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.0238032341003418,
      "learning_rate": 3.7785396117868465e-05,
      "loss": 0.8152,
      "step": 7482
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.0467618703842163,
      "learning_rate": 3.776186700462132e-05,
      "loss": 0.9008,
      "step": 7483
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9790921807289124,
      "learning_rate": 3.7738343514237106e-05,
      "loss": 0.8523,
      "step": 7484
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.1401054859161377,
      "learning_rate": 3.7714825648841036e-05,
      "loss": 0.9407,
      "step": 7485
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.1373331546783447,
      "learning_rate": 3.769131341055777e-05,
      "loss": 0.8899,
      "step": 7486
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9715731739997864,
      "learning_rate": 3.766780680151158e-05,
      "loss": 0.7283,
      "step": 7487
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9903486967086792,
      "learning_rate": 3.764430582382612e-05,
      "loss": 0.7439,
      "step": 7488
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9357986450195312,
      "learning_rate": 3.7620810479624544e-05,
      "loss": 0.9172,
      "step": 7489
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9297086000442505,
      "learning_rate": 3.759732077102962e-05,
      "loss": 0.7959,
      "step": 7490
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9089120030403137,
      "learning_rate": 3.7573836700163365e-05,
      "loss": 0.781,
      "step": 7491
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.0298839807510376,
      "learning_rate": 3.7550358269147515e-05,
      "loss": 0.746,
      "step": 7492
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.8338559865951538,
      "learning_rate": 3.752688548010327e-05,
      "loss": 0.7932,
      "step": 7493
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.0771530866622925,
      "learning_rate": 3.750341833515113e-05,
      "loss": 0.9928,
      "step": 7494
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.01557195186615,
      "learning_rate": 3.747995683641134e-05,
      "loss": 0.8054,
      "step": 7495
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9451504945755005,
      "learning_rate": 3.745650098600344e-05,
      "loss": 0.8988,
      "step": 7496
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.8821951746940613,
      "learning_rate": 3.743305078604653e-05,
      "loss": 0.8446,
      "step": 7497
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9468172788619995,
      "learning_rate": 3.740960623865927e-05,
      "loss": 0.8008,
      "step": 7498
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9664888978004456,
      "learning_rate": 3.738616734595969e-05,
      "loss": 0.8104,
      "step": 7499
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.1887441873550415,
      "learning_rate": 3.736273411006538e-05,
      "loss": 0.8909,
      "step": 7500
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.8941977024078369,
      "learning_rate": 3.7339306533093345e-05,
      "loss": 0.8011,
      "step": 7501
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.956028401851654,
      "learning_rate": 3.731588461716022e-05,
      "loss": 0.9863,
      "step": 7502
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9884627461433411,
      "learning_rate": 3.7292468364382e-05,
      "loss": 0.6904,
      "step": 7503
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.0923556089401245,
      "learning_rate": 3.726905777687419e-05,
      "loss": 0.7597,
      "step": 7504
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9658195972442627,
      "learning_rate": 3.724565285675187e-05,
      "loss": 0.8697,
      "step": 7505
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.0650930404663086,
      "learning_rate": 3.722225360612949e-05,
      "loss": 0.8486,
      "step": 7506
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.019537329673767,
      "learning_rate": 3.719886002712101e-05,
      "loss": 0.8022,
      "step": 7507
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.1393005847930908,
      "learning_rate": 3.7175472121839984e-05,
      "loss": 0.8624,
      "step": 7508
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.059464454650879,
      "learning_rate": 3.7152089892399344e-05,
      "loss": 0.845,
      "step": 7509
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9705213308334351,
      "learning_rate": 3.7128713340911535e-05,
      "loss": 0.9656,
      "step": 7510
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9300762414932251,
      "learning_rate": 3.710534246948847e-05,
      "loss": 0.7548,
      "step": 7511
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9506428241729736,
      "learning_rate": 3.708197728024163e-05,
      "loss": 0.7996,
      "step": 7512
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.0184721946716309,
      "learning_rate": 3.705861777528191e-05,
      "loss": 0.8819,
      "step": 7513
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.042346715927124,
      "learning_rate": 3.703526395671967e-05,
      "loss": 0.8268,
      "step": 7514
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9002046585083008,
      "learning_rate": 3.7011915826664846e-05,
      "loss": 0.7559,
      "step": 7515
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9072455763816833,
      "learning_rate": 3.69885733872268e-05,
      "loss": 0.8073,
      "step": 7516
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.0319515466690063,
      "learning_rate": 3.696523664051434e-05,
      "loss": 0.8689,
      "step": 7517
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.8975298404693604,
      "learning_rate": 3.6941905588635884e-05,
      "loss": 0.8382,
      "step": 7518
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9671549201011658,
      "learning_rate": 3.691858023369923e-05,
      "loss": 0.6946,
      "step": 7519
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.8181380033493042,
      "learning_rate": 3.689526057781167e-05,
      "loss": 0.8054,
      "step": 7520
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9802835583686829,
      "learning_rate": 3.687194662308e-05,
      "loss": 0.7742,
      "step": 7521
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.8900817036628723,
      "learning_rate": 3.6848638371610554e-05,
      "loss": 0.7382,
      "step": 7522
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9533427953720093,
      "learning_rate": 3.6825335825509065e-05,
      "loss": 0.895,
      "step": 7523
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9201135635375977,
      "learning_rate": 3.680203898688075e-05,
      "loss": 0.812,
      "step": 7524
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9332418441772461,
      "learning_rate": 3.6778747857830465e-05,
      "loss": 0.8165,
      "step": 7525
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.863610029220581,
      "learning_rate": 3.675546244046228e-05,
      "loss": 0.6864,
      "step": 7526
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9479172229766846,
      "learning_rate": 3.673218273687998e-05,
      "loss": 0.8253,
      "step": 7527
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.965356707572937,
      "learning_rate": 3.670890874918681e-05,
      "loss": 0.6394,
      "step": 7528
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9276130199432373,
      "learning_rate": 3.668564047948532e-05,
      "loss": 0.7869,
      "step": 7529
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9348182678222656,
      "learning_rate": 3.666237792987776e-05,
      "loss": 0.9145,
      "step": 7530
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.1533371210098267,
      "learning_rate": 3.6639121102465735e-05,
      "loss": 0.887,
      "step": 7531
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.0226624011993408,
      "learning_rate": 3.6615869999350327e-05,
      "loss": 0.7732,
      "step": 7532
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9505677223205566,
      "learning_rate": 3.6592624622632234e-05,
      "loss": 0.9497,
      "step": 7533
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.0824247598648071,
      "learning_rate": 3.656938497441148e-05,
      "loss": 0.9477,
      "step": 7534
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.097638726234436,
      "learning_rate": 3.6546151056787655e-05,
      "loss": 0.8119,
      "step": 7535
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9324548244476318,
      "learning_rate": 3.652292287185979e-05,
      "loss": 0.8084,
      "step": 7536
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9264864921569824,
      "learning_rate": 3.6499700421726455e-05,
      "loss": 0.8599,
      "step": 7537
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.041636347770691,
      "learning_rate": 3.6476483708485654e-05,
      "loss": 0.7238,
      "step": 7538
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.001818060874939,
      "learning_rate": 3.6453272734234846e-05,
      "loss": 0.9248,
      "step": 7539
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.1082067489624023,
      "learning_rate": 3.643006750107107e-05,
      "loss": 0.9334,
      "step": 7540
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9608253240585327,
      "learning_rate": 3.6406868011090776e-05,
      "loss": 0.8448,
      "step": 7541
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.1235332489013672,
      "learning_rate": 3.6383674266389854e-05,
      "loss": 0.9784,
      "step": 7542
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.0410866737365723,
      "learning_rate": 3.636048626906381e-05,
      "loss": 0.711,
      "step": 7543
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.8764690160751343,
      "learning_rate": 3.6337304021207495e-05,
      "loss": 0.7616,
      "step": 7544
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9428754448890686,
      "learning_rate": 3.631412752491531e-05,
      "loss": 0.7947,
      "step": 7545
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.0265016555786133,
      "learning_rate": 3.6290956782281075e-05,
      "loss": 0.8633,
      "step": 7546
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.8968940377235413,
      "learning_rate": 3.626779179539821e-05,
      "loss": 0.8276,
      "step": 7547
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9372925162315369,
      "learning_rate": 3.6244632566359516e-05,
      "loss": 0.6949,
      "step": 7548
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.956941545009613,
      "learning_rate": 3.622147909725724e-05,
      "loss": 0.8436,
      "step": 7549
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0152337551116943,
      "learning_rate": 3.619833139018326e-05,
      "loss": 0.7951,
      "step": 7550
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0626733303070068,
      "learning_rate": 3.6175189447228784e-05,
      "loss": 0.7575,
      "step": 7551
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9407270550727844,
      "learning_rate": 3.615205327048454e-05,
      "loss": 0.83,
      "step": 7552
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.140998125076294,
      "learning_rate": 3.6128922862040805e-05,
      "loss": 0.9363,
      "step": 7553
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.03515625,
      "learning_rate": 3.610579822398726e-05,
      "loss": 0.8549,
      "step": 7554
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0622241497039795,
      "learning_rate": 3.6082679358413076e-05,
      "loss": 0.6806,
      "step": 7555
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.2061697244644165,
      "learning_rate": 3.6059566267406875e-05,
      "loss": 0.9062,
      "step": 7556
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9492170810699463,
      "learning_rate": 3.603645895305686e-05,
      "loss": 0.797,
      "step": 7557
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9267086982727051,
      "learning_rate": 3.601335741745063e-05,
      "loss": 0.6643,
      "step": 7558
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9608106017112732,
      "learning_rate": 3.599026166267523e-05,
      "loss": 0.7513,
      "step": 7559
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9860916137695312,
      "learning_rate": 3.596717169081733e-05,
      "loss": 0.828,
      "step": 7560
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9661557078361511,
      "learning_rate": 3.594408750396286e-05,
      "loss": 0.9028,
      "step": 7561
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.138761281967163,
      "learning_rate": 3.592100910419738e-05,
      "loss": 0.7307,
      "step": 7562
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9983358383178711,
      "learning_rate": 3.589793649360599e-05,
      "loss": 0.8597,
      "step": 7563
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0276944637298584,
      "learning_rate": 3.587486967427303e-05,
      "loss": 0.8872,
      "step": 7564
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0254305601119995,
      "learning_rate": 3.5851808648282544e-05,
      "loss": 0.8991,
      "step": 7565
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0396320819854736,
      "learning_rate": 3.582875341771794e-05,
      "loss": 0.8653,
      "step": 7566
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9767132997512817,
      "learning_rate": 3.5805703984662096e-05,
      "loss": 0.8427,
      "step": 7567
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0015809535980225,
      "learning_rate": 3.5782660351197474e-05,
      "loss": 0.9211,
      "step": 7568
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.8895262479782104,
      "learning_rate": 3.575962251940589e-05,
      "loss": 0.7588,
      "step": 7569
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9425777196884155,
      "learning_rate": 3.573659049136867e-05,
      "loss": 0.9028,
      "step": 7570
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9402073621749878,
      "learning_rate": 3.5713564269166625e-05,
      "loss": 0.8775,
      "step": 7571
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9315981864929199,
      "learning_rate": 3.5690543854880075e-05,
      "loss": 0.8352,
      "step": 7572
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0591684579849243,
      "learning_rate": 3.566752925058877e-05,
      "loss": 0.8352,
      "step": 7573
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.968318521976471,
      "learning_rate": 3.564452045837192e-05,
      "loss": 0.7904,
      "step": 7574
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.0411642789840698,
      "learning_rate": 3.562151748030831e-05,
      "loss": 0.9067,
      "step": 7575
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9507133364677429,
      "learning_rate": 3.559852031847607e-05,
      "loss": 0.8044,
      "step": 7576
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9419892430305481,
      "learning_rate": 3.5575528974952854e-05,
      "loss": 0.8228,
      "step": 7577
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9809172749519348,
      "learning_rate": 3.555254345181586e-05,
      "loss": 0.7958,
      "step": 7578
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9046222567558289,
      "learning_rate": 3.552956375114166e-05,
      "loss": 0.69,
      "step": 7579
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9287858605384827,
      "learning_rate": 3.5506589875006346e-05,
      "loss": 0.8254,
      "step": 7580
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.0500366687774658,
      "learning_rate": 3.548362182548545e-05,
      "loss": 0.7536,
      "step": 7581
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.0967304706573486,
      "learning_rate": 3.546065960465405e-05,
      "loss": 0.7549,
      "step": 7582
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9044837951660156,
      "learning_rate": 3.543770321458664e-05,
      "loss": 0.751,
      "step": 7583
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.1360193490982056,
      "learning_rate": 3.5414752657357165e-05,
      "loss": 0.8121,
      "step": 7584
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.8218927383422852,
      "learning_rate": 3.5391807935039137e-05,
      "loss": 0.6496,
      "step": 7585
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9573684930801392,
      "learning_rate": 3.5368869049705446e-05,
      "loss": 0.9244,
      "step": 7586
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.8766704201698303,
      "learning_rate": 3.534593600342847e-05,
      "loss": 0.7003,
      "step": 7587
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9023606777191162,
      "learning_rate": 3.532300879828013e-05,
      "loss": 0.9326,
      "step": 7588
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.227485179901123,
      "learning_rate": 3.530008743633175e-05,
      "loss": 0.6886,
      "step": 7589
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9296458959579468,
      "learning_rate": 3.527717191965415e-05,
      "loss": 0.8175,
      "step": 7590
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.934276819229126,
      "learning_rate": 3.5254262250317563e-05,
      "loss": 0.7418,
      "step": 7591
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9455251097679138,
      "learning_rate": 3.523135843039183e-05,
      "loss": 0.8427,
      "step": 7592
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.2484594583511353,
      "learning_rate": 3.520846046194614e-05,
      "loss": 0.6991,
      "step": 7593
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.0081783533096313,
      "learning_rate": 3.518556834704917e-05,
      "loss": 0.7616,
      "step": 7594
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9542840123176575,
      "learning_rate": 3.516268208776918e-05,
      "loss": 0.726,
      "step": 7595
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9109475016593933,
      "learning_rate": 3.513980168617369e-05,
      "loss": 0.8553,
      "step": 7596
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.874394953250885,
      "learning_rate": 3.511692714432989e-05,
      "loss": 0.7769,
      "step": 7597
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9292196035385132,
      "learning_rate": 3.5094058464304415e-05,
      "loss": 0.7252,
      "step": 7598
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.8869231939315796,
      "learning_rate": 3.507119564816319e-05,
      "loss": 0.9495,
      "step": 7599
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9962003231048584,
      "learning_rate": 3.5048338697971864e-05,
      "loss": 0.9234,
      "step": 7600
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.1789199113845825,
      "learning_rate": 3.502548761579536e-05,
      "loss": 0.7692,
      "step": 7601
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9038863778114319,
      "learning_rate": 3.5002642403698136e-05,
      "loss": 0.8049,
      "step": 7602
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.963026225566864,
      "learning_rate": 3.49798030637442e-05,
      "loss": 0.8774,
      "step": 7603
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9255886673927307,
      "learning_rate": 3.495696959799688e-05,
      "loss": 0.6909,
      "step": 7604
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.0083898305892944,
      "learning_rate": 3.493414200851914e-05,
      "loss": 0.8241,
      "step": 7605
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9958054423332214,
      "learning_rate": 3.4911320297373204e-05,
      "loss": 0.8323,
      "step": 7606
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9574832916259766,
      "learning_rate": 3.488850446662095e-05,
      "loss": 0.7418,
      "step": 7607
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.8526789546012878,
      "learning_rate": 3.486569451832373e-05,
      "loss": 0.8173,
      "step": 7608
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9620357155799866,
      "learning_rate": 3.4842890454542156e-05,
      "loss": 0.7538,
      "step": 7609
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.8619489073753357,
      "learning_rate": 3.482009227733654e-05,
      "loss": 0.6205,
      "step": 7610
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9371187090873718,
      "learning_rate": 3.479729998876653e-05,
      "loss": 0.6764,
      "step": 7611
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9781648516654968,
      "learning_rate": 3.4774513590891254e-05,
      "loss": 0.8035,
      "step": 7612
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9841399788856506,
      "learning_rate": 3.475173308576941e-05,
      "loss": 1.0268,
      "step": 7613
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.0783793926239014,
      "learning_rate": 3.472895847545905e-05,
      "loss": 0.8981,
      "step": 7614
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.8323302865028381,
      "learning_rate": 3.470618976201773e-05,
      "loss": 0.6629,
      "step": 7615
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9074571132659912,
      "learning_rate": 3.468342694750243e-05,
      "loss": 0.7539,
      "step": 7616
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9729734063148499,
      "learning_rate": 3.4660670033969724e-05,
      "loss": 0.8901,
      "step": 7617
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.10355544090271,
      "learning_rate": 3.4637919023475526e-05,
      "loss": 0.8687,
      "step": 7618
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.8386011719703674,
      "learning_rate": 3.461517391807524e-05,
      "loss": 0.7951,
      "step": 7619
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.236383080482483,
      "learning_rate": 3.459243471982381e-05,
      "loss": 0.8513,
      "step": 7620
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9927204251289368,
      "learning_rate": 3.456970143077557e-05,
      "loss": 0.824,
      "step": 7621
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9276738166809082,
      "learning_rate": 3.454697405298429e-05,
      "loss": 0.8413,
      "step": 7622
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9685863852500916,
      "learning_rate": 3.4524252588503367e-05,
      "loss": 0.7606,
      "step": 7623
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.143633484840393,
      "learning_rate": 3.450153703938549e-05,
      "loss": 0.931,
      "step": 7624
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.8996179699897766,
      "learning_rate": 3.4478827407682887e-05,
      "loss": 0.7748,
      "step": 7625
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.97540283203125,
      "learning_rate": 3.445612369544722e-05,
      "loss": 0.8509,
      "step": 7626
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9543501734733582,
      "learning_rate": 3.44334259047297e-05,
      "loss": 0.8943,
      "step": 7627
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.833642303943634,
      "learning_rate": 3.4410734037580906e-05,
      "loss": 0.674,
      "step": 7628
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8594216108322144,
      "learning_rate": 3.438804809605091e-05,
      "loss": 0.7516,
      "step": 7629
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.0499866008758545,
      "learning_rate": 3.436536808218932e-05,
      "loss": 0.8261,
      "step": 7630
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.960899293422699,
      "learning_rate": 3.434269399804505e-05,
      "loss": 0.8132,
      "step": 7631
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.034379243850708,
      "learning_rate": 3.432002584566663e-05,
      "loss": 0.7147,
      "step": 7632
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.0016978979110718,
      "learning_rate": 3.429736362710201e-05,
      "loss": 0.7408,
      "step": 7633
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.077154278755188,
      "learning_rate": 3.4274707344398594e-05,
      "loss": 0.8858,
      "step": 7634
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.0097711086273193,
      "learning_rate": 3.4252056999603235e-05,
      "loss": 0.7698,
      "step": 7635
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.964712381362915,
      "learning_rate": 3.422941259476221e-05,
      "loss": 0.8318,
      "step": 7636
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8586922287940979,
      "learning_rate": 3.420677413192141e-05,
      "loss": 0.818,
      "step": 7637
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9306658506393433,
      "learning_rate": 3.4184141613126055e-05,
      "loss": 0.8381,
      "step": 7638
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8760762810707092,
      "learning_rate": 3.416151504042082e-05,
      "loss": 0.7533,
      "step": 7639
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9743070006370544,
      "learning_rate": 3.413889441584999e-05,
      "loss": 0.8644,
      "step": 7640
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.0653449296951294,
      "learning_rate": 3.4116279741457066e-05,
      "loss": 0.9253,
      "step": 7641
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8696021437644958,
      "learning_rate": 3.4093671019285256e-05,
      "loss": 0.8,
      "step": 7642
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8965446949005127,
      "learning_rate": 3.407106825137716e-05,
      "loss": 0.7362,
      "step": 7643
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9524657130241394,
      "learning_rate": 3.40484714397747e-05,
      "loss": 0.8251,
      "step": 7644
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.0200276374816895,
      "learning_rate": 3.4025880586519454e-05,
      "loss": 0.9365,
      "step": 7645
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9486342668533325,
      "learning_rate": 3.4003295693652357e-05,
      "loss": 0.8376,
      "step": 7646
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.1421313285827637,
      "learning_rate": 3.3980716763213794e-05,
      "loss": 0.9754,
      "step": 7647
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9667962193489075,
      "learning_rate": 3.39581437972437e-05,
      "loss": 0.7702,
      "step": 7648
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8326226472854614,
      "learning_rate": 3.3935576797781385e-05,
      "loss": 0.8857,
      "step": 7649
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9404334425926208,
      "learning_rate": 3.391301576686565e-05,
      "loss": 0.7651,
      "step": 7650
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.2030810117721558,
      "learning_rate": 3.389046070653473e-05,
      "loss": 0.8877,
      "step": 7651
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9697878956794739,
      "learning_rate": 3.38679116188264e-05,
      "loss": 0.7881,
      "step": 7652
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.8281930088996887,
      "learning_rate": 3.384536850577782e-05,
      "loss": 0.7713,
      "step": 7653
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.0146692991256714,
      "learning_rate": 3.382283136942559e-05,
      "loss": 0.8504,
      "step": 7654
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9004440903663635,
      "learning_rate": 3.380030021180589e-05,
      "loss": 0.8524,
      "step": 7655
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.0009188652038574,
      "learning_rate": 3.377777503495424e-05,
      "loss": 0.9172,
      "step": 7656
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9263618588447571,
      "learning_rate": 3.375525584090564e-05,
      "loss": 0.7716,
      "step": 7657
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9557112455368042,
      "learning_rate": 3.3732742631694625e-05,
      "loss": 0.9176,
      "step": 7658
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.944005012512207,
      "learning_rate": 3.3710235409355116e-05,
      "loss": 0.9597,
      "step": 7659
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9867224097251892,
      "learning_rate": 3.36877341759205e-05,
      "loss": 0.7239,
      "step": 7660
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.8566451668739319,
      "learning_rate": 3.3665238933423614e-05,
      "loss": 0.7674,
      "step": 7661
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9859318137168884,
      "learning_rate": 3.364274968389684e-05,
      "loss": 0.8769,
      "step": 7662
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.8600419163703918,
      "learning_rate": 3.3620266429371905e-05,
      "loss": 0.8136,
      "step": 7663
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9154042601585388,
      "learning_rate": 3.3597789171880036e-05,
      "loss": 0.7982,
      "step": 7664
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.0354702472686768,
      "learning_rate": 3.357531791345196e-05,
      "loss": 0.9725,
      "step": 7665
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.1528338193893433,
      "learning_rate": 3.355285265611784e-05,
      "loss": 0.718,
      "step": 7666
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.0112946033477783,
      "learning_rate": 3.3530393401907214e-05,
      "loss": 0.8584,
      "step": 7667
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.0175222158432007,
      "learning_rate": 3.3507940152849224e-05,
      "loss": 1.0192,
      "step": 7668
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.7980990409851074,
      "learning_rate": 3.348549291097238e-05,
      "loss": 0.4968,
      "step": 7669
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.037209391593933,
      "learning_rate": 3.3463051678304636e-05,
      "loss": 0.8696,
      "step": 7670
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9768759608268738,
      "learning_rate": 3.344061645687341e-05,
      "loss": 0.6877,
      "step": 7671
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.903754711151123,
      "learning_rate": 3.3418187248705654e-05,
      "loss": 0.8375,
      "step": 7672
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.042909860610962,
      "learning_rate": 3.33957640558277e-05,
      "loss": 0.8982,
      "step": 7673
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.0868935585021973,
      "learning_rate": 3.337334688026533e-05,
      "loss": 0.8986,
      "step": 7674
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.075210452079773,
      "learning_rate": 3.335093572404389e-05,
      "loss": 0.7463,
      "step": 7675
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9595418572425842,
      "learning_rate": 3.3328530589187966e-05,
      "loss": 0.6873,
      "step": 7676
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.04567289352417,
      "learning_rate": 3.3306131477721826e-05,
      "loss": 0.8131,
      "step": 7677
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9727736115455627,
      "learning_rate": 3.328373839166914e-05,
      "loss": 0.7715,
      "step": 7678
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9727533459663391,
      "learning_rate": 3.326135133305288e-05,
      "loss": 0.802,
      "step": 7679
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9269750714302063,
      "learning_rate": 3.323897030389569e-05,
      "loss": 0.8919,
      "step": 7680
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9509264230728149,
      "learning_rate": 3.321659530621953e-05,
      "loss": 0.6519,
      "step": 7681
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9853764772415161,
      "learning_rate": 3.319422634204583e-05,
      "loss": 0.759,
      "step": 7682
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.8609706163406372,
      "learning_rate": 3.3171863413395554e-05,
      "loss": 0.7326,
      "step": 7683
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0264180898666382,
      "learning_rate": 3.314950652228904e-05,
      "loss": 0.9665,
      "step": 7684
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0378308296203613,
      "learning_rate": 3.312715567074611e-05,
      "loss": 0.7407,
      "step": 7685
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.097079873085022,
      "learning_rate": 3.3104810860785993e-05,
      "loss": 0.8816,
      "step": 7686
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0564568042755127,
      "learning_rate": 3.3082472094427494e-05,
      "loss": 0.8362,
      "step": 7687
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.1680207252502441,
      "learning_rate": 3.3060139373688756e-05,
      "loss": 0.7163,
      "step": 7688
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.986595869064331,
      "learning_rate": 3.303781270058738e-05,
      "loss": 0.7215,
      "step": 7689
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9398879408836365,
      "learning_rate": 3.301549207714051e-05,
      "loss": 0.8275,
      "step": 7690
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9397369623184204,
      "learning_rate": 3.299317750536467e-05,
      "loss": 0.8014,
      "step": 7691
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.077769160270691,
      "learning_rate": 3.2970868987275825e-05,
      "loss": 0.7777,
      "step": 7692
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0061651468276978,
      "learning_rate": 3.294856652488948e-05,
      "loss": 0.9547,
      "step": 7693
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0514088869094849,
      "learning_rate": 3.292627012022049e-05,
      "loss": 1.0425,
      "step": 7694
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0771855115890503,
      "learning_rate": 3.2903979775283244e-05,
      "loss": 0.957,
      "step": 7695
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0071425437927246,
      "learning_rate": 3.288169549209149e-05,
      "loss": 0.9134,
      "step": 7696
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9473686814308167,
      "learning_rate": 3.2859417272658556e-05,
      "loss": 0.8254,
      "step": 7697
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9439135789871216,
      "learning_rate": 3.283714511899711e-05,
      "loss": 0.742,
      "step": 7698
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.301753044128418,
      "learning_rate": 3.281487903311932e-05,
      "loss": 0.861,
      "step": 7699
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9460572600364685,
      "learning_rate": 3.279261901703683e-05,
      "loss": 0.7992,
      "step": 7700
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.025091290473938,
      "learning_rate": 3.277036507276068e-05,
      "loss": 0.8681,
      "step": 7701
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9683058261871338,
      "learning_rate": 3.2748117202301364e-05,
      "loss": 0.7922,
      "step": 7702
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9131573438644409,
      "learning_rate": 3.27258754076689e-05,
      "loss": 0.7017,
      "step": 7703
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9854358434677124,
      "learning_rate": 3.270363969087271e-05,
      "loss": 0.8099,
      "step": 7704
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.8888185620307922,
      "learning_rate": 3.268141005392163e-05,
      "loss": 0.828,
      "step": 7705
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.1130229234695435,
      "learning_rate": 3.2659186498823966e-05,
      "loss": 0.7093,
      "step": 7706
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.1332718133926392,
      "learning_rate": 3.263696902758755e-05,
      "loss": 0.8494,
      "step": 7707
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9259494543075562,
      "learning_rate": 3.2614757642219575e-05,
      "loss": 0.8141,
      "step": 7708
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9104101061820984,
      "learning_rate": 3.259255234472668e-05,
      "loss": 0.864,
      "step": 7709
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9896255135536194,
      "learning_rate": 3.2570353137115085e-05,
      "loss": 0.9056,
      "step": 7710
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.1445319652557373,
      "learning_rate": 3.254816002139023e-05,
      "loss": 0.904,
      "step": 7711
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9868961572647095,
      "learning_rate": 3.252597299955721e-05,
      "loss": 0.9026,
      "step": 7712
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.0095990896224976,
      "learning_rate": 3.2503792073620557e-05,
      "loss": 0.8835,
      "step": 7713
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.8554115295410156,
      "learning_rate": 3.248161724558406e-05,
      "loss": 0.7983,
      "step": 7714
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.078898549079895,
      "learning_rate": 3.245944851745119e-05,
      "loss": 0.9255,
      "step": 7715
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.127135992050171,
      "learning_rate": 3.243728589122473e-05,
      "loss": 0.8126,
      "step": 7716
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9248011112213135,
      "learning_rate": 3.2415129368906925e-05,
      "loss": 0.7171,
      "step": 7717
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9495300054550171,
      "learning_rate": 3.239297895249955e-05,
      "loss": 0.8587,
      "step": 7718
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.125105619430542,
      "learning_rate": 3.2370834644003735e-05,
      "loss": 0.7936,
      "step": 7719
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.7966009974479675,
      "learning_rate": 3.234869644542009e-05,
      "loss": 0.7706,
      "step": 7720
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.8813653588294983,
      "learning_rate": 3.232656435874866e-05,
      "loss": 0.7009,
      "step": 7721
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.049356460571289,
      "learning_rate": 3.230443838598899e-05,
      "loss": 0.8723,
      "step": 7722
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9487774968147278,
      "learning_rate": 3.228231852914003e-05,
      "loss": 0.8226,
      "step": 7723
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9744672775268555,
      "learning_rate": 3.226020479020013e-05,
      "loss": 0.894,
      "step": 7724
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9349337220191956,
      "learning_rate": 3.2238097171167236e-05,
      "loss": 0.8119,
      "step": 7725
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.1085752248764038,
      "learning_rate": 3.221599567403859e-05,
      "loss": 0.9592,
      "step": 7726
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9792479872703552,
      "learning_rate": 3.219390030081091e-05,
      "loss": 0.7131,
      "step": 7727
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.1602656841278076,
      "learning_rate": 3.2171811053480447e-05,
      "loss": 0.9017,
      "step": 7728
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9393881559371948,
      "learning_rate": 3.2149727934042816e-05,
      "loss": 0.7762,
      "step": 7729
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.8422985076904297,
      "learning_rate": 3.2127650944493104e-05,
      "loss": 0.7676,
      "step": 7730
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9782288074493408,
      "learning_rate": 3.21055800868258e-05,
      "loss": 0.7589,
      "step": 7731
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9461728930473328,
      "learning_rate": 3.208351536303495e-05,
      "loss": 0.938,
      "step": 7732
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.0639896392822266,
      "learning_rate": 3.206145677511395e-05,
      "loss": 0.8675,
      "step": 7733
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9363032579421997,
      "learning_rate": 3.2039404325055624e-05,
      "loss": 0.7122,
      "step": 7734
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.0001636743545532,
      "learning_rate": 3.201735801485236e-05,
      "loss": 0.8037,
      "step": 7735
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.1112109422683716,
      "learning_rate": 3.199531784649589e-05,
      "loss": 0.7545,
      "step": 7736
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9331488013267517,
      "learning_rate": 3.1973283821977375e-05,
      "loss": 0.8296,
      "step": 7737
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.0405724048614502,
      "learning_rate": 3.1951255943287525e-05,
      "loss": 0.761,
      "step": 7738
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.0256736278533936,
      "learning_rate": 3.1929234212416414e-05,
      "loss": 0.7192,
      "step": 7739
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.0716664791107178,
      "learning_rate": 3.190721863135357e-05,
      "loss": 0.8446,
      "step": 7740
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.8863771557807922,
      "learning_rate": 3.188520920208796e-05,
      "loss": 0.7584,
      "step": 7741
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.024251103401184,
      "learning_rate": 3.1863205926608064e-05,
      "loss": 0.7456,
      "step": 7742
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9324749708175659,
      "learning_rate": 3.1841208806901723e-05,
      "loss": 0.8365,
      "step": 7743
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.8571458458900452,
      "learning_rate": 3.1819217844956214e-05,
      "loss": 0.5684,
      "step": 7744
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9923720955848694,
      "learning_rate": 3.179723304275841e-05,
      "loss": 0.7769,
      "step": 7745
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.8974207639694214,
      "learning_rate": 3.177525440229436e-05,
      "loss": 0.6983,
      "step": 7746
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.0157970190048218,
      "learning_rate": 3.1753281925549806e-05,
      "loss": 0.8361,
      "step": 7747
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.8553011417388916,
      "learning_rate": 3.173131561450986e-05,
      "loss": 0.7455,
      "step": 7748
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9296185374259949,
      "learning_rate": 3.170935547115896e-05,
      "loss": 0.7586,
      "step": 7749
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9083219766616821,
      "learning_rate": 3.168740149748116e-05,
      "loss": 0.7981,
      "step": 7750
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.947044312953949,
      "learning_rate": 3.166545369545986e-05,
      "loss": 0.7639,
      "step": 7751
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.0168018341064453,
      "learning_rate": 3.164351206707786e-05,
      "loss": 0.8325,
      "step": 7752
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9610939025878906,
      "learning_rate": 3.162157661431756e-05,
      "loss": 0.7578,
      "step": 7753
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.0094791650772095,
      "learning_rate": 3.1599647339160656e-05,
      "loss": 0.7559,
      "step": 7754
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9927675724029541,
      "learning_rate": 3.157772424358831e-05,
      "loss": 0.6803,
      "step": 7755
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9912436604499817,
      "learning_rate": 3.155580732958116e-05,
      "loss": 0.7178,
      "step": 7756
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.0584968328475952,
      "learning_rate": 3.1533896599119294e-05,
      "loss": 0.8434,
      "step": 7757
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9658499360084534,
      "learning_rate": 3.151199205418227e-05,
      "loss": 0.782,
      "step": 7758
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.974124014377594,
      "learning_rate": 3.149009369674891e-05,
      "loss": 0.8063,
      "step": 7759
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0381770133972168,
      "learning_rate": 3.146820152879774e-05,
      "loss": 0.8826,
      "step": 7760
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0948477983474731,
      "learning_rate": 3.144631555230652e-05,
      "loss": 0.9372,
      "step": 7761
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.00632905960083,
      "learning_rate": 3.14244357692525e-05,
      "loss": 0.8618,
      "step": 7762
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0492795705795288,
      "learning_rate": 3.140256218161248e-05,
      "loss": 1.0229,
      "step": 7763
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0523566007614136,
      "learning_rate": 3.1380694791362574e-05,
      "loss": 0.6893,
      "step": 7764
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.890155553817749,
      "learning_rate": 3.1358833600478366e-05,
      "loss": 0.7875,
      "step": 7765
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9153689742088318,
      "learning_rate": 3.1336978610934854e-05,
      "loss": 0.871,
      "step": 7766
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.8962323069572449,
      "learning_rate": 3.13151298247066e-05,
      "loss": 0.8319,
      "step": 7767
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0025181770324707,
      "learning_rate": 3.129328724376747e-05,
      "loss": 0.8146,
      "step": 7768
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9153618812561035,
      "learning_rate": 3.1271450870090794e-05,
      "loss": 0.9646,
      "step": 7769
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.1361137628555298,
      "learning_rate": 3.124962070564942e-05,
      "loss": 1.0801,
      "step": 7770
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0264286994934082,
      "learning_rate": 3.122779675241555e-05,
      "loss": 0.8301,
      "step": 7771
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9460259675979614,
      "learning_rate": 3.120597901236082e-05,
      "loss": 0.9595,
      "step": 7772
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9575111865997314,
      "learning_rate": 3.118416748745642e-05,
      "loss": 0.7663,
      "step": 7773
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9907057881355286,
      "learning_rate": 3.116236217967285e-05,
      "loss": 0.6793,
      "step": 7774
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.090024471282959,
      "learning_rate": 3.114056309098009e-05,
      "loss": 0.8412,
      "step": 7775
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.066033124923706,
      "learning_rate": 3.111877022334754e-05,
      "loss": 0.8021,
      "step": 7776
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0076799392700195,
      "learning_rate": 3.1096983578744135e-05,
      "loss": 0.8397,
      "step": 7777
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0081795454025269,
      "learning_rate": 3.1075203159138136e-05,
      "loss": 0.6691,
      "step": 7778
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9767215847969055,
      "learning_rate": 3.105342896649725e-05,
      "loss": 0.7953,
      "step": 7779
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9515060186386108,
      "learning_rate": 3.103166100278875e-05,
      "loss": 0.7639,
      "step": 7780
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.8694470524787903,
      "learning_rate": 3.10098992699791e-05,
      "loss": 0.7725,
      "step": 7781
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9532203674316406,
      "learning_rate": 3.098814377003445e-05,
      "loss": 0.9264,
      "step": 7782
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0332255363464355,
      "learning_rate": 3.0966394504920317e-05,
      "loss": 0.7379,
      "step": 7783
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9040732979774475,
      "learning_rate": 3.0944651476601514e-05,
      "loss": 0.82,
      "step": 7784
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9034704566001892,
      "learning_rate": 3.092291468704249e-05,
      "loss": 0.7021,
      "step": 7785
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9701902866363525,
      "learning_rate": 3.090118413820698e-05,
      "loss": 0.8983,
      "step": 7786
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9603883028030396,
      "learning_rate": 3.0879459832058275e-05,
      "loss": 0.7758,
      "step": 7787
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.8348279595375061,
      "learning_rate": 3.0857741770559025e-05,
      "loss": 0.7068,
      "step": 7788
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.021385908126831,
      "learning_rate": 3.0836029955671286e-05,
      "loss": 0.8074,
      "step": 7789
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9936757683753967,
      "learning_rate": 3.08143243893567e-05,
      "loss": 0.8646,
      "step": 7790
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.013936996459961,
      "learning_rate": 3.079262507357611e-05,
      "loss": 0.8782,
      "step": 7791
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9899959564208984,
      "learning_rate": 3.077093201028999e-05,
      "loss": 0.7816,
      "step": 7792
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9571762084960938,
      "learning_rate": 3.074924520145825e-05,
      "loss": 0.9315,
      "step": 7793
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9784216284751892,
      "learning_rate": 3.072756464904006e-05,
      "loss": 0.8097,
      "step": 7794
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9677906632423401,
      "learning_rate": 3.070589035499422e-05,
      "loss": 0.9012,
      "step": 7795
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9957283139228821,
      "learning_rate": 3.0684222321278825e-05,
      "loss": 0.8718,
      "step": 7796
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9356372356414795,
      "learning_rate": 3.066256054985146e-05,
      "loss": 0.7494,
      "step": 7797
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.0880048274993896,
      "learning_rate": 3.064090504266919e-05,
      "loss": 0.8242,
      "step": 7798
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.0468814373016357,
      "learning_rate": 3.061925580168843e-05,
      "loss": 0.8962,
      "step": 7799
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9572826623916626,
      "learning_rate": 3.0597612828865086e-05,
      "loss": 0.9395,
      "step": 7800
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.040055513381958,
      "learning_rate": 3.057597612615444e-05,
      "loss": 0.8238,
      "step": 7801
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.0333307981491089,
      "learning_rate": 3.0554345695511285e-05,
      "loss": 0.8394,
      "step": 7802
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.0509532690048218,
      "learning_rate": 3.053272153888982e-05,
      "loss": 0.8622,
      "step": 7803
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.0283617973327637,
      "learning_rate": 3.05111036582436e-05,
      "loss": 0.8605,
      "step": 7804
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9885609745979309,
      "learning_rate": 3.048949205552575e-05,
      "loss": 0.8786,
      "step": 7805
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.250054955482483,
      "learning_rate": 3.0467886732688722e-05,
      "loss": 0.9515,
      "step": 7806
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9774721264839172,
      "learning_rate": 3.0446287691684406e-05,
      "loss": 0.7565,
      "step": 7807
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9257248044013977,
      "learning_rate": 3.0424694934464227e-05,
      "loss": 0.7745,
      "step": 7808
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9825694561004639,
      "learning_rate": 3.0403108462978914e-05,
      "loss": 0.791,
      "step": 7809
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.0289263725280762,
      "learning_rate": 3.0381528279178706e-05,
      "loss": 0.7493,
      "step": 7810
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0324395895004272,
      "learning_rate": 3.03599543850132e-05,
      "loss": 0.8192,
      "step": 7811
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.1024703979492188,
      "learning_rate": 3.033838678243155e-05,
      "loss": 0.7256,
      "step": 7812
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.948047399520874,
      "learning_rate": 3.0316825473382227e-05,
      "loss": 0.7802,
      "step": 7813
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.02178156375885,
      "learning_rate": 3.0295270459813152e-05,
      "loss": 0.8537,
      "step": 7814
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9260463118553162,
      "learning_rate": 3.0273721743671745e-05,
      "loss": 0.8665,
      "step": 7815
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0703625679016113,
      "learning_rate": 3.0252179326904794e-05,
      "loss": 0.8492,
      "step": 7816
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9132555723190308,
      "learning_rate": 3.0230643211458498e-05,
      "loss": 0.789,
      "step": 7817
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0014147758483887,
      "learning_rate": 3.02091133992786e-05,
      "loss": 0.7995,
      "step": 7818
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0417718887329102,
      "learning_rate": 3.018758989231013e-05,
      "loss": 0.8715,
      "step": 7819
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.1779361963272095,
      "learning_rate": 3.016607269249765e-05,
      "loss": 0.8885,
      "step": 7820
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9315588474273682,
      "learning_rate": 3.0144561801785067e-05,
      "loss": 0.8422,
      "step": 7821
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9599744081497192,
      "learning_rate": 3.0123057222115836e-05,
      "loss": 0.9343,
      "step": 7822
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9746417999267578,
      "learning_rate": 3.0101558955432753e-05,
      "loss": 0.825,
      "step": 7823
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0187450647354126,
      "learning_rate": 3.0080067003678003e-05,
      "loss": 0.7454,
      "step": 7824
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0553938150405884,
      "learning_rate": 3.0058581368793394e-05,
      "loss": 0.8736,
      "step": 7825
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0891188383102417,
      "learning_rate": 3.003710205271988e-05,
      "loss": 0.9672,
      "step": 7826
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.1065350770950317,
      "learning_rate": 3.0015629057398063e-05,
      "loss": 0.8041,
      "step": 7827
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0310038328170776,
      "learning_rate": 2.9994162384767987e-05,
      "loss": 0.9414,
      "step": 7828
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9977184534072876,
      "learning_rate": 2.9972702036768895e-05,
      "loss": 0.8039,
      "step": 7829
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9130735993385315,
      "learning_rate": 2.9951248015339716e-05,
      "loss": 0.6527,
      "step": 7830
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.019484519958496,
      "learning_rate": 2.9929800322418656e-05,
      "loss": 0.7389,
      "step": 7831
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9089261889457703,
      "learning_rate": 2.9908358959943372e-05,
      "loss": 0.7772,
      "step": 7832
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0536820888519287,
      "learning_rate": 2.988692392985104e-05,
      "loss": 0.8583,
      "step": 7833
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.911730945110321,
      "learning_rate": 2.9865495234078135e-05,
      "loss": 0.734,
      "step": 7834
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9350784420967102,
      "learning_rate": 2.984407287456065e-05,
      "loss": 0.5422,
      "step": 7835
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9062708616256714,
      "learning_rate": 2.982265685323392e-05,
      "loss": 0.6798,
      "step": 7836
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9832141399383545,
      "learning_rate": 2.9801247172032833e-05,
      "loss": 0.7894,
      "step": 7837
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9334800839424133,
      "learning_rate": 2.9779843832891597e-05,
      "loss": 0.7549,
      "step": 7838
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9104373455047607,
      "learning_rate": 2.9758446837743857e-05,
      "loss": 0.7599,
      "step": 7839
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9861698746681213,
      "learning_rate": 2.973705618852277e-05,
      "loss": 0.8173,
      "step": 7840
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.1251966953277588,
      "learning_rate": 2.9715671887160823e-05,
      "loss": 0.6929,
      "step": 7841
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9798271656036377,
      "learning_rate": 2.9694293935589946e-05,
      "loss": 0.6609,
      "step": 7842
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9728927612304688,
      "learning_rate": 2.9672922335741582e-05,
      "loss": 0.9444,
      "step": 7843
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9364487528800964,
      "learning_rate": 2.9651557089546478e-05,
      "loss": 0.8314,
      "step": 7844
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.0102734565734863,
      "learning_rate": 2.9630198198934887e-05,
      "loss": 0.9747,
      "step": 7845
    },
    {
      "epoch": 2.96,
      "grad_norm": Infinity,
      "learning_rate": 2.9630198198934887e-05,
      "loss": 0.8039,
      "step": 7846
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9795091152191162,
      "learning_rate": 2.9608845665836428e-05,
      "loss": 0.9232,
      "step": 7847
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9266605973243713,
      "learning_rate": 2.958749949218024e-05,
      "loss": 0.7849,
      "step": 7848
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9341822266578674,
      "learning_rate": 2.956615967989479e-05,
      "loss": 0.8157,
      "step": 7849
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.0152347087860107,
      "learning_rate": 2.9544826230907985e-05,
      "loss": 0.7568,
      "step": 7850
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.007606029510498,
      "learning_rate": 2.952349914714726e-05,
      "loss": 0.7783,
      "step": 7851
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.13630211353302,
      "learning_rate": 2.950217843053934e-05,
      "loss": 0.7545,
      "step": 7852
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.030192494392395,
      "learning_rate": 2.9480864083010407e-05,
      "loss": 0.9237,
      "step": 7853
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.0209753513336182,
      "learning_rate": 2.9459556106486165e-05,
      "loss": 0.7991,
      "step": 7854
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.0387178659439087,
      "learning_rate": 2.9438254502891626e-05,
      "loss": 0.8891,
      "step": 7855
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.0339391231536865,
      "learning_rate": 2.9416959274151278e-05,
      "loss": 0.7507,
      "step": 7856
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.942881166934967,
      "learning_rate": 2.939567042218898e-05,
      "loss": 0.8252,
      "step": 7857
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9975326061248779,
      "learning_rate": 2.937438794892813e-05,
      "loss": 0.7556,
      "step": 7858
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.1567434072494507,
      "learning_rate": 2.935311185629146e-05,
      "loss": 0.8249,
      "step": 7859
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.0078626871109009,
      "learning_rate": 2.9331842146201083e-05,
      "loss": 0.8609,
      "step": 7860
    },
    {
      "epoch": 2.96,
      "eval_loss": 0.9807450175285339,
      "eval_runtime": 17.7684,
      "eval_samples_per_second": 153.587,
      "eval_steps_per_second": 38.439,
      "step": 7860
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.062860369682312,
      "learning_rate": 2.9310578820578726e-05,
      "loss": 0.8314,
      "step": 7861
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.8805136680603027,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 0.8393,
      "step": 7862
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9360345602035522,
      "learning_rate": 2.92680713304212e-05,
      "loss": 0.6865,
      "step": 7863
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.0126088857650757,
      "learning_rate": 2.9246827169726466e-05,
      "loss": 0.8536,
      "step": 7864
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9170997738838196,
      "learning_rate": 2.9225589401180242e-05,
      "loss": 0.8483,
      "step": 7865
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.2047983407974243,
      "learning_rate": 2.9204358026701317e-05,
      "loss": 0.8257,
      "step": 7866
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9925597906112671,
      "learning_rate": 2.91831330482078e-05,
      "loss": 0.7623,
      "step": 7867
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.0094153881072998,
      "learning_rate": 2.9161914467617225e-05,
      "loss": 0.7209,
      "step": 7868
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.136904239654541,
      "learning_rate": 2.9140702286846623e-05,
      "loss": 0.7682,
      "step": 7869
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9452815055847168,
      "learning_rate": 2.911949650781236e-05,
      "loss": 0.7995,
      "step": 7870
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.024186372756958,
      "learning_rate": 2.9098297132430265e-05,
      "loss": 0.8549,
      "step": 7871
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.1391983032226562,
      "learning_rate": 2.9077104162615544e-05,
      "loss": 0.9441,
      "step": 7872
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.920432448387146,
      "learning_rate": 2.9055917600282933e-05,
      "loss": 0.7391,
      "step": 7873
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.1416221857070923,
      "learning_rate": 2.903473744734648e-05,
      "loss": 0.9024,
      "step": 7874
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.0054394006729126,
      "learning_rate": 2.901356370571967e-05,
      "loss": 0.8802,
      "step": 7875
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9628970623016357,
      "learning_rate": 2.8992396377315488e-05,
      "loss": 0.8882,
      "step": 7876
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.8088098764419556,
      "learning_rate": 2.897123546404624e-05,
      "loss": 0.743,
      "step": 7877
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9848136901855469,
      "learning_rate": 2.895008096782368e-05,
      "loss": 0.7384,
      "step": 7878
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.069791316986084,
      "learning_rate": 2.8928932890559047e-05,
      "loss": 0.9572,
      "step": 7879
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9519852995872498,
      "learning_rate": 2.890779123416293e-05,
      "loss": 0.816,
      "step": 7880
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9871480464935303,
      "learning_rate": 2.8886656000545342e-05,
      "loss": 0.7453,
      "step": 7881
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9371827244758606,
      "learning_rate": 2.8865527191615716e-05,
      "loss": 0.6879,
      "step": 7882
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.038935661315918,
      "learning_rate": 2.8844404809282978e-05,
      "loss": 0.8275,
      "step": 7883
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9645636677742004,
      "learning_rate": 2.882328885545539e-05,
      "loss": 0.8081,
      "step": 7884
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.0269420146942139,
      "learning_rate": 2.8802179332040615e-05,
      "loss": 0.7397,
      "step": 7885
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.91978520154953,
      "learning_rate": 2.8781076240945836e-05,
      "loss": 0.7835,
      "step": 7886
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9566662907600403,
      "learning_rate": 2.875997958407759e-05,
      "loss": 0.8303,
      "step": 7887
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.030970573425293,
      "learning_rate": 2.8738889363341803e-05,
      "loss": 0.8084,
      "step": 7888
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0632834434509277,
      "learning_rate": 2.8717805580643896e-05,
      "loss": 0.8508,
      "step": 7889
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.8999619483947754,
      "learning_rate": 2.869672823788867e-05,
      "loss": 0.8523,
      "step": 7890
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.036726713180542,
      "learning_rate": 2.867565733698033e-05,
      "loss": 0.662,
      "step": 7891
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.04769766330719,
      "learning_rate": 2.8654592879822484e-05,
      "loss": 0.8417,
      "step": 7892
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0014748573303223,
      "learning_rate": 2.8633534868318234e-05,
      "loss": 0.7431,
      "step": 7893
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0378514528274536,
      "learning_rate": 2.8612483304370042e-05,
      "loss": 0.9826,
      "step": 7894
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9022136926651001,
      "learning_rate": 2.8591438189879748e-05,
      "loss": 0.8215,
      "step": 7895
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.11886727809906,
      "learning_rate": 2.857039952674877e-05,
      "loss": 0.9066,
      "step": 7896
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9915139079093933,
      "learning_rate": 2.8549367316877697e-05,
      "loss": 0.9116,
      "step": 7897
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.1565651893615723,
      "learning_rate": 2.852834156216674e-05,
      "loss": 0.7037,
      "step": 7898
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0059210062026978,
      "learning_rate": 2.8507322264515502e-05,
      "loss": 0.8627,
      "step": 7899
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.8295075297355652,
      "learning_rate": 2.8486309425822856e-05,
      "loss": 0.793,
      "step": 7900
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9874873757362366,
      "learning_rate": 2.8465303047987267e-05,
      "loss": 0.9042,
      "step": 7901
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.987204909324646,
      "learning_rate": 2.8444303132906513e-05,
      "loss": 0.6511,
      "step": 7902
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.1425936222076416,
      "learning_rate": 2.8423309682477804e-05,
      "loss": 1.0602,
      "step": 7903
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.1023626327514648,
      "learning_rate": 2.8402322698597827e-05,
      "loss": 0.8942,
      "step": 7904
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0934299230575562,
      "learning_rate": 2.8381342183162608e-05,
      "loss": 0.8911,
      "step": 7905
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.047400712966919,
      "learning_rate": 2.8360368138067627e-05,
      "loss": 0.8076,
      "step": 7906
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0535387992858887,
      "learning_rate": 2.833940056520772e-05,
      "loss": 0.8008,
      "step": 7907
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0056748390197754,
      "learning_rate": 2.8318439466477263e-05,
      "loss": 0.7463,
      "step": 7908
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.132751226425171,
      "learning_rate": 2.829748484376995e-05,
      "loss": 0.979,
      "step": 7909
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.916599690914154,
      "learning_rate": 2.827653669897886e-05,
      "loss": 1.0211,
      "step": 7910
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9852902293205261,
      "learning_rate": 2.825559503399663e-05,
      "loss": 0.8493,
      "step": 7911
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9834513664245605,
      "learning_rate": 2.823465985071515e-05,
      "loss": 0.7369,
      "step": 7912
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0297553539276123,
      "learning_rate": 2.82137311510258e-05,
      "loss": 0.7365,
      "step": 7913
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.1064317226409912,
      "learning_rate": 2.8192808936819426e-05,
      "loss": 0.7336,
      "step": 7914
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.0252870321273804,
      "learning_rate": 2.8171893209986177e-05,
      "loss": 0.9184,
      "step": 7915
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9886939525604248,
      "learning_rate": 2.81509839724157e-05,
      "loss": 0.8178,
      "step": 7916
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9476537108421326,
      "learning_rate": 2.8130081225996974e-05,
      "loss": 0.7419,
      "step": 7917
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9663602113723755,
      "learning_rate": 2.810918497261852e-05,
      "loss": 0.7896,
      "step": 7918
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.045780897140503,
      "learning_rate": 2.8088295214168147e-05,
      "loss": 0.6423,
      "step": 7919
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9942068457603455,
      "learning_rate": 2.806741195253312e-05,
      "loss": 1.0146,
      "step": 7920
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9563878178596497,
      "learning_rate": 2.8046535189600166e-05,
      "loss": 0.7833,
      "step": 7921
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9429323673248291,
      "learning_rate": 2.802566492725537e-05,
      "loss": 0.76,
      "step": 7922
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.932558536529541,
      "learning_rate": 2.8004801167384186e-05,
      "loss": 0.7937,
      "step": 7923
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.8561239242553711,
      "learning_rate": 2.7983943911871614e-05,
      "loss": 0.7019,
      "step": 7924
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.0463411808013916,
      "learning_rate": 2.796309316260196e-05,
      "loss": 0.8701,
      "step": 7925
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.936529278755188,
      "learning_rate": 2.7942248921458957e-05,
      "loss": 0.9756,
      "step": 7926
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.0102530717849731,
      "learning_rate": 2.7921411190325752e-05,
      "loss": 0.9135,
      "step": 7927
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9837690591812134,
      "learning_rate": 2.7900579971084972e-05,
      "loss": 0.9447,
      "step": 7928
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9683241844177246,
      "learning_rate": 2.7879755265618555e-05,
      "loss": 0.8231,
      "step": 7929
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.8771883249282837,
      "learning_rate": 2.785893707580789e-05,
      "loss": 0.784,
      "step": 7930
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.1681004762649536,
      "learning_rate": 2.7838125403533854e-05,
      "loss": 0.809,
      "step": 7931
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9769678115844727,
      "learning_rate": 2.7817320250676538e-05,
      "loss": 0.7648,
      "step": 7932
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.0148745775222778,
      "learning_rate": 2.7796521619115644e-05,
      "loss": 0.7919,
      "step": 7933
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.0846303701400757,
      "learning_rate": 2.7775729510730274e-05,
      "loss": 0.8266,
      "step": 7934
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.0771740674972534,
      "learning_rate": 2.7754943927398747e-05,
      "loss": 0.7788,
      "step": 7935
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9924818277359009,
      "learning_rate": 2.7734164870999014e-05,
      "loss": 0.7903,
      "step": 7936
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.150328278541565,
      "learning_rate": 2.771339234340832e-05,
      "loss": 0.7434,
      "step": 7937
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.939105749130249,
      "learning_rate": 2.7692626346503304e-05,
      "loss": 0.9171,
      "step": 7938
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9367903470993042,
      "learning_rate": 2.7671866882160134e-05,
      "loss": 0.8812,
      "step": 7939
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.0428826808929443,
      "learning_rate": 2.765111395225424e-05,
      "loss": 0.8777,
      "step": 7940
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0221951007843018,
      "learning_rate": 2.763036755866061e-05,
      "loss": 0.9882,
      "step": 7941
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9698917269706726,
      "learning_rate": 2.760962770325346e-05,
      "loss": 0.7228,
      "step": 7942
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0437679290771484,
      "learning_rate": 2.7588894387906585e-05,
      "loss": 0.8834,
      "step": 7943
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8600085377693176,
      "learning_rate": 2.7568167614493158e-05,
      "loss": 0.7781,
      "step": 7944
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9688100218772888,
      "learning_rate": 2.7547447384885627e-05,
      "loss": 0.8397,
      "step": 7945
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.173047661781311,
      "learning_rate": 2.7526733700956033e-05,
      "loss": 0.9385,
      "step": 7946
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9797694683074951,
      "learning_rate": 2.7506026564575694e-05,
      "loss": 0.7222,
      "step": 7947
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9776696562767029,
      "learning_rate": 2.7485325977615373e-05,
      "loss": 0.682,
      "step": 7948
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9774590134620667,
      "learning_rate": 2.7464631941945306e-05,
      "loss": 0.9819,
      "step": 7949
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.900507926940918,
      "learning_rate": 2.7443944459435045e-05,
      "loss": 0.6877,
      "step": 7950
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0018991231918335,
      "learning_rate": 2.7423263531953592e-05,
      "loss": 0.7992,
      "step": 7951
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0273499488830566,
      "learning_rate": 2.7402589161369318e-05,
      "loss": 0.9862,
      "step": 7952
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0104960203170776,
      "learning_rate": 2.7381921349550098e-05,
      "loss": 0.8221,
      "step": 7953
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8761460185050964,
      "learning_rate": 2.7361260098363116e-05,
      "loss": 0.8058,
      "step": 7954
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0159964561462402,
      "learning_rate": 2.734060540967499e-05,
      "loss": 0.929,
      "step": 7955
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.067510962486267,
      "learning_rate": 2.731995728535178e-05,
      "loss": 0.7421,
      "step": 7956
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.1651482582092285,
      "learning_rate": 2.7299315727258933e-05,
      "loss": 0.9555,
      "step": 7957
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.047707438468933,
      "learning_rate": 2.727868073726124e-05,
      "loss": 0.8832,
      "step": 7958
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.2311276197433472,
      "learning_rate": 2.7258052317223025e-05,
      "loss": 0.8284,
      "step": 7959
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0092966556549072,
      "learning_rate": 2.723743046900793e-05,
      "loss": 0.7252,
      "step": 7960
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9509356617927551,
      "learning_rate": 2.7216815194479006e-05,
      "loss": 0.7634,
      "step": 7961
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0490330457687378,
      "learning_rate": 2.71962064954987e-05,
      "loss": 0.7566,
      "step": 7962
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.1528916358947754,
      "learning_rate": 2.717560437392894e-05,
      "loss": 0.798,
      "step": 7963
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.2248393297195435,
      "learning_rate": 2.7155008831631014e-05,
      "loss": 0.9466,
      "step": 7964
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9904489517211914,
      "learning_rate": 2.7134419870465556e-05,
      "loss": 0.9531,
      "step": 7965
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9626322984695435,
      "learning_rate": 2.711383749229276e-05,
      "loss": 0.7947,
      "step": 7966
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9614479541778564,
      "learning_rate": 2.7093261698972005e-05,
      "loss": 0.8558,
      "step": 7967
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.983052670955658,
      "learning_rate": 2.7072692492362262e-05,
      "loss": 0.8654,
      "step": 7968
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9287702441215515,
      "learning_rate": 2.7052129874321863e-05,
      "loss": 0.7081,
      "step": 7969
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0216176509857178,
      "learning_rate": 2.7031573846708503e-05,
      "loss": 0.7837,
      "step": 7970
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0010219812393188,
      "learning_rate": 2.7011024411379303e-05,
      "loss": 0.8028,
      "step": 7971
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9296990036964417,
      "learning_rate": 2.6990481570190752e-05,
      "loss": 0.7227,
      "step": 7972
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8896356225013733,
      "learning_rate": 2.6969945324998834e-05,
      "loss": 0.7893,
      "step": 7973
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.018528699874878,
      "learning_rate": 2.6949415677658875e-05,
      "loss": 0.8802,
      "step": 7974
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0096875429153442,
      "learning_rate": 2.692889263002556e-05,
      "loss": 1.0419,
      "step": 7975
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9647845029830933,
      "learning_rate": 2.6908376183953133e-05,
      "loss": 0.9225,
      "step": 7976
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0570937395095825,
      "learning_rate": 2.688786634129501e-05,
      "loss": 0.9227,
      "step": 7977
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0197943449020386,
      "learning_rate": 2.68673631039042e-05,
      "loss": 0.864,
      "step": 7978
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9794496893882751,
      "learning_rate": 2.6846866473633125e-05,
      "loss": 0.7179,
      "step": 7979
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.003331184387207,
      "learning_rate": 2.6826376452333414e-05,
      "loss": 0.968,
      "step": 7980
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.1445320844650269,
      "learning_rate": 2.68058930418563e-05,
      "loss": 0.8471,
      "step": 7981
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0460976362228394,
      "learning_rate": 2.678541624405234e-05,
      "loss": 0.8627,
      "step": 7982
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9557591676712036,
      "learning_rate": 2.6764946060771444e-05,
      "loss": 0.7688,
      "step": 7983
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9907974600791931,
      "learning_rate": 2.6744482493863043e-05,
      "loss": 0.8528,
      "step": 7984
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.045852541923523,
      "learning_rate": 2.672402554517589e-05,
      "loss": 0.869,
      "step": 7985
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0354161262512207,
      "learning_rate": 2.670357521655813e-05,
      "loss": 0.8685,
      "step": 7986
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9614201784133911,
      "learning_rate": 2.6683131509857317e-05,
      "loss": 0.8454,
      "step": 7987
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9889244437217712,
      "learning_rate": 2.6662694426920487e-05,
      "loss": 0.7511,
      "step": 7988
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0666682720184326,
      "learning_rate": 2.6642263969593983e-05,
      "loss": 0.8812,
      "step": 7989
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8674406409263611,
      "learning_rate": 2.6621840139723542e-05,
      "loss": 0.8711,
      "step": 7990
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9273693561553955,
      "learning_rate": 2.6601422939154407e-05,
      "loss": 0.8599,
      "step": 7991
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9630395770072937,
      "learning_rate": 2.658101236973114e-05,
      "loss": 0.7526,
      "step": 7992
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.034271001815796,
      "learning_rate": 2.656060843329766e-05,
      "loss": 0.9276,
      "step": 7993
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0980192422866821,
      "learning_rate": 2.6540211131697446e-05,
      "loss": 0.7594,
      "step": 7994
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0123710632324219,
      "learning_rate": 2.6519820466773214e-05,
      "loss": 0.935,
      "step": 7995
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9829866290092468,
      "learning_rate": 2.6499436440367165e-05,
      "loss": 0.7288,
      "step": 7996
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0120404958724976,
      "learning_rate": 2.6479059054320843e-05,
      "loss": 0.8496,
      "step": 7997
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0722243785858154,
      "learning_rate": 2.6458688310475287e-05,
      "loss": 0.9978,
      "step": 7998
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9350081086158752,
      "learning_rate": 2.6438324210670863e-05,
      "loss": 0.7417,
      "step": 7999
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.000784158706665,
      "learning_rate": 2.6417966756747303e-05,
      "loss": 0.9489,
      "step": 8000
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9217638969421387,
      "learning_rate": 2.6397615950543854e-05,
      "loss": 0.8366,
      "step": 8001
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.081331729888916,
      "learning_rate": 2.637727179389907e-05,
      "loss": 0.8578,
      "step": 8002
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0921717882156372,
      "learning_rate": 2.6356934288650903e-05,
      "loss": 1.0387,
      "step": 8003
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9894843101501465,
      "learning_rate": 2.6336603436636786e-05,
      "loss": 0.9828,
      "step": 8004
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.8409438133239746,
      "learning_rate": 2.6316279239693465e-05,
      "loss": 0.7992,
      "step": 8005
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9000970721244812,
      "learning_rate": 2.6295961699657112e-05,
      "loss": 0.7412,
      "step": 8006
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0334932804107666,
      "learning_rate": 2.6275650818363294e-05,
      "loss": 0.8841,
      "step": 8007
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8540332317352295,
      "learning_rate": 2.625534659764701e-05,
      "loss": 0.8148,
      "step": 8008
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8911861777305603,
      "learning_rate": 2.623504903934264e-05,
      "loss": 0.6785,
      "step": 8009
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9815496206283569,
      "learning_rate": 2.621475814528389e-05,
      "loss": 0.7354,
      "step": 8010
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9368183016777039,
      "learning_rate": 2.619447391730403e-05,
      "loss": 0.6508,
      "step": 8011
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8219999074935913,
      "learning_rate": 2.6174196357235515e-05,
      "loss": 0.7268,
      "step": 8012
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.880169153213501,
      "learning_rate": 2.6153925466910346e-05,
      "loss": 0.8781,
      "step": 8013
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9440558552742004,
      "learning_rate": 2.613366124815997e-05,
      "loss": 0.7267,
      "step": 8014
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8765121102333069,
      "learning_rate": 2.6113403702814998e-05,
      "loss": 0.5788,
      "step": 8015
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9855409860610962,
      "learning_rate": 2.609315283270569e-05,
      "loss": 0.78,
      "step": 8016
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9116550087928772,
      "learning_rate": 2.6072908639661563e-05,
      "loss": 0.7349,
      "step": 8017
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0588949918746948,
      "learning_rate": 2.605267112551154e-05,
      "loss": 0.8308,
      "step": 8018
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9325016140937805,
      "learning_rate": 2.6032440292084026e-05,
      "loss": 0.7554,
      "step": 8019
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0242242813110352,
      "learning_rate": 2.6012216141206723e-05,
      "loss": 0.76,
      "step": 8020
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.89713454246521,
      "learning_rate": 2.5991998674706773e-05,
      "loss": 0.6944,
      "step": 8021
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9252524971961975,
      "learning_rate": 2.5971787894410684e-05,
      "loss": 0.8082,
      "step": 8022
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9150708317756653,
      "learning_rate": 2.595158380214444e-05,
      "loss": 0.7163,
      "step": 8023
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8370439410209656,
      "learning_rate": 2.5931386399733336e-05,
      "loss": 0.6609,
      "step": 8024
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7891771793365479,
      "learning_rate": 2.5911195689002064e-05,
      "loss": 0.685,
      "step": 8025
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0361526012420654,
      "learning_rate": 2.589101167177481e-05,
      "loss": 0.7476,
      "step": 8026
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8735950589179993,
      "learning_rate": 2.587083434987505e-05,
      "loss": 0.667,
      "step": 8027
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.890738308429718,
      "learning_rate": 2.5850663725125646e-05,
      "loss": 0.6981,
      "step": 8028
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9237583875656128,
      "learning_rate": 2.5830499799348973e-05,
      "loss": 0.7732,
      "step": 8029
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9599441885948181,
      "learning_rate": 2.581034257436671e-05,
      "loss": 0.7914,
      "step": 8030
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0949323177337646,
      "learning_rate": 2.579019205199992e-05,
      "loss": 0.6668,
      "step": 8031
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9242334961891174,
      "learning_rate": 2.5770048234069078e-05,
      "loss": 0.7198,
      "step": 8032
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8583904504776001,
      "learning_rate": 2.5749911122394122e-05,
      "loss": 0.6114,
      "step": 8033
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9906371235847473,
      "learning_rate": 2.5729780718794294e-05,
      "loss": 0.5988,
      "step": 8034
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8994998335838318,
      "learning_rate": 2.570965702508823e-05,
      "loss": 0.659,
      "step": 8035
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9113233685493469,
      "learning_rate": 2.5689540043094053e-05,
      "loss": 0.7409,
      "step": 8036
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9642217755317688,
      "learning_rate": 2.5669429774629182e-05,
      "loss": 0.7677,
      "step": 8037
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9988697171211243,
      "learning_rate": 2.564932622151045e-05,
      "loss": 0.7387,
      "step": 8038
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9452222585678101,
      "learning_rate": 2.5629229385554142e-05,
      "loss": 0.7943,
      "step": 8039
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8983279466629028,
      "learning_rate": 2.5609139268575887e-05,
      "loss": 0.6106,
      "step": 8040
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.1064125299453735,
      "learning_rate": 2.5589055872390687e-05,
      "loss": 0.6748,
      "step": 8041
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9767341017723083,
      "learning_rate": 2.556897919881295e-05,
      "loss": 0.8871,
      "step": 8042
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0216561555862427,
      "learning_rate": 2.554890924965655e-05,
      "loss": 0.6963,
      "step": 8043
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9855220913887024,
      "learning_rate": 2.5528846026734655e-05,
      "loss": 0.7099,
      "step": 8044
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.8925521969795227,
      "learning_rate": 2.5508789531859833e-05,
      "loss": 0.6253,
      "step": 8045
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.99262535572052,
      "learning_rate": 2.548873976684417e-05,
      "loss": 0.8307,
      "step": 8046
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0256638526916504,
      "learning_rate": 2.546869673349892e-05,
      "loss": 0.7035,
      "step": 8047
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.081619143486023,
      "learning_rate": 2.5448660433634943e-05,
      "loss": 0.8618,
      "step": 8048
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0850001573562622,
      "learning_rate": 2.5428630869062432e-05,
      "loss": 0.7361,
      "step": 8049
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0508164167404175,
      "learning_rate": 2.5408608041590844e-05,
      "loss": 0.6631,
      "step": 8050
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0052679777145386,
      "learning_rate": 2.538859195302922e-05,
      "loss": 0.8446,
      "step": 8051
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0153065919876099,
      "learning_rate": 2.5368582605185854e-05,
      "loss": 0.7123,
      "step": 8052
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.293020248413086,
      "learning_rate": 2.534857999986847e-05,
      "loss": 0.8768,
      "step": 8053
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9398112893104553,
      "learning_rate": 2.5328584138884224e-05,
      "loss": 0.5715,
      "step": 8054
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0233824253082275,
      "learning_rate": 2.530859502403963e-05,
      "loss": 0.7703,
      "step": 8055
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.950976550579071,
      "learning_rate": 2.5288612657140554e-05,
      "loss": 0.6238,
      "step": 8056
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9790272116661072,
      "learning_rate": 2.5268637039992293e-05,
      "loss": 0.7897,
      "step": 8057
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9668529033660889,
      "learning_rate": 2.524866817439958e-05,
      "loss": 0.7398,
      "step": 8058
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0357766151428223,
      "learning_rate": 2.5228706062166453e-05,
      "loss": 0.8221,
      "step": 8059
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9277655482292175,
      "learning_rate": 2.520875070509634e-05,
      "loss": 0.5306,
      "step": 8060
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9563776850700378,
      "learning_rate": 2.518880210499217e-05,
      "loss": 0.7841,
      "step": 8061
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9881957173347473,
      "learning_rate": 2.516886026365616e-05,
      "loss": 0.9117,
      "step": 8062
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9559555053710938,
      "learning_rate": 2.514892518288988e-05,
      "loss": 0.7256,
      "step": 8063
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.1902036666870117,
      "learning_rate": 2.512899686449446e-05,
      "loss": 0.7106,
      "step": 8064
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0802545547485352,
      "learning_rate": 2.5109075310270237e-05,
      "loss": 0.7911,
      "step": 8065
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.9147621393203735,
      "learning_rate": 2.508916052201703e-05,
      "loss": 0.6567,
      "step": 8066
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.2552417516708374,
      "learning_rate": 2.5069252501534e-05,
      "loss": 0.7036,
      "step": 8067
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.8978458046913147,
      "learning_rate": 2.504935125061978e-05,
      "loss": 0.6869,
      "step": 8068
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.044152021408081,
      "learning_rate": 2.50294567710723e-05,
      "loss": 0.7955,
      "step": 8069
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0546258687973022,
      "learning_rate": 2.50095690646889e-05,
      "loss": 0.8624,
      "step": 8070
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.995029091835022,
      "learning_rate": 2.4989688133266366e-05,
      "loss": 0.5802,
      "step": 8071
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.9530028104782104,
      "learning_rate": 2.4969813978600798e-05,
      "loss": 0.6684,
      "step": 8072
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.956859827041626,
      "learning_rate": 2.4949946602487683e-05,
      "loss": 0.6876,
      "step": 8073
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.086984634399414,
      "learning_rate": 2.4930086006721998e-05,
      "loss": 0.8746,
      "step": 8074
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.1155555248260498,
      "learning_rate": 2.4910232193097994e-05,
      "loss": 0.7205,
      "step": 8075
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0337733030319214,
      "learning_rate": 2.4890385163409356e-05,
      "loss": 0.8114,
      "step": 8076
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.074081301689148,
      "learning_rate": 2.4870544919449113e-05,
      "loss": 0.7197,
      "step": 8077
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.086875557899475,
      "learning_rate": 2.485071146300979e-05,
      "loss": 0.7494,
      "step": 8078
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0329999923706055,
      "learning_rate": 2.483088479588319e-05,
      "loss": 0.7762,
      "step": 8079
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.9137189388275146,
      "learning_rate": 2.4811064919860517e-05,
      "loss": 0.6833,
      "step": 8080
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0238802433013916,
      "learning_rate": 2.4791251836732464e-05,
      "loss": 0.6937,
      "step": 8081
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0225166082382202,
      "learning_rate": 2.477144554828893e-05,
      "loss": 0.7475,
      "step": 8082
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0679553747177124,
      "learning_rate": 2.4751646056319334e-05,
      "loss": 0.7664,
      "step": 8083
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.071401596069336,
      "learning_rate": 2.4731853362612522e-05,
      "loss": 0.7482,
      "step": 8084
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0860387086868286,
      "learning_rate": 2.4712067468956535e-05,
      "loss": 0.868,
      "step": 8085
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0081230401992798,
      "learning_rate": 2.4692288377139005e-05,
      "loss": 0.7201,
      "step": 8086
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.9174724817276001,
      "learning_rate": 2.467251608894683e-05,
      "loss": 0.705,
      "step": 8087
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.9237522482872009,
      "learning_rate": 2.4652750606166286e-05,
      "loss": 0.6706,
      "step": 8088
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.8903366327285767,
      "learning_rate": 2.463299193058315e-05,
      "loss": 0.6837,
      "step": 8089
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0237739086151123,
      "learning_rate": 2.4613240063982456e-05,
      "loss": 0.7245,
      "step": 8090
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.9601891040802002,
      "learning_rate": 2.459349500814868e-05,
      "loss": 0.648,
      "step": 8091
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.1389763355255127,
      "learning_rate": 2.4573756764865664e-05,
      "loss": 0.7162,
      "step": 8092
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.9855803847312927,
      "learning_rate": 2.4554025335916652e-05,
      "loss": 0.7631,
      "step": 8093
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.1630808115005493,
      "learning_rate": 2.4534300723084346e-05,
      "loss": 0.7188,
      "step": 8094
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.9986269474029541,
      "learning_rate": 2.451458292815062e-05,
      "loss": 0.7144,
      "step": 8095
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.0782290697097778,
      "learning_rate": 2.4494871952896947e-05,
      "loss": 0.6935,
      "step": 8096
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.0699889659881592,
      "learning_rate": 2.4475167799104105e-05,
      "loss": 0.7349,
      "step": 8097
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.0597455501556396,
      "learning_rate": 2.445547046855219e-05,
      "loss": 0.9119,
      "step": 8098
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.894646406173706,
      "learning_rate": 2.443577996302081e-05,
      "loss": 0.6899,
      "step": 8099
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.8739500045776367,
      "learning_rate": 2.441609628428888e-05,
      "loss": 0.6192,
      "step": 8100
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.0510575771331787,
      "learning_rate": 2.4396419434134687e-05,
      "loss": 0.659,
      "step": 8101
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.963117778301239,
      "learning_rate": 2.4376749414335908e-05,
      "loss": 0.79,
      "step": 8102
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.1128013134002686,
      "learning_rate": 2.435708622666967e-05,
      "loss": 0.7671,
      "step": 8103
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.016755223274231,
      "learning_rate": 2.4337429872912397e-05,
      "loss": 0.7767,
      "step": 8104
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9316563606262207,
      "learning_rate": 2.4317780354839914e-05,
      "loss": 0.6054,
      "step": 8105
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9017807841300964,
      "learning_rate": 2.4298137674227484e-05,
      "loss": 0.6065,
      "step": 8106
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9984809756278992,
      "learning_rate": 2.42785018328497e-05,
      "loss": 0.7664,
      "step": 8107
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9379798173904419,
      "learning_rate": 2.4258872832480507e-05,
      "loss": 0.6929,
      "step": 8108
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.8566086888313293,
      "learning_rate": 2.4239250674893345e-05,
      "loss": 0.681,
      "step": 8109
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.1864845752716064,
      "learning_rate": 2.4219635361860915e-05,
      "loss": 0.7157,
      "step": 8110
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9995941519737244,
      "learning_rate": 2.420002689515537e-05,
      "loss": 0.7106,
      "step": 8111
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9735347032546997,
      "learning_rate": 2.418042527654819e-05,
      "loss": 0.7568,
      "step": 8112
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.1523607969284058,
      "learning_rate": 2.416083050781033e-05,
      "loss": 0.6933,
      "step": 8113
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9361030459403992,
      "learning_rate": 2.4141242590712023e-05,
      "loss": 0.7024,
      "step": 8114
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.0254050493240356,
      "learning_rate": 2.412166152702291e-05,
      "loss": 0.758,
      "step": 8115
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9816838502883911,
      "learning_rate": 2.4102087318512122e-05,
      "loss": 0.6979,
      "step": 8116
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.2214492559432983,
      "learning_rate": 2.4082519966947948e-05,
      "loss": 0.7578,
      "step": 8117
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9360090494155884,
      "learning_rate": 2.4062959474098257e-05,
      "loss": 0.5617,
      "step": 8118
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.072795033454895,
      "learning_rate": 2.404340584173027e-05,
      "loss": 0.7889,
      "step": 8119
    },
    {
      "epoch": 3.04,
      "grad_norm": 1.0196350812911987,
      "learning_rate": 2.4023859071610456e-05,
      "loss": 0.835,
      "step": 8120
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.964723527431488,
      "learning_rate": 2.400431916550482e-05,
      "loss": 0.5792,
      "step": 8121
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.9333589673042297,
      "learning_rate": 2.3984786125178628e-05,
      "loss": 0.8131,
      "step": 8122
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.1545500755310059,
      "learning_rate": 2.3965259952396646e-05,
      "loss": 0.5291,
      "step": 8123
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9813075661659241,
      "learning_rate": 2.394574064892292e-05,
      "loss": 0.6772,
      "step": 8124
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.0234146118164062,
      "learning_rate": 2.3926228216520874e-05,
      "loss": 0.6749,
      "step": 8125
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9401314854621887,
      "learning_rate": 2.390672265695345e-05,
      "loss": 0.7176,
      "step": 8126
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9289931654930115,
      "learning_rate": 2.388722397198273e-05,
      "loss": 0.5677,
      "step": 8127
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9933103322982788,
      "learning_rate": 2.386773216337037e-05,
      "loss": 0.5898,
      "step": 8128
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.1409776210784912,
      "learning_rate": 2.384824723287741e-05,
      "loss": 0.6539,
      "step": 8129
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.0672084093093872,
      "learning_rate": 2.382876918226409e-05,
      "loss": 0.8329,
      "step": 8130
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9981968402862549,
      "learning_rate": 2.3809298013290215e-05,
      "loss": 0.7065,
      "step": 8131
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.1193089485168457,
      "learning_rate": 2.3789833727714883e-05,
      "loss": 0.7507,
      "step": 8132
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.978188157081604,
      "learning_rate": 2.377037632729654e-05,
      "loss": 0.7365,
      "step": 8133
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.0872188806533813,
      "learning_rate": 2.3750925813793113e-05,
      "loss": 0.7797,
      "step": 8134
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9092377424240112,
      "learning_rate": 2.3731482188961818e-05,
      "loss": 0.6338,
      "step": 8135
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.0486384630203247,
      "learning_rate": 2.371204545455927e-05,
      "loss": 0.6887,
      "step": 8136
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.1345365047454834,
      "learning_rate": 2.369261561234145e-05,
      "loss": 0.7896,
      "step": 8137
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9475464224815369,
      "learning_rate": 2.3673192664063792e-05,
      "loss": 0.8051,
      "step": 8138
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.993828296661377,
      "learning_rate": 2.3653776611481014e-05,
      "loss": 0.8178,
      "step": 8139
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9162449240684509,
      "learning_rate": 2.3634367456347218e-05,
      "loss": 0.565,
      "step": 8140
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9278566241264343,
      "learning_rate": 2.3614965200415962e-05,
      "loss": 0.5785,
      "step": 8141
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.0386141538619995,
      "learning_rate": 2.3595569845440112e-05,
      "loss": 0.6928,
      "step": 8142
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9798884391784668,
      "learning_rate": 2.3576181393171902e-05,
      "loss": 0.7504,
      "step": 8143
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9258533716201782,
      "learning_rate": 2.355679984536302e-05,
      "loss": 0.747,
      "step": 8144
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9842424392700195,
      "learning_rate": 2.3537425203764453e-05,
      "loss": 0.7275,
      "step": 8145
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.005286455154419,
      "learning_rate": 2.3518057470126586e-05,
      "loss": 0.7637,
      "step": 8146
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.024659276008606,
      "learning_rate": 2.349869664619917e-05,
      "loss": 0.7919,
      "step": 8147
    },
    {
      "epoch": 3.05,
      "grad_norm": 1.0532301664352417,
      "learning_rate": 2.347934273373138e-05,
      "loss": 0.6967,
      "step": 8148
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.939990758895874,
      "learning_rate": 2.3459995734471717e-05,
      "loss": 0.6713,
      "step": 8149
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.9598864912986755,
      "learning_rate": 2.344065565016804e-05,
      "loss": 0.7162,
      "step": 8150
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.9580551385879517,
      "learning_rate": 2.3421322482567676e-05,
      "loss": 0.7047,
      "step": 8151
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.1836827993392944,
      "learning_rate": 2.3401996233417245e-05,
      "loss": 0.717,
      "step": 8152
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.9279395937919617,
      "learning_rate": 2.338267690446272e-05,
      "loss": 0.7933,
      "step": 8153
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.9856902956962585,
      "learning_rate": 2.3363364497449557e-05,
      "loss": 0.8583,
      "step": 8154
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.974711000919342,
      "learning_rate": 2.334405901412251e-05,
      "loss": 0.6172,
      "step": 8155
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0841654539108276,
      "learning_rate": 2.3324760456225692e-05,
      "loss": 0.6824,
      "step": 8156
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.1081563234329224,
      "learning_rate": 2.3305468825502618e-05,
      "loss": 0.731,
      "step": 8157
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.1369671821594238,
      "learning_rate": 2.328618412369621e-05,
      "loss": 0.7533,
      "step": 8158
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0664194822311401,
      "learning_rate": 2.326690635254872e-05,
      "loss": 0.7159,
      "step": 8159
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0796053409576416,
      "learning_rate": 2.3247635513801746e-05,
      "loss": 0.8131,
      "step": 8160
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0047087669372559,
      "learning_rate": 2.32283716091964e-05,
      "loss": 0.5756,
      "step": 8161
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0937080383300781,
      "learning_rate": 2.3209114640472927e-05,
      "loss": 0.6566,
      "step": 8162
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0683318376541138,
      "learning_rate": 2.318986460937116e-05,
      "loss": 0.7376,
      "step": 8163
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.132534146308899,
      "learning_rate": 2.3170621517630296e-05,
      "loss": 0.815,
      "step": 8164
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.047959566116333,
      "learning_rate": 2.315138536698871e-05,
      "loss": 0.7764,
      "step": 8165
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.3007171154022217,
      "learning_rate": 2.313215615918436e-05,
      "loss": 0.7575,
      "step": 8166
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0939583778381348,
      "learning_rate": 2.311293389595448e-05,
      "loss": 0.7491,
      "step": 8167
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0535889863967896,
      "learning_rate": 2.309371857903567e-05,
      "loss": 0.7649,
      "step": 8168
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0150707960128784,
      "learning_rate": 2.307451021016397e-05,
      "loss": 0.6423,
      "step": 8169
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.1029785871505737,
      "learning_rate": 2.3055308791074713e-05,
      "loss": 0.8505,
      "step": 8170
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.003851294517517,
      "learning_rate": 2.3036114323502655e-05,
      "loss": 0.7927,
      "step": 8171
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.0146918296813965,
      "learning_rate": 2.301692680918186e-05,
      "loss": 0.6712,
      "step": 8172
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.8853873014450073,
      "learning_rate": 2.2997746249845887e-05,
      "loss": 0.8684,
      "step": 8173
    },
    {
      "epoch": 3.06,
      "grad_norm": 1.076791763305664,
      "learning_rate": 2.297857264722756e-05,
      "loss": 0.7162,
      "step": 8174
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.9796596765518188,
      "learning_rate": 2.295940600305907e-05,
      "loss": 0.7357,
      "step": 8175
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9705610871315002,
      "learning_rate": 2.2940246319072067e-05,
      "loss": 0.6966,
      "step": 8176
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9848082065582275,
      "learning_rate": 2.29210935969975e-05,
      "loss": 0.5788,
      "step": 8177
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.0097825527191162,
      "learning_rate": 2.2901947838565674e-05,
      "loss": 0.7224,
      "step": 8178
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.1544628143310547,
      "learning_rate": 2.288280904550637e-05,
      "loss": 0.7584,
      "step": 8179
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.1377264261245728,
      "learning_rate": 2.286367721954863e-05,
      "loss": 0.8853,
      "step": 8180
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.1240458488464355,
      "learning_rate": 2.2844552362420902e-05,
      "loss": 0.7935,
      "step": 8181
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9429728388786316,
      "learning_rate": 2.2825434475850994e-05,
      "loss": 0.714,
      "step": 8182
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.0275639295578003,
      "learning_rate": 2.2806323561566146e-05,
      "loss": 0.6896,
      "step": 8183
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9426898956298828,
      "learning_rate": 2.27872196212929e-05,
      "loss": 0.7512,
      "step": 8184
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.1296406984329224,
      "learning_rate": 2.276812265675715e-05,
      "loss": 0.7115,
      "step": 8185
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9507439732551575,
      "learning_rate": 2.2749032669684254e-05,
      "loss": 0.6555,
      "step": 8186
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.0723408460617065,
      "learning_rate": 2.2729949661798878e-05,
      "loss": 0.6829,
      "step": 8187
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.965126097202301,
      "learning_rate": 2.2710873634825003e-05,
      "loss": 0.6265,
      "step": 8188
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.0534261465072632,
      "learning_rate": 2.269180459048612e-05,
      "loss": 0.6901,
      "step": 8189
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9463297128677368,
      "learning_rate": 2.2672742530504977e-05,
      "loss": 0.6038,
      "step": 8190
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9742079973220825,
      "learning_rate": 2.2653687456603723e-05,
      "loss": 0.6789,
      "step": 8191
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9606778025627136,
      "learning_rate": 2.263463937050384e-05,
      "loss": 0.6895,
      "step": 8192
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.0363831520080566,
      "learning_rate": 2.2615598273926276e-05,
      "loss": 0.8327,
      "step": 8193
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9853733777999878,
      "learning_rate": 2.259656416859126e-05,
      "loss": 0.7493,
      "step": 8194
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.133048176765442,
      "learning_rate": 2.257753705621839e-05,
      "loss": 0.8727,
      "step": 8195
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9047926068305969,
      "learning_rate": 2.255851693852674e-05,
      "loss": 0.6436,
      "step": 8196
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.0124412775039673,
      "learning_rate": 2.2539503817234553e-05,
      "loss": 0.7302,
      "step": 8197
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.0014770030975342,
      "learning_rate": 2.2520497694059604e-05,
      "loss": 0.4809,
      "step": 8198
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.2341759204864502,
      "learning_rate": 2.250149857071907e-05,
      "loss": 0.7098,
      "step": 8199
    },
    {
      "epoch": 3.07,
      "grad_norm": 1.1458659172058105,
      "learning_rate": 2.2482506448929285e-05,
      "loss": 0.7554,
      "step": 8200
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.9872215390205383,
      "learning_rate": 2.2463521330406177e-05,
      "loss": 0.7369,
      "step": 8201
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.2173840999603271,
      "learning_rate": 2.24445432168649e-05,
      "loss": 0.7936,
      "step": 8202
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0458061695098877,
      "learning_rate": 2.242557211001999e-05,
      "loss": 0.5335,
      "step": 8203
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0405919551849365,
      "learning_rate": 2.240660801158545e-05,
      "loss": 0.5984,
      "step": 8204
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.2485703229904175,
      "learning_rate": 2.2387650923274538e-05,
      "loss": 0.8346,
      "step": 8205
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.9710096120834351,
      "learning_rate": 2.2368700846799927e-05,
      "loss": 0.7823,
      "step": 8206
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.104174017906189,
      "learning_rate": 2.2349757783873627e-05,
      "loss": 0.8513,
      "step": 8207
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.9613421559333801,
      "learning_rate": 2.2330821736207076e-05,
      "loss": 0.7485,
      "step": 8208
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.000362515449524,
      "learning_rate": 2.2311892705511018e-05,
      "loss": 0.638,
      "step": 8209
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0569934844970703,
      "learning_rate": 2.2292970693495564e-05,
      "loss": 0.7297,
      "step": 8210
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.1103142499923706,
      "learning_rate": 2.2274055701870254e-05,
      "loss": 0.8294,
      "step": 8211
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0584805011749268,
      "learning_rate": 2.225514773234393e-05,
      "loss": 0.7223,
      "step": 8212
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.3667136430740356,
      "learning_rate": 2.2236246786624792e-05,
      "loss": 0.7441,
      "step": 8213
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0754313468933105,
      "learning_rate": 2.2217352866420492e-05,
      "loss": 0.9487,
      "step": 8214
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.003045916557312,
      "learning_rate": 2.2198465973437943e-05,
      "loss": 0.786,
      "step": 8215
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0200042724609375,
      "learning_rate": 2.21795861093835e-05,
      "loss": 0.6676,
      "step": 8216
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0485609769821167,
      "learning_rate": 2.2160713275962797e-05,
      "loss": 0.7465,
      "step": 8217
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0643272399902344,
      "learning_rate": 2.214184747488095e-05,
      "loss": 0.8616,
      "step": 8218
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.9702640771865845,
      "learning_rate": 2.2122988707842353e-05,
      "loss": 0.8085,
      "step": 8219
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.027295470237732,
      "learning_rate": 2.210413697655076e-05,
      "loss": 0.7033,
      "step": 8220
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.9336124062538147,
      "learning_rate": 2.208529228270937e-05,
      "loss": 0.7115,
      "step": 8221
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0429178476333618,
      "learning_rate": 2.2066454628020683e-05,
      "loss": 0.5722,
      "step": 8222
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0630872249603271,
      "learning_rate": 2.204762401418652e-05,
      "loss": 0.8002,
      "step": 8223
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0240941047668457,
      "learning_rate": 2.2028800442908193e-05,
      "loss": 0.7495,
      "step": 8224
    },
    {
      "epoch": 3.08,
      "grad_norm": 1.0652430057525635,
      "learning_rate": 2.2009983915886278e-05,
      "loss": 0.7304,
      "step": 8225
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.9660937190055847,
      "learning_rate": 2.1991174434820737e-05,
      "loss": 0.6872,
      "step": 8226
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.9742222428321838,
      "learning_rate": 2.197237200141088e-05,
      "loss": 0.7752,
      "step": 8227
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.9536305665969849,
      "learning_rate": 2.195357661735544e-05,
      "loss": 0.7186,
      "step": 8228
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.9891760945320129,
      "learning_rate": 2.193478828435246e-05,
      "loss": 0.7275,
      "step": 8229
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.9621010422706604,
      "learning_rate": 2.1916007004099325e-05,
      "loss": 0.7683,
      "step": 8230
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.9448545575141907,
      "learning_rate": 2.18972327782929e-05,
      "loss": 0.6895,
      "step": 8231
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0644571781158447,
      "learning_rate": 2.1878465608629217e-05,
      "loss": 0.7991,
      "step": 8232
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.1056584119796753,
      "learning_rate": 2.1859705496803852e-05,
      "loss": 0.7556,
      "step": 8233
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0642894506454468,
      "learning_rate": 2.184095244451172e-05,
      "loss": 0.8562,
      "step": 8234
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0452674627304077,
      "learning_rate": 2.1822206453446937e-05,
      "loss": 0.7826,
      "step": 8235
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0155723094940186,
      "learning_rate": 2.1803467525303202e-05,
      "loss": 0.7858,
      "step": 8236
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.063292384147644,
      "learning_rate": 2.1784735661773415e-05,
      "loss": 0.8412,
      "step": 8237
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0433480739593506,
      "learning_rate": 2.176601086454988e-05,
      "loss": 0.8532,
      "step": 8238
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0794016122817993,
      "learning_rate": 2.174729313532433e-05,
      "loss": 0.7179,
      "step": 8239
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0173252820968628,
      "learning_rate": 2.1728582475787796e-05,
      "loss": 0.6623,
      "step": 8240
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0129539966583252,
      "learning_rate": 2.1709878887630654e-05,
      "loss": 0.7927,
      "step": 8241
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.8960475921630859,
      "learning_rate": 2.169118237254264e-05,
      "loss": 0.6922,
      "step": 8242
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.071644902229309,
      "learning_rate": 2.167249293221293e-05,
      "loss": 0.6555,
      "step": 8243
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0081762075424194,
      "learning_rate": 2.1653810568330034e-05,
      "loss": 0.8254,
      "step": 8244
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.9863021373748779,
      "learning_rate": 2.163513528258172e-05,
      "loss": 0.7309,
      "step": 8245
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.1220043897628784,
      "learning_rate": 2.1616467076655244e-05,
      "loss": 0.7653,
      "step": 8246
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.1087371110916138,
      "learning_rate": 2.1597805952237172e-05,
      "loss": 0.6634,
      "step": 8247
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.054997205734253,
      "learning_rate": 2.1579151911013385e-05,
      "loss": 0.6708,
      "step": 8248
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.9785958528518677,
      "learning_rate": 2.1560504954669238e-05,
      "loss": 0.7047,
      "step": 8249
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.1232924461364746,
      "learning_rate": 2.1541865084889346e-05,
      "loss": 0.7633,
      "step": 8250
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0624607801437378,
      "learning_rate": 2.152323230335771e-05,
      "loss": 0.7953,
      "step": 8251
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.9319671392440796,
      "learning_rate": 2.150460661175768e-05,
      "loss": 0.6803,
      "step": 8252
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.9834529161453247,
      "learning_rate": 2.1485988011772028e-05,
      "loss": 0.7041,
      "step": 8253
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0230741500854492,
      "learning_rate": 2.1467376505082814e-05,
      "loss": 0.8982,
      "step": 8254
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0135674476623535,
      "learning_rate": 2.144877209337145e-05,
      "loss": 0.6944,
      "step": 8255
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.057533860206604,
      "learning_rate": 2.14301747783188e-05,
      "loss": 0.7322,
      "step": 8256
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.014972448348999,
      "learning_rate": 2.1411584561605002e-05,
      "loss": 0.8024,
      "step": 8257
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0498735904693604,
      "learning_rate": 2.1393001444909544e-05,
      "loss": 0.6009,
      "step": 8258
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0380078554153442,
      "learning_rate": 2.137442542991136e-05,
      "loss": 0.7128,
      "step": 8259
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.9782800078392029,
      "learning_rate": 2.1355856518288665e-05,
      "loss": 0.7792,
      "step": 8260
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.202033281326294,
      "learning_rate": 2.1337294711719048e-05,
      "loss": 0.7171,
      "step": 8261
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.139710545539856,
      "learning_rate": 2.1318740011879446e-05,
      "loss": 0.8548,
      "step": 8262
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0056865215301514,
      "learning_rate": 2.1300192420446207e-05,
      "loss": 0.6546,
      "step": 8263
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0196723937988281,
      "learning_rate": 2.1281651939094992e-05,
      "loss": 0.6197,
      "step": 8264
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.1030669212341309,
      "learning_rate": 2.1263118569500795e-05,
      "loss": 0.5977,
      "step": 8265
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.026012659072876,
      "learning_rate": 2.1244592313338085e-05,
      "loss": 0.7487,
      "step": 8266
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.9869826436042786,
      "learning_rate": 2.122607317228049e-05,
      "loss": 0.7398,
      "step": 8267
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0317878723144531,
      "learning_rate": 2.1207561148001166e-05,
      "loss": 0.7687,
      "step": 8268
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.1159306764602661,
      "learning_rate": 2.1189056242172635e-05,
      "loss": 0.7357,
      "step": 8269
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0868357419967651,
      "learning_rate": 2.1170558456466582e-05,
      "loss": 0.7171,
      "step": 8270
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.9717540740966797,
      "learning_rate": 2.1152067792554265e-05,
      "loss": 0.6985,
      "step": 8271
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.030918002128601,
      "learning_rate": 2.1133584252106197e-05,
      "loss": 0.7555,
      "step": 8272
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.1515148878097534,
      "learning_rate": 2.1115107836792214e-05,
      "loss": 0.7737,
      "step": 8273
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.1085578203201294,
      "learning_rate": 2.1096638548281633e-05,
      "loss": 0.7265,
      "step": 8274
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0248810052871704,
      "learning_rate": 2.1078176388242964e-05,
      "loss": 0.7183,
      "step": 8275
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0286072492599487,
      "learning_rate": 2.1059721358344276e-05,
      "loss": 0.8281,
      "step": 8276
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.9900050163269043,
      "learning_rate": 2.104127346025274e-05,
      "loss": 0.6648,
      "step": 8277
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.054434061050415,
      "learning_rate": 2.1022832695635087e-05,
      "loss": 0.6206,
      "step": 8278
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.0165799856185913,
      "learning_rate": 2.100439906615739e-05,
      "loss": 0.5553,
      "step": 8279
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.0425355434417725,
      "learning_rate": 2.0985972573484913e-05,
      "loss": 0.8124,
      "step": 8280
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.0266938209533691,
      "learning_rate": 2.0967553219282455e-05,
      "loss": 0.7354,
      "step": 8281
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.1215537786483765,
      "learning_rate": 2.094914100521409e-05,
      "loss": 0.8392,
      "step": 8282
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.157520055770874,
      "learning_rate": 2.0930735932943225e-05,
      "loss": 0.7311,
      "step": 8283
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.1327176094055176,
      "learning_rate": 2.0912338004132705e-05,
      "loss": 0.7501,
      "step": 8284
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.0814790725708008,
      "learning_rate": 2.0893947220444643e-05,
      "loss": 0.6353,
      "step": 8285
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9800668954849243,
      "learning_rate": 2.087556358354056e-05,
      "loss": 0.7522,
      "step": 8286
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9925199151039124,
      "learning_rate": 2.0857187095081276e-05,
      "loss": 0.6596,
      "step": 8287
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9486314058303833,
      "learning_rate": 2.083881775672707e-05,
      "loss": 0.7288,
      "step": 8288
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.0026705265045166,
      "learning_rate": 2.0820455570137465e-05,
      "loss": 0.6446,
      "step": 8289
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9508515000343323,
      "learning_rate": 2.080210053697136e-05,
      "loss": 0.6274,
      "step": 8290
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.0909723043441772,
      "learning_rate": 2.0783752658887066e-05,
      "loss": 0.8206,
      "step": 8291
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.0272409915924072,
      "learning_rate": 2.0765411937542212e-05,
      "loss": 0.6601,
      "step": 8292
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9843588471412659,
      "learning_rate": 2.074707837459374e-05,
      "loss": 0.6564,
      "step": 8293
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9665646553039551,
      "learning_rate": 2.0728751971698036e-05,
      "loss": 0.6414,
      "step": 8294
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9018855690956116,
      "learning_rate": 2.0710432730510743e-05,
      "loss": 0.6759,
      "step": 8295
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.0668045282363892,
      "learning_rate": 2.069212065268693e-05,
      "loss": 0.6784,
      "step": 8296
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9695737361907959,
      "learning_rate": 2.067381573988094e-05,
      "loss": 0.591,
      "step": 8297
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.165960669517517,
      "learning_rate": 2.0655517993746598e-05,
      "loss": 0.572,
      "step": 8298
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.021818995475769,
      "learning_rate": 2.0637227415936945e-05,
      "loss": 0.7529,
      "step": 8299
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.1195276975631714,
      "learning_rate": 2.061894400810441e-05,
      "loss": 0.5968,
      "step": 8300
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9856497645378113,
      "learning_rate": 2.0600667771900895e-05,
      "loss": 0.7196,
      "step": 8301
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.1288400888442993,
      "learning_rate": 2.0582398708977435e-05,
      "loss": 0.6801,
      "step": 8302
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9925099611282349,
      "learning_rate": 2.056413682098459e-05,
      "loss": 0.7295,
      "step": 8303
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.0284779071807861,
      "learning_rate": 2.0545882109572246e-05,
      "loss": 0.6894,
      "step": 8304
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9781960844993591,
      "learning_rate": 2.0527634576389576e-05,
      "loss": 0.683,
      "step": 8305
    },
    {
      "epoch": 3.11,
      "grad_norm": 1.0912878513336182,
      "learning_rate": 2.0509394223085165e-05,
      "loss": 0.7205,
      "step": 8306
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.0725929737091064,
      "learning_rate": 2.0491161051306874e-05,
      "loss": 0.5883,
      "step": 8307
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.9850541353225708,
      "learning_rate": 2.0472935062702038e-05,
      "loss": 0.6227,
      "step": 8308
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.0655592679977417,
      "learning_rate": 2.045471625891723e-05,
      "loss": 0.653,
      "step": 8309
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.053954839706421,
      "learning_rate": 2.04365046415984e-05,
      "loss": 0.6677,
      "step": 8310
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.0735995769500732,
      "learning_rate": 2.041830021239094e-05,
      "loss": 0.7451,
      "step": 8311
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.9468885660171509,
      "learning_rate": 2.0400102972939418e-05,
      "loss": 0.6263,
      "step": 8312
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.064113736152649,
      "learning_rate": 2.03819129248879e-05,
      "loss": 0.7406,
      "step": 8313
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.0384057760238647,
      "learning_rate": 2.0363730069879806e-05,
      "loss": 0.7383,
      "step": 8314
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.9800799489021301,
      "learning_rate": 2.034555440955773e-05,
      "loss": 0.7165,
      "step": 8315
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.0471279621124268,
      "learning_rate": 2.032738594556386e-05,
      "loss": 0.6934,
      "step": 8316
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.0083075761795044,
      "learning_rate": 2.0309224679539552e-05,
      "loss": 0.6893,
      "step": 8317
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.2627086639404297,
      "learning_rate": 2.0291070613125562e-05,
      "loss": 0.7502,
      "step": 8318
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.155874252319336,
      "learning_rate": 2.027292374796206e-05,
      "loss": 0.8237,
      "step": 8319
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.1413154602050781,
      "learning_rate": 2.0254784085688483e-05,
      "loss": 0.6782,
      "step": 8320
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.0282950401306152,
      "learning_rate": 2.023665162794365e-05,
      "loss": 0.8202,
      "step": 8321
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.0178747177124023,
      "learning_rate": 2.0218526376365698e-05,
      "loss": 0.761,
      "step": 8322
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.9160252213478088,
      "learning_rate": 2.0200408332592192e-05,
      "loss": 0.6193,
      "step": 8323
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.9832085967063904,
      "learning_rate": 2.0182297498259962e-05,
      "loss": 0.7775,
      "step": 8324
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.1044410467147827,
      "learning_rate": 2.01641938750052e-05,
      "loss": 0.7243,
      "step": 8325
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.993820309638977,
      "learning_rate": 2.0146097464463532e-05,
      "loss": 0.8018,
      "step": 8326
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.218582272529602,
      "learning_rate": 2.0128008268269815e-05,
      "loss": 0.7955,
      "step": 8327
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.0267633199691772,
      "learning_rate": 2.010992628805829e-05,
      "loss": 0.7333,
      "step": 8328
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.0718587636947632,
      "learning_rate": 2.0091851525462624e-05,
      "loss": 0.7573,
      "step": 8329
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.2271544933319092,
      "learning_rate": 2.0073783982115723e-05,
      "loss": 0.6795,
      "step": 8330
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.3247065544128418,
      "learning_rate": 2.0055723659649904e-05,
      "loss": 0.6999,
      "step": 8331
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.084662675857544,
      "learning_rate": 2.0037670559696787e-05,
      "loss": 0.7694,
      "step": 8332
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.1112602949142456,
      "learning_rate": 2.0019624683887405e-05,
      "loss": 0.8213,
      "step": 8333
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9747363328933716,
      "learning_rate": 2.0001586033852092e-05,
      "loss": 0.8292,
      "step": 8334
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9815605878829956,
      "learning_rate": 1.99835546112205e-05,
      "loss": 0.6228,
      "step": 8335
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0785510540008545,
      "learning_rate": 1.9965530417621727e-05,
      "loss": 0.7332,
      "step": 8336
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.8619665503501892,
      "learning_rate": 1.994751345468413e-05,
      "loss": 0.6881,
      "step": 8337
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0108859539031982,
      "learning_rate": 1.9929503724035402e-05,
      "loss": 0.7435,
      "step": 8338
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.1934397220611572,
      "learning_rate": 1.9911501227302687e-05,
      "loss": 0.6765,
      "step": 8339
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0977833271026611,
      "learning_rate": 1.9893505966112382e-05,
      "loss": 0.7105,
      "step": 8340
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.3163998126983643,
      "learning_rate": 1.9875517942090237e-05,
      "loss": 0.7407,
      "step": 8341
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0598845481872559,
      "learning_rate": 1.985753715686136e-05,
      "loss": 0.7661,
      "step": 8342
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0815600156784058,
      "learning_rate": 1.983956361205027e-05,
      "loss": 0.7122,
      "step": 8343
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0810781717300415,
      "learning_rate": 1.9821597309280737e-05,
      "loss": 0.6486,
      "step": 8344
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0087732076644897,
      "learning_rate": 1.9803638250175892e-05,
      "loss": 0.7152,
      "step": 8345
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9768717885017395,
      "learning_rate": 1.9785686436358307e-05,
      "loss": 0.6203,
      "step": 8346
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0470556020736694,
      "learning_rate": 1.976774186944973e-05,
      "loss": 0.8537,
      "step": 8347
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0474168062210083,
      "learning_rate": 1.9749804551071405e-05,
      "loss": 0.7626,
      "step": 8348
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0206280946731567,
      "learning_rate": 1.9731874482843914e-05,
      "loss": 0.7308,
      "step": 8349
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.1213023662567139,
      "learning_rate": 1.971395166638703e-05,
      "loss": 0.6462,
      "step": 8350
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0654561519622803,
      "learning_rate": 1.969603610332007e-05,
      "loss": 0.6034,
      "step": 8351
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0602190494537354,
      "learning_rate": 1.967812779526156e-05,
      "loss": 0.906,
      "step": 8352
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0256491899490356,
      "learning_rate": 1.9660226743829402e-05,
      "loss": 0.7568,
      "step": 8353
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9291660189628601,
      "learning_rate": 1.9642332950640897e-05,
      "loss": 0.7301,
      "step": 8354
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9917504787445068,
      "learning_rate": 1.9624446417312625e-05,
      "loss": 0.7092,
      "step": 8355
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9856034517288208,
      "learning_rate": 1.960656714546054e-05,
      "loss": 0.7454,
      "step": 8356
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9775485992431641,
      "learning_rate": 1.9588695136699897e-05,
      "loss": 0.7247,
      "step": 8357
    },
    {
      "epoch": 3.13,
      "grad_norm": 1.0556694269180298,
      "learning_rate": 1.957083039264539e-05,
      "loss": 0.6457,
      "step": 8358
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.1901068687438965,
      "learning_rate": 1.955297291491096e-05,
      "loss": 0.6972,
      "step": 8359
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.0057767629623413,
      "learning_rate": 1.953512270510992e-05,
      "loss": 0.7852,
      "step": 8360
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.8664202690124512,
      "learning_rate": 1.951727976485498e-05,
      "loss": 0.6413,
      "step": 8361
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.9768220782279968,
      "learning_rate": 1.9499444095758114e-05,
      "loss": 0.5435,
      "step": 8362
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.004700779914856,
      "learning_rate": 1.9481615699430654e-05,
      "loss": 0.8204,
      "step": 8363
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.197533130645752,
      "learning_rate": 1.9463794577483352e-05,
      "loss": 0.8795,
      "step": 8364
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.0851786136627197,
      "learning_rate": 1.9445980731526214e-05,
      "loss": 0.7796,
      "step": 8365
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.118904709815979,
      "learning_rate": 1.9428174163168622e-05,
      "loss": 0.7285,
      "step": 8366
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.0331822633743286,
      "learning_rate": 1.9410374874019277e-05,
      "loss": 0.8337,
      "step": 8367
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.1711002588272095,
      "learning_rate": 1.9392582865686282e-05,
      "loss": 0.788,
      "step": 8368
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.0771785974502563,
      "learning_rate": 1.937479813977703e-05,
      "loss": 0.6528,
      "step": 8369
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.1298748254776,
      "learning_rate": 1.9357020697898244e-05,
      "loss": 0.8977,
      "step": 8370
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.9739071130752563,
      "learning_rate": 1.933925054165606e-05,
      "loss": 0.6568,
      "step": 8371
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.9915308356285095,
      "learning_rate": 1.9321487672655893e-05,
      "loss": 0.729,
      "step": 8372
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.9432351589202881,
      "learning_rate": 1.930373209250249e-05,
      "loss": 0.6333,
      "step": 8373
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.5936599969863892,
      "learning_rate": 1.928598380280001e-05,
      "loss": 0.8911,
      "step": 8374
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.3058140277862549,
      "learning_rate": 1.9268242805151902e-05,
      "loss": 0.7907,
      "step": 8375
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.2683802843093872,
      "learning_rate": 1.925050910116095e-05,
      "loss": 0.6817,
      "step": 8376
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.9748633503913879,
      "learning_rate": 1.9232782692429276e-05,
      "loss": 0.7043,
      "step": 8377
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.175662875175476,
      "learning_rate": 1.9215063580558402e-05,
      "loss": 0.5989,
      "step": 8378
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.0589122772216797,
      "learning_rate": 1.919735176714913e-05,
      "loss": 0.8654,
      "step": 8379
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.1583337783813477,
      "learning_rate": 1.9179647253801602e-05,
      "loss": 0.7416,
      "step": 8380
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.354985237121582,
      "learning_rate": 1.9161950042115383e-05,
      "loss": 0.6838,
      "step": 8381
    },
    {
      "epoch": 3.14,
      "grad_norm": 1.0164586305618286,
      "learning_rate": 1.914426013368924e-05,
      "loss": 0.5713,
      "step": 8382
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.9318915009498596,
      "learning_rate": 1.9126577530121372e-05,
      "loss": 0.5725,
      "step": 8383
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.9165032505989075,
      "learning_rate": 1.9108902233009384e-05,
      "loss": 0.5954,
      "step": 8384
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9548971056938171,
      "learning_rate": 1.909123424395002e-05,
      "loss": 0.6958,
      "step": 8385
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0613524913787842,
      "learning_rate": 1.907357356453956e-05,
      "loss": 0.6677,
      "step": 8386
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0344734191894531,
      "learning_rate": 1.9055920196373523e-05,
      "loss": 0.658,
      "step": 8387
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0572896003723145,
      "learning_rate": 1.9038274141046774e-05,
      "loss": 0.6133,
      "step": 8388
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9751765131950378,
      "learning_rate": 1.9020635400153575e-05,
      "loss": 0.6706,
      "step": 8389
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0860122442245483,
      "learning_rate": 1.900300397528746e-05,
      "loss": 0.6943,
      "step": 8390
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.1493300199508667,
      "learning_rate": 1.898537986804133e-05,
      "loss": 0.8076,
      "step": 8391
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9801821112632751,
      "learning_rate": 1.8967763080007405e-05,
      "loss": 0.7855,
      "step": 8392
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9901358485221863,
      "learning_rate": 1.895015361277731e-05,
      "loss": 0.6668,
      "step": 8393
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.1115587949752808,
      "learning_rate": 1.8932551467941927e-05,
      "loss": 0.8024,
      "step": 8394
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0727943181991577,
      "learning_rate": 1.89149566470915e-05,
      "loss": 0.6848,
      "step": 8395
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0310070514678955,
      "learning_rate": 1.8897369151815658e-05,
      "loss": 0.6983,
      "step": 8396
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0392391681671143,
      "learning_rate": 1.8879788983703305e-05,
      "loss": 0.657,
      "step": 8397
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9275290369987488,
      "learning_rate": 1.8862216144342692e-05,
      "loss": 0.7654,
      "step": 8398
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0170843601226807,
      "learning_rate": 1.8844650635321483e-05,
      "loss": 0.7559,
      "step": 8399
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.961005687713623,
      "learning_rate": 1.8827092458226582e-05,
      "loss": 0.541,
      "step": 8400
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9901219606399536,
      "learning_rate": 1.8809541614644287e-05,
      "loss": 0.6676,
      "step": 8401
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.1597495079040527,
      "learning_rate": 1.8791998106160168e-05,
      "loss": 0.7707,
      "step": 8402
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0873172283172607,
      "learning_rate": 1.8774461934359257e-05,
      "loss": 0.7916,
      "step": 8403
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0745017528533936,
      "learning_rate": 1.875693310082581e-05,
      "loss": 0.8036,
      "step": 8404
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.1398003101348877,
      "learning_rate": 1.873941160714343e-05,
      "loss": 0.694,
      "step": 8405
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0003429651260376,
      "learning_rate": 1.8721897454895154e-05,
      "loss": 0.7549,
      "step": 8406
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9602251052856445,
      "learning_rate": 1.870439064566324e-05,
      "loss": 0.6852,
      "step": 8407
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.992182731628418,
      "learning_rate": 1.868689118102931e-05,
      "loss": 0.7082,
      "step": 8408
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9336217045783997,
      "learning_rate": 1.8669399062574388e-05,
      "loss": 0.6543,
      "step": 8409
    },
    {
      "epoch": 3.15,
      "grad_norm": 1.0972353219985962,
      "learning_rate": 1.865191429187878e-05,
      "loss": 0.7897,
      "step": 8410
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.9045754075050354,
      "learning_rate": 1.863443687052211e-05,
      "loss": 0.8147,
      "step": 8411
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.9818060398101807,
      "learning_rate": 1.861696680008337e-05,
      "loss": 0.6525,
      "step": 8412
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.0161360502243042,
      "learning_rate": 1.8599504082140896e-05,
      "loss": 0.7463,
      "step": 8413
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.059190273284912,
      "learning_rate": 1.858204871827235e-05,
      "loss": 0.6992,
      "step": 8414
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.0103576183319092,
      "learning_rate": 1.856460071005469e-05,
      "loss": 0.5976,
      "step": 8415
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.082658290863037,
      "learning_rate": 1.8547160059064327e-05,
      "loss": 0.9191,
      "step": 8416
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.0552284717559814,
      "learning_rate": 1.852972676687681e-05,
      "loss": 0.775,
      "step": 8417
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.014639139175415,
      "learning_rate": 1.8512300835067196e-05,
      "loss": 0.6727,
      "step": 8418
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.9792020916938782,
      "learning_rate": 1.8494882265209867e-05,
      "loss": 0.6894,
      "step": 8419
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.9599397778511047,
      "learning_rate": 1.8477471058878393e-05,
      "loss": 0.6819,
      "step": 8420
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.0084049701690674,
      "learning_rate": 1.846006721764586e-05,
      "loss": 0.7823,
      "step": 8421
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.0630186796188354,
      "learning_rate": 1.844267074308458e-05,
      "loss": 0.7474,
      "step": 8422
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.2924787998199463,
      "learning_rate": 1.842528163676619e-05,
      "loss": 0.8045,
      "step": 8423
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.060523509979248,
      "learning_rate": 1.840789990026176e-05,
      "loss": 0.5792,
      "step": 8424
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.9210415482521057,
      "learning_rate": 1.8390525535141567e-05,
      "loss": 0.5422,
      "step": 8425
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.0876710414886475,
      "learning_rate": 1.8373158542975377e-05,
      "loss": 0.7913,
      "step": 8426
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.1454886198043823,
      "learning_rate": 1.835579892533209e-05,
      "loss": 0.7024,
      "step": 8427
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.7660450339317322,
      "learning_rate": 1.833844668378011e-05,
      "loss": 0.3893,
      "step": 8428
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.1798911094665527,
      "learning_rate": 1.8321101819887143e-05,
      "loss": 0.6849,
      "step": 8429
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.0423330068588257,
      "learning_rate": 1.830376433522012e-05,
      "loss": 0.7309,
      "step": 8430
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.9713183045387268,
      "learning_rate": 1.828643423134545e-05,
      "loss": 0.5753,
      "step": 8431
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.1074738502502441,
      "learning_rate": 1.8269111509828783e-05,
      "loss": 0.7362,
      "step": 8432
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.0143823623657227,
      "learning_rate": 1.8251796172235113e-05,
      "loss": 0.7425,
      "step": 8433
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.9743937253952026,
      "learning_rate": 1.823448822012882e-05,
      "loss": 0.769,
      "step": 8434
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.0144141912460327,
      "learning_rate": 1.8217187655073564e-05,
      "loss": 0.773,
      "step": 8435
    },
    {
      "epoch": 3.16,
      "grad_norm": 1.1118022203445435,
      "learning_rate": 1.819989447863235e-05,
      "loss": 0.7126,
      "step": 8436
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.0982333421707153,
      "learning_rate": 1.81826086923675e-05,
      "loss": 0.7507,
      "step": 8437
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.9950075745582581,
      "learning_rate": 1.816533029784072e-05,
      "loss": 0.7521,
      "step": 8438
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.059446096420288,
      "learning_rate": 1.8148059296613006e-05,
      "loss": 0.8481,
      "step": 8439
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.8937699794769287,
      "learning_rate": 1.813079569024466e-05,
      "loss": 0.739,
      "step": 8440
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.996634304523468,
      "learning_rate": 1.8113539480295415e-05,
      "loss": 0.7205,
      "step": 8441
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.9936226010322571,
      "learning_rate": 1.8096290668324233e-05,
      "loss": 0.6932,
      "step": 8442
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.1790931224822998,
      "learning_rate": 1.8079049255889434e-05,
      "loss": 0.6121,
      "step": 8443
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.0803405046463013,
      "learning_rate": 1.806181524454872e-05,
      "loss": 0.7606,
      "step": 8444
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.0857096910476685,
      "learning_rate": 1.8044588635859062e-05,
      "loss": 0.677,
      "step": 8445
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.0761653184890747,
      "learning_rate": 1.8027369431376795e-05,
      "loss": 0.7297,
      "step": 8446
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.0815399885177612,
      "learning_rate": 1.8010157632657543e-05,
      "loss": 0.7515,
      "step": 8447
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.0184310674667358,
      "learning_rate": 1.799295324125635e-05,
      "loss": 0.6938,
      "step": 8448
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.9802287817001343,
      "learning_rate": 1.7975756258727506e-05,
      "loss": 0.7026,
      "step": 8449
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.067026972770691,
      "learning_rate": 1.7958566686624644e-05,
      "loss": 0.7018,
      "step": 8450
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.9424697756767273,
      "learning_rate": 1.7941384526500817e-05,
      "loss": 0.6128,
      "step": 8451
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.9209526777267456,
      "learning_rate": 1.792420977990823e-05,
      "loss": 0.674,
      "step": 8452
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.089989423751831,
      "learning_rate": 1.790704244839857e-05,
      "loss": 0.7304,
      "step": 8453
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.0849412679672241,
      "learning_rate": 1.7889882533522852e-05,
      "loss": 0.7714,
      "step": 8454
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.956847071647644,
      "learning_rate": 1.787273003683133e-05,
      "loss": 0.6448,
      "step": 8455
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.9841948747634888,
      "learning_rate": 1.7855584959873652e-05,
      "loss": 0.7607,
      "step": 8456
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.0886846780776978,
      "learning_rate": 1.7838447304198748e-05,
      "loss": 0.6215,
      "step": 8457
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.1162586212158203,
      "learning_rate": 1.7821317071354958e-05,
      "loss": 0.8674,
      "step": 8458
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.9415084719657898,
      "learning_rate": 1.7804194262889874e-05,
      "loss": 0.5889,
      "step": 8459
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.0757049322128296,
      "learning_rate": 1.7787078880350427e-05,
      "loss": 0.8791,
      "step": 8460
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.0685847997665405,
      "learning_rate": 1.7769970925282964e-05,
      "loss": 0.6804,
      "step": 8461
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.9905105233192444,
      "learning_rate": 1.775287039923299e-05,
      "loss": 0.7063,
      "step": 8462
    },
    {
      "epoch": 3.17,
      "grad_norm": 1.1176837682724,
      "learning_rate": 1.7735777303745494e-05,
      "loss": 0.6039,
      "step": 8463
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.1858307123184204,
      "learning_rate": 1.771869164036478e-05,
      "loss": 0.7345,
      "step": 8464
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0211575031280518,
      "learning_rate": 1.7701613410634365e-05,
      "loss": 0.6766,
      "step": 8465
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.013595461845398,
      "learning_rate": 1.7684542616097222e-05,
      "loss": 0.7852,
      "step": 8466
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.9854930639266968,
      "learning_rate": 1.7667479258295595e-05,
      "loss": 0.655,
      "step": 8467
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.1102757453918457,
      "learning_rate": 1.7650423338771016e-05,
      "loss": 0.5896,
      "step": 8468
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.9293200373649597,
      "learning_rate": 1.7633374859064456e-05,
      "loss": 0.6267,
      "step": 8469
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0363894701004028,
      "learning_rate": 1.7616333820716114e-05,
      "loss": 0.7013,
      "step": 8470
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0113532543182373,
      "learning_rate": 1.759930022526556e-05,
      "loss": 0.8036,
      "step": 8471
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0682631731033325,
      "learning_rate": 1.758227407425166e-05,
      "loss": 0.8466,
      "step": 8472
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0091655254364014,
      "learning_rate": 1.7565255369212662e-05,
      "loss": 0.7186,
      "step": 8473
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.241448998451233,
      "learning_rate": 1.7548244111686108e-05,
      "loss": 0.7603,
      "step": 8474
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0603574514389038,
      "learning_rate": 1.753124030320883e-05,
      "loss": 0.6922,
      "step": 8475
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0982904434204102,
      "learning_rate": 1.7514243945317077e-05,
      "loss": 0.8319,
      "step": 8476
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.1419003009796143,
      "learning_rate": 1.7497255039546355e-05,
      "loss": 0.5457,
      "step": 8477
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0816574096679688,
      "learning_rate": 1.7480273587431483e-05,
      "loss": 0.7376,
      "step": 8478
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.9737247228622437,
      "learning_rate": 1.7463299590506688e-05,
      "loss": 0.7309,
      "step": 8479
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.9271056652069092,
      "learning_rate": 1.7446333050305452e-05,
      "loss": 0.7235,
      "step": 8480
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0294498205184937,
      "learning_rate": 1.7429373968360617e-05,
      "loss": 0.5945,
      "step": 8481
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.008527159690857,
      "learning_rate": 1.7412422346204293e-05,
      "loss": 0.6917,
      "step": 8482
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.070875644683838,
      "learning_rate": 1.739547818536804e-05,
      "loss": 0.7865,
      "step": 8483
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.936811625957489,
      "learning_rate": 1.7378541487382617e-05,
      "loss": 0.6024,
      "step": 8484
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0893816947937012,
      "learning_rate": 1.7361612253778148e-05,
      "loss": 0.6987,
      "step": 8485
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.8953719735145569,
      "learning_rate": 1.7344690486084137e-05,
      "loss": 0.7713,
      "step": 8486
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0382462739944458,
      "learning_rate": 1.7327776185829347e-05,
      "loss": 0.6407,
      "step": 8487
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0641716718673706,
      "learning_rate": 1.7310869354541868e-05,
      "loss": 0.7244,
      "step": 8488
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.0489912033081055,
      "learning_rate": 1.729396999374918e-05,
      "loss": 0.7314,
      "step": 8489
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.0546748638153076,
      "learning_rate": 1.727707810497803e-05,
      "loss": 0.7245,
      "step": 8490
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.9147816896438599,
      "learning_rate": 1.726019368975449e-05,
      "loss": 0.7484,
      "step": 8491
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.0633926391601562,
      "learning_rate": 1.7243316749603953e-05,
      "loss": 0.7567,
      "step": 8492
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.3747823238372803,
      "learning_rate": 1.7226447286051205e-05,
      "loss": 0.9072,
      "step": 8493
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.119113802909851,
      "learning_rate": 1.720958530062029e-05,
      "loss": 0.7639,
      "step": 8494
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.0871083736419678,
      "learning_rate": 1.7192730794834556e-05,
      "loss": 0.7009,
      "step": 8495
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.029677391052246,
      "learning_rate": 1.7175883770216782e-05,
      "loss": 0.7718,
      "step": 8496
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.0006197690963745,
      "learning_rate": 1.715904422828892e-05,
      "loss": 0.7307,
      "step": 8497
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.0989919900894165,
      "learning_rate": 1.714221217057237e-05,
      "loss": 0.6307,
      "step": 8498
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.0064196586608887,
      "learning_rate": 1.7125387598587862e-05,
      "loss": 0.7414,
      "step": 8499
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.2774009704589844,
      "learning_rate": 1.7108570513855294e-05,
      "loss": 0.7625,
      "step": 8500
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.0326265096664429,
      "learning_rate": 1.7091760917894072e-05,
      "loss": 0.7549,
      "step": 8501
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.1957862377166748,
      "learning_rate": 1.707495881222283e-05,
      "loss": 0.8141,
      "step": 8502
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.9958105683326721,
      "learning_rate": 1.7058164198359518e-05,
      "loss": 0.7295,
      "step": 8503
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.1243839263916016,
      "learning_rate": 1.7041377077821473e-05,
      "loss": 0.818,
      "step": 8504
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.1555945873260498,
      "learning_rate": 1.70245974521253e-05,
      "loss": 0.8248,
      "step": 8505
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.077142596244812,
      "learning_rate": 1.700782532278694e-05,
      "loss": 0.6863,
      "step": 8506
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.1040760278701782,
      "learning_rate": 1.699106069132165e-05,
      "loss": 0.831,
      "step": 8507
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.0852550268173218,
      "learning_rate": 1.6974303559244043e-05,
      "loss": 0.7538,
      "step": 8508
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.9948079586029053,
      "learning_rate": 1.6957553928068017e-05,
      "loss": 0.711,
      "step": 8509
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.127382516860962,
      "learning_rate": 1.6940811799306787e-05,
      "loss": 0.8126,
      "step": 8510
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.0706111192703247,
      "learning_rate": 1.6924077174472964e-05,
      "loss": 0.7377,
      "step": 8511
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.056753158569336,
      "learning_rate": 1.6907350055078385e-05,
      "loss": 0.7637,
      "step": 8512
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.017273187637329,
      "learning_rate": 1.6890630442634237e-05,
      "loss": 0.8499,
      "step": 8513
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.9729218482971191,
      "learning_rate": 1.6873918338651084e-05,
      "loss": 0.5847,
      "step": 8514
    },
    {
      "epoch": 3.19,
      "grad_norm": 1.0979191064834595,
      "learning_rate": 1.6857213744638757e-05,
      "loss": 0.8617,
      "step": 8515
    },
    {
      "epoch": 3.19,
      "eval_loss": 1.024443507194519,
      "eval_runtime": 17.8152,
      "eval_samples_per_second": 153.183,
      "eval_steps_per_second": 38.338,
      "step": 8515
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0588101148605347,
      "learning_rate": 1.684051666210642e-05,
      "loss": 0.732,
      "step": 8516
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.9992765188217163,
      "learning_rate": 1.6823827092562517e-05,
      "loss": 0.6711,
      "step": 8517
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.1244597434997559,
      "learning_rate": 1.6807145037514927e-05,
      "loss": 0.8851,
      "step": 8518
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.172288179397583,
      "learning_rate": 1.6790470498470744e-05,
      "loss": 0.8216,
      "step": 8519
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0661565065383911,
      "learning_rate": 1.67738034769364e-05,
      "loss": 0.7913,
      "step": 8520
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.1494472026824951,
      "learning_rate": 1.67571439744177e-05,
      "loss": 0.7965,
      "step": 8521
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.1071324348449707,
      "learning_rate": 1.6740491992419738e-05,
      "loss": 0.7102,
      "step": 8522
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.1426312923431396,
      "learning_rate": 1.6723847532446868e-05,
      "loss": 0.6675,
      "step": 8523
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0886788368225098,
      "learning_rate": 1.6707210596002908e-05,
      "loss": 0.7135,
      "step": 8524
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.00278902053833,
      "learning_rate": 1.6690581184590858e-05,
      "loss": 0.7578,
      "step": 8525
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0630741119384766,
      "learning_rate": 1.6673959299713092e-05,
      "loss": 0.7996,
      "step": 8526
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0693148374557495,
      "learning_rate": 1.66573449428713e-05,
      "loss": 0.6811,
      "step": 8527
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0704936981201172,
      "learning_rate": 1.6640738115566535e-05,
      "loss": 0.6992,
      "step": 8528
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0450294017791748,
      "learning_rate": 1.6624138819299096e-05,
      "loss": 0.5514,
      "step": 8529
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.9689821600914001,
      "learning_rate": 1.660754705556862e-05,
      "loss": 0.678,
      "step": 8530
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0036914348602295,
      "learning_rate": 1.6590962825874146e-05,
      "loss": 0.7014,
      "step": 8531
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.012209415435791,
      "learning_rate": 1.657438613171387e-05,
      "loss": 0.6978,
      "step": 8532
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.1781607866287231,
      "learning_rate": 1.655781697458546e-05,
      "loss": 0.8294,
      "step": 8533
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.04071044921875,
      "learning_rate": 1.6541255355985885e-05,
      "loss": 0.665,
      "step": 8534
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0346769094467163,
      "learning_rate": 1.6524701277411294e-05,
      "loss": 0.6857,
      "step": 8535
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0179566144943237,
      "learning_rate": 1.650815474035733e-05,
      "loss": 0.7526,
      "step": 8536
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.9964904189109802,
      "learning_rate": 1.6491615746318857e-05,
      "loss": 0.6176,
      "step": 8537
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.1095975637435913,
      "learning_rate": 1.6475084296790055e-05,
      "loss": 0.7163,
      "step": 8538
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.030268669128418,
      "learning_rate": 1.6458560393264498e-05,
      "loss": 0.7735,
      "step": 8539
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0631005764007568,
      "learning_rate": 1.6442044037234992e-05,
      "loss": 0.8213,
      "step": 8540
    },
    {
      "epoch": 3.2,
      "grad_norm": 1.0204867124557495,
      "learning_rate": 1.6425535230193712e-05,
      "loss": 0.6159,
      "step": 8541
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0659743547439575,
      "learning_rate": 1.6409033973632093e-05,
      "loss": 0.5956,
      "step": 8542
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0134565830230713,
      "learning_rate": 1.639254026904099e-05,
      "loss": 0.7202,
      "step": 8543
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0637378692626953,
      "learning_rate": 1.6376054117910488e-05,
      "loss": 0.6858,
      "step": 8544
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.1654237508773804,
      "learning_rate": 1.6359575521730008e-05,
      "loss": 0.7727,
      "step": 8545
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.005456805229187,
      "learning_rate": 1.6343104481988314e-05,
      "loss": 0.7331,
      "step": 8546
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.046491265296936,
      "learning_rate": 1.6326641000173482e-05,
      "loss": 0.6548,
      "step": 8547
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0644466876983643,
      "learning_rate": 1.6310185077772845e-05,
      "loss": 0.8189,
      "step": 8548
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.129550576210022,
      "learning_rate": 1.6293736716273156e-05,
      "loss": 0.7129,
      "step": 8549
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.9435514807701111,
      "learning_rate": 1.6277295917160418e-05,
      "loss": 0.7029,
      "step": 8550
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.1534204483032227,
      "learning_rate": 1.6260862681919963e-05,
      "loss": 0.7695,
      "step": 8551
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0409247875213623,
      "learning_rate": 1.6244437012036407e-05,
      "loss": 0.7999,
      "step": 8552
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.9615367650985718,
      "learning_rate": 1.622801890899376e-05,
      "loss": 0.781,
      "step": 8553
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.2633025646209717,
      "learning_rate": 1.6211608374275288e-05,
      "loss": 0.6339,
      "step": 8554
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0522356033325195,
      "learning_rate": 1.6195205409363577e-05,
      "loss": 0.7447,
      "step": 8555
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.074739694595337,
      "learning_rate": 1.6178810015740575e-05,
      "loss": 0.7639,
      "step": 8556
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.2386527061462402,
      "learning_rate": 1.6162422194887495e-05,
      "loss": 0.7242,
      "step": 8557
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0208560228347778,
      "learning_rate": 1.6146041948284863e-05,
      "loss": 0.6953,
      "step": 8558
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0259406566619873,
      "learning_rate": 1.6129669277412575e-05,
      "loss": 0.6951,
      "step": 8559
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.9286192059516907,
      "learning_rate": 1.61133041837498e-05,
      "loss": 0.5632,
      "step": 8560
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.2213603258132935,
      "learning_rate": 1.6096946668775027e-05,
      "loss": 0.7521,
      "step": 8561
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0560847520828247,
      "learning_rate": 1.608059673396605e-05,
      "loss": 0.7106,
      "step": 8562
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.9044788479804993,
      "learning_rate": 1.6064254380800015e-05,
      "loss": 0.6428,
      "step": 8563
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0410083532333374,
      "learning_rate": 1.604791961075336e-05,
      "loss": 0.7179,
      "step": 8564
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.0153900384902954,
      "learning_rate": 1.603159242530181e-05,
      "loss": 0.7029,
      "step": 8565
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.1317006349563599,
      "learning_rate": 1.60152728259205e-05,
      "loss": 0.8345,
      "step": 8566
    },
    {
      "epoch": 3.21,
      "grad_norm": 1.1148796081542969,
      "learning_rate": 1.599896081408373e-05,
      "loss": 0.5944,
      "step": 8567
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.1138668060302734,
      "learning_rate": 1.598265639126524e-05,
      "loss": 0.7457,
      "step": 8568
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.0455981492996216,
      "learning_rate": 1.5966359558938083e-05,
      "loss": 0.5945,
      "step": 8569
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.0855661630630493,
      "learning_rate": 1.5950070318574495e-05,
      "loss": 0.8794,
      "step": 8570
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.9405040740966797,
      "learning_rate": 1.5933788671646177e-05,
      "loss": 0.6542,
      "step": 8571
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.9426544904708862,
      "learning_rate": 1.5917514619624086e-05,
      "loss": 0.6317,
      "step": 8572
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.1999326944351196,
      "learning_rate": 1.5901248163978444e-05,
      "loss": 0.7998,
      "step": 8573
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.1196178197860718,
      "learning_rate": 1.5884989306178898e-05,
      "loss": 0.7533,
      "step": 8574
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.26893150806427,
      "learning_rate": 1.5868738047694298e-05,
      "loss": 0.7444,
      "step": 8575
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.1446245908737183,
      "learning_rate": 1.5852494389992866e-05,
      "loss": 0.7874,
      "step": 8576
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.1162298917770386,
      "learning_rate": 1.5836258334542108e-05,
      "loss": 0.672,
      "step": 8577
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.9779715538024902,
      "learning_rate": 1.5820029882808875e-05,
      "loss": 0.71,
      "step": 8578
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.0594226121902466,
      "learning_rate": 1.5803809036259364e-05,
      "loss": 0.6393,
      "step": 8579
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.1087273359298706,
      "learning_rate": 1.5787595796358934e-05,
      "loss": 0.6155,
      "step": 8580
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.1025971174240112,
      "learning_rate": 1.5771390164572442e-05,
      "loss": 0.6663,
      "step": 8581
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.0680676698684692,
      "learning_rate": 1.575519214236395e-05,
      "loss": 0.6824,
      "step": 8582
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.0872606039047241,
      "learning_rate": 1.5739001731196834e-05,
      "loss": 0.7579,
      "step": 8583
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.043330192565918,
      "learning_rate": 1.5722818932533846e-05,
      "loss": 0.7228,
      "step": 8584
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.933159589767456,
      "learning_rate": 1.570664374783699e-05,
      "loss": 0.7782,
      "step": 8585
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.1514307260513306,
      "learning_rate": 1.5690476178567602e-05,
      "loss": 0.8036,
      "step": 8586
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.9390965700149536,
      "learning_rate": 1.5674316226186315e-05,
      "loss": 0.5448,
      "step": 8587
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.1575497388839722,
      "learning_rate": 1.5658163892153123e-05,
      "loss": 0.8682,
      "step": 8588
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.2446956634521484,
      "learning_rate": 1.5642019177927293e-05,
      "loss": 0.6674,
      "step": 8589
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.154819369316101,
      "learning_rate": 1.562588208496736e-05,
      "loss": 0.7743,
      "step": 8590
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.0550135374069214,
      "learning_rate": 1.5609752614731288e-05,
      "loss": 0.7196,
      "step": 8591
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.9840099215507507,
      "learning_rate": 1.5593630768676248e-05,
      "loss": 0.8009,
      "step": 8592
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.001564383506775,
      "learning_rate": 1.557751654825873e-05,
      "loss": 0.69,
      "step": 8593
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.978463351726532,
      "learning_rate": 1.5561409954934626e-05,
      "loss": 0.595,
      "step": 8594
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.1010429859161377,
      "learning_rate": 1.5545310990159034e-05,
      "loss": 0.8155,
      "step": 8595
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.0334906578063965,
      "learning_rate": 1.552921965538642e-05,
      "loss": 0.7159,
      "step": 8596
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.9989932179450989,
      "learning_rate": 1.5513135952070513e-05,
      "loss": 0.5766,
      "step": 8597
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.0468556880950928,
      "learning_rate": 1.5497059881664434e-05,
      "loss": 0.742,
      "step": 8598
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.1808243989944458,
      "learning_rate": 1.5480991445620542e-05,
      "loss": 0.6228,
      "step": 8599
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.983851969242096,
      "learning_rate": 1.5464930645390497e-05,
      "loss": 0.7298,
      "step": 8600
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.2330890893936157,
      "learning_rate": 1.5448877482425383e-05,
      "loss": 0.8293,
      "step": 8601
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.9612021446228027,
      "learning_rate": 1.543283195817541e-05,
      "loss": 0.6732,
      "step": 8602
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.1558693647384644,
      "learning_rate": 1.5416794074090258e-05,
      "loss": 0.7459,
      "step": 8603
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.186774730682373,
      "learning_rate": 1.540076383161889e-05,
      "loss": 0.5871,
      "step": 8604
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.099134922027588,
      "learning_rate": 1.5384741232209475e-05,
      "loss": 0.8113,
      "step": 8605
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.1274503469467163,
      "learning_rate": 1.536872627730962e-05,
      "loss": 0.8418,
      "step": 8606
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.1017522811889648,
      "learning_rate": 1.5352718968366143e-05,
      "loss": 0.5725,
      "step": 8607
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.9729222655296326,
      "learning_rate": 1.5336719306825254e-05,
      "loss": 0.7812,
      "step": 8608
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.1359949111938477,
      "learning_rate": 1.532072729413241e-05,
      "loss": 0.5936,
      "step": 8609
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.124929428100586,
      "learning_rate": 1.5304742931732384e-05,
      "loss": 0.7215,
      "step": 8610
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.090544581413269,
      "learning_rate": 1.528876622106934e-05,
      "loss": 0.7558,
      "step": 8611
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.1104398965835571,
      "learning_rate": 1.5272797163586584e-05,
      "loss": 0.6447,
      "step": 8612
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.1279990673065186,
      "learning_rate": 1.5256835760726873e-05,
      "loss": 0.7851,
      "step": 8613
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.0726474523544312,
      "learning_rate": 1.524088201393229e-05,
      "loss": 0.8126,
      "step": 8614
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.9466400742530823,
      "learning_rate": 1.5224935924644069e-05,
      "loss": 0.6535,
      "step": 8615
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.9639216065406799,
      "learning_rate": 1.5208997494302912e-05,
      "loss": 0.678,
      "step": 8616
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.9507066011428833,
      "learning_rate": 1.5193066724348748e-05,
      "loss": 0.6841,
      "step": 8617
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.081803560256958,
      "learning_rate": 1.5177143616220801e-05,
      "loss": 0.6915,
      "step": 8618
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.0938347578048706,
      "learning_rate": 1.5161228171357699e-05,
      "loss": 0.777,
      "step": 8619
    },
    {
      "epoch": 3.23,
      "grad_norm": 1.016236662864685,
      "learning_rate": 1.514532039119726e-05,
      "loss": 0.7903,
      "step": 8620
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.954439103603363,
      "learning_rate": 1.5129420277176686e-05,
      "loss": 0.6745,
      "step": 8621
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.1077604293823242,
      "learning_rate": 1.5113527830732433e-05,
      "loss": 0.6794,
      "step": 8622
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0425703525543213,
      "learning_rate": 1.5097643053300336e-05,
      "loss": 0.6571,
      "step": 8623
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0474046468734741,
      "learning_rate": 1.5081765946315462e-05,
      "loss": 0.7083,
      "step": 8624
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0750906467437744,
      "learning_rate": 1.5065896511212219e-05,
      "loss": 0.7906,
      "step": 8625
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.188757300376892,
      "learning_rate": 1.5050034749424346e-05,
      "loss": 0.7777,
      "step": 8626
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.1414625644683838,
      "learning_rate": 1.5034180662384857e-05,
      "loss": 0.7967,
      "step": 8627
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.9663220643997192,
      "learning_rate": 1.5018334251526045e-05,
      "loss": 0.7399,
      "step": 8628
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.9402247071266174,
      "learning_rate": 1.500249551827958e-05,
      "loss": 0.666,
      "step": 8629
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.951943576335907,
      "learning_rate": 1.4986664464076406e-05,
      "loss": 0.5305,
      "step": 8630
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.1231836080551147,
      "learning_rate": 1.4970841090346743e-05,
      "loss": 0.7541,
      "step": 8631
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0575339794158936,
      "learning_rate": 1.495502539852014e-05,
      "loss": 0.6026,
      "step": 8632
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.1052583456039429,
      "learning_rate": 1.4939217390025484e-05,
      "loss": 0.8259,
      "step": 8633
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0891767740249634,
      "learning_rate": 1.4923417066290923e-05,
      "loss": 0.5539,
      "step": 8634
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0149139165878296,
      "learning_rate": 1.4907624428743905e-05,
      "loss": 0.5651,
      "step": 8635
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.1041271686553955,
      "learning_rate": 1.4891839478811254e-05,
      "loss": 0.7539,
      "step": 8636
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0739388465881348,
      "learning_rate": 1.4876062217919028e-05,
      "loss": 0.9052,
      "step": 8637
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0401972532272339,
      "learning_rate": 1.486029264749258e-05,
      "loss": 0.7009,
      "step": 8638
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0499413013458252,
      "learning_rate": 1.4844530768956656e-05,
      "loss": 0.7223,
      "step": 8639
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0902019739151,
      "learning_rate": 1.4828776583735226e-05,
      "loss": 0.7638,
      "step": 8640
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0302083492279053,
      "learning_rate": 1.4813030093251601e-05,
      "loss": 0.7538,
      "step": 8641
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0842262506484985,
      "learning_rate": 1.479729129892835e-05,
      "loss": 0.7025,
      "step": 8642
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.1679307222366333,
      "learning_rate": 1.478156020218745e-05,
      "loss": 0.7282,
      "step": 8643
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.2327594757080078,
      "learning_rate": 1.4765836804450062e-05,
      "loss": 0.6795,
      "step": 8644
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.1471444368362427,
      "learning_rate": 1.4750121107136717e-05,
      "loss": 0.7757,
      "step": 8645
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.1676257848739624,
      "learning_rate": 1.4734413111667279e-05,
      "loss": 0.7545,
      "step": 8646
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0293692350387573,
      "learning_rate": 1.47187128194608e-05,
      "loss": 0.6328,
      "step": 8647
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.1812989711761475,
      "learning_rate": 1.470302023193576e-05,
      "loss": 0.9163,
      "step": 8648
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0162055492401123,
      "learning_rate": 1.4687335350509945e-05,
      "loss": 0.8524,
      "step": 8649
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0623490810394287,
      "learning_rate": 1.4671658176600288e-05,
      "loss": 0.6188,
      "step": 8650
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.1873791217803955,
      "learning_rate": 1.4655988711623203e-05,
      "loss": 0.7706,
      "step": 8651
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.1203824281692505,
      "learning_rate": 1.4640326956994333e-05,
      "loss": 0.7751,
      "step": 8652
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.9470168352127075,
      "learning_rate": 1.462467291412859e-05,
      "loss": 0.4697,
      "step": 8653
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0874981880187988,
      "learning_rate": 1.4609026584440277e-05,
      "loss": 0.7878,
      "step": 8654
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.1855725049972534,
      "learning_rate": 1.459338796934293e-05,
      "loss": 0.7535,
      "step": 8655
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.9554637670516968,
      "learning_rate": 1.457775707024942e-05,
      "loss": 0.8223,
      "step": 8656
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0436716079711914,
      "learning_rate": 1.456213388857186e-05,
      "loss": 0.6818,
      "step": 8657
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0497807264328003,
      "learning_rate": 1.454651842572179e-05,
      "loss": 0.5926,
      "step": 8658
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.9534757137298584,
      "learning_rate": 1.4530910683109933e-05,
      "loss": 0.6847,
      "step": 8659
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.4324437379837036,
      "learning_rate": 1.4515310662146353e-05,
      "loss": 0.8848,
      "step": 8660
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.312010407447815,
      "learning_rate": 1.4499718364240466e-05,
      "loss": 0.6975,
      "step": 8661
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.912691593170166,
      "learning_rate": 1.4484133790800925e-05,
      "loss": 0.7674,
      "step": 8662
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.9492802619934082,
      "learning_rate": 1.4468556943235678e-05,
      "loss": 0.65,
      "step": 8663
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.9945290088653564,
      "learning_rate": 1.4452987822952068e-05,
      "loss": 0.632,
      "step": 8664
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0324013233184814,
      "learning_rate": 1.4437426431356627e-05,
      "loss": 0.6922,
      "step": 8665
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.9701200723648071,
      "learning_rate": 1.442187276985526e-05,
      "loss": 0.7314,
      "step": 8666
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.101178526878357,
      "learning_rate": 1.4406326839853124e-05,
      "loss": 0.6631,
      "step": 8667
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0897502899169922,
      "learning_rate": 1.439078864275476e-05,
      "loss": 0.7812,
      "step": 8668
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0324066877365112,
      "learning_rate": 1.4375258179963913e-05,
      "loss": 0.6217,
      "step": 8669
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0438801050186157,
      "learning_rate": 1.435973545288366e-05,
      "loss": 0.5779,
      "step": 8670
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.1602433919906616,
      "learning_rate": 1.4344220462916435e-05,
      "loss": 0.682,
      "step": 8671
    },
    {
      "epoch": 3.25,
      "grad_norm": 1.0837311744689941,
      "learning_rate": 1.4328713211463918e-05,
      "loss": 0.6618,
      "step": 8672
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.2518019676208496,
      "learning_rate": 1.4313213699927074e-05,
      "loss": 0.6676,
      "step": 8673
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.1111434698104858,
      "learning_rate": 1.4297721929706232e-05,
      "loss": 0.6813,
      "step": 8674
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.1337478160858154,
      "learning_rate": 1.4282237902200957e-05,
      "loss": 0.8429,
      "step": 8675
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.1212034225463867,
      "learning_rate": 1.4266761618810166e-05,
      "loss": 0.8792,
      "step": 8676
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.0106837749481201,
      "learning_rate": 1.4251293080932027e-05,
      "loss": 0.6951,
      "step": 8677
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.9145600199699402,
      "learning_rate": 1.4235832289964058e-05,
      "loss": 0.5715,
      "step": 8678
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.0105172395706177,
      "learning_rate": 1.422037924730305e-05,
      "loss": 0.6379,
      "step": 8679
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.9723754525184631,
      "learning_rate": 1.4204933954345078e-05,
      "loss": 0.7053,
      "step": 8680
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.1022402048110962,
      "learning_rate": 1.4189496412485592e-05,
      "loss": 0.7015,
      "step": 8681
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.9736297726631165,
      "learning_rate": 1.4174066623119209e-05,
      "loss": 0.6791,
      "step": 8682
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.9871993660926819,
      "learning_rate": 1.4158644587639958e-05,
      "loss": 0.7333,
      "step": 8683
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.1076475381851196,
      "learning_rate": 1.414323030744118e-05,
      "loss": 0.5906,
      "step": 8684
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.9620296955108643,
      "learning_rate": 1.4127823783915384e-05,
      "loss": 0.6474,
      "step": 8685
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.2747806310653687,
      "learning_rate": 1.411242501845451e-05,
      "loss": 0.9142,
      "step": 8686
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.1167364120483398,
      "learning_rate": 1.409703401244975e-05,
      "loss": 0.848,
      "step": 8687
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.0693984031677246,
      "learning_rate": 1.4081650767291566e-05,
      "loss": 0.7764,
      "step": 8688
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.134645938873291,
      "learning_rate": 1.406627528436979e-05,
      "loss": 0.6858,
      "step": 8689
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.0922201871871948,
      "learning_rate": 1.405090756507349e-05,
      "loss": 0.7843,
      "step": 8690
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.0321450233459473,
      "learning_rate": 1.4035547610791045e-05,
      "loss": 0.6242,
      "step": 8691
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.1480622291564941,
      "learning_rate": 1.402019542291011e-05,
      "loss": 0.5951,
      "step": 8692
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.991938054561615,
      "learning_rate": 1.4004851002817743e-05,
      "loss": 0.7209,
      "step": 8693
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.1692122220993042,
      "learning_rate": 1.3989514351900169e-05,
      "loss": 0.7154,
      "step": 8694
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.0382134914398193,
      "learning_rate": 1.3974185471542967e-05,
      "loss": 0.6915,
      "step": 8695
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.116464376449585,
      "learning_rate": 1.3958864363131052e-05,
      "loss": 0.5436,
      "step": 8696
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.9933493733406067,
      "learning_rate": 1.3943551028048585e-05,
      "loss": 0.7195,
      "step": 8697
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.1251953840255737,
      "learning_rate": 1.3928245467679003e-05,
      "loss": 0.7727,
      "step": 8698
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.0879409313201904,
      "learning_rate": 1.391294768340513e-05,
      "loss": 0.7749,
      "step": 8699
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.1146827936172485,
      "learning_rate": 1.389765767660901e-05,
      "loss": 0.7039,
      "step": 8700
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.0438424348831177,
      "learning_rate": 1.3882375448672013e-05,
      "loss": 0.6917,
      "step": 8701
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.1099449396133423,
      "learning_rate": 1.386710100097477e-05,
      "loss": 0.8279,
      "step": 8702
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.9072893261909485,
      "learning_rate": 1.3851834334897296e-05,
      "loss": 0.6391,
      "step": 8703
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.0982481241226196,
      "learning_rate": 1.3836575451818811e-05,
      "loss": 0.766,
      "step": 8704
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.0673723220825195,
      "learning_rate": 1.382132435311786e-05,
      "loss": 0.6397,
      "step": 8705
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.1340081691741943,
      "learning_rate": 1.3806081040172336e-05,
      "loss": 0.9067,
      "step": 8706
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.9446349143981934,
      "learning_rate": 1.3790845514359363e-05,
      "loss": 0.6443,
      "step": 8707
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.8833301663398743,
      "learning_rate": 1.3775617777055362e-05,
      "loss": 0.5435,
      "step": 8708
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.1198652982711792,
      "learning_rate": 1.3760397829636107e-05,
      "loss": 0.6997,
      "step": 8709
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.961736798286438,
      "learning_rate": 1.3745185673476624e-05,
      "loss": 0.7083,
      "step": 8710
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.1060811281204224,
      "learning_rate": 1.3729981309951245e-05,
      "loss": 0.8627,
      "step": 8711
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.188815712928772,
      "learning_rate": 1.3714784740433572e-05,
      "loss": 0.8701,
      "step": 8712
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.0773437023162842,
      "learning_rate": 1.3699595966296575e-05,
      "loss": 0.8038,
      "step": 8713
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.1758430004119873,
      "learning_rate": 1.3684414988912442e-05,
      "loss": 0.7178,
      "step": 8714
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.9726375937461853,
      "learning_rate": 1.3669241809652689e-05,
      "loss": 0.7951,
      "step": 8715
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.1504243612289429,
      "learning_rate": 1.3654076429888174e-05,
      "loss": 0.6706,
      "step": 8716
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.5023800134658813,
      "learning_rate": 1.3638918850988913e-05,
      "loss": 0.6932,
      "step": 8717
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.0286999940872192,
      "learning_rate": 1.3623769074324366e-05,
      "loss": 0.5883,
      "step": 8718
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.0527045726776123,
      "learning_rate": 1.360862710126326e-05,
      "loss": 0.6929,
      "step": 8719
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.133133053779602,
      "learning_rate": 1.3593492933173512e-05,
      "loss": 0.8106,
      "step": 8720
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.1675955057144165,
      "learning_rate": 1.3578366571422462e-05,
      "loss": 0.6847,
      "step": 8721
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.9431855082511902,
      "learning_rate": 1.3563248017376685e-05,
      "loss": 0.7171,
      "step": 8722
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.1312373876571655,
      "learning_rate": 1.3548137272402006e-05,
      "loss": 0.6124,
      "step": 8723
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.1006453037261963,
      "learning_rate": 1.353303433786367e-05,
      "loss": 0.8427,
      "step": 8724
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.9970380663871765,
      "learning_rate": 1.3517939215126107e-05,
      "loss": 0.7132,
      "step": 8725
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.1562236547470093,
      "learning_rate": 1.3502851905553071e-05,
      "loss": 0.7736,
      "step": 8726
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0601738691329956,
      "learning_rate": 1.3487772410507592e-05,
      "loss": 0.7469,
      "step": 8727
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0913394689559937,
      "learning_rate": 1.3472700731352084e-05,
      "loss": 0.7211,
      "step": 8728
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.2850486040115356,
      "learning_rate": 1.345763686944813e-05,
      "loss": 0.7933,
      "step": 8729
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.9110239744186401,
      "learning_rate": 1.3442580826156681e-05,
      "loss": 0.7444,
      "step": 8730
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.1767686605453491,
      "learning_rate": 1.3427532602837977e-05,
      "loss": 0.8879,
      "step": 8731
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0180143117904663,
      "learning_rate": 1.3412492200851523e-05,
      "loss": 0.7334,
      "step": 8732
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0770057439804077,
      "learning_rate": 1.339745962155613e-05,
      "loss": 0.8614,
      "step": 8733
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.9675548672676086,
      "learning_rate": 1.3382434866309945e-05,
      "loss": 0.6501,
      "step": 8734
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0657414197921753,
      "learning_rate": 1.3367417936470328e-05,
      "loss": 0.6647,
      "step": 8735
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0963503122329712,
      "learning_rate": 1.3352408833394004e-05,
      "loss": 0.778,
      "step": 8736
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.146285057067871,
      "learning_rate": 1.3337407558436909e-05,
      "loss": 0.6269,
      "step": 8737
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0742048025131226,
      "learning_rate": 1.3322414112954384e-05,
      "loss": 0.8457,
      "step": 8738
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.1320629119873047,
      "learning_rate": 1.3307428498300978e-05,
      "loss": 0.6728,
      "step": 8739
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0801959037780762,
      "learning_rate": 1.329245071583054e-05,
      "loss": 0.642,
      "step": 8740
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.2920372486114502,
      "learning_rate": 1.3277480766896256e-05,
      "loss": 0.7352,
      "step": 8741
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0408315658569336,
      "learning_rate": 1.3262518652850564e-05,
      "loss": 0.8106,
      "step": 8742
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0819870233535767,
      "learning_rate": 1.3247564375045184e-05,
      "loss": 0.7297,
      "step": 8743
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.9985828995704651,
      "learning_rate": 1.3232617934831205e-05,
      "loss": 0.6875,
      "step": 8744
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0895895957946777,
      "learning_rate": 1.3217679333558908e-05,
      "loss": 0.7017,
      "step": 8745
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0744102001190186,
      "learning_rate": 1.3202748572577927e-05,
      "loss": 0.6248,
      "step": 8746
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.0612046718597412,
      "learning_rate": 1.318782565323714e-05,
      "loss": 0.6461,
      "step": 8747
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.1198526620864868,
      "learning_rate": 1.31729105768848e-05,
      "loss": 0.7627,
      "step": 8748
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.1907988786697388,
      "learning_rate": 1.3158003344868375e-05,
      "loss": 0.8191,
      "step": 8749
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.981118381023407,
      "learning_rate": 1.3143103958534631e-05,
      "loss": 0.6671,
      "step": 8750
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.9814044833183289,
      "learning_rate": 1.3128212419229713e-05,
      "loss": 0.5717,
      "step": 8751
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.9863961338996887,
      "learning_rate": 1.311332872829888e-05,
      "loss": 0.7015,
      "step": 8752
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.032778263092041,
      "learning_rate": 1.3098452887086854e-05,
      "loss": 0.6434,
      "step": 8753
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0656863451004028,
      "learning_rate": 1.3083584896937607e-05,
      "loss": 0.8532,
      "step": 8754
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.2324187755584717,
      "learning_rate": 1.3068724759194317e-05,
      "loss": 0.6971,
      "step": 8755
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0473054647445679,
      "learning_rate": 1.3053872475199546e-05,
      "loss": 0.7747,
      "step": 8756
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.198223352432251,
      "learning_rate": 1.3039028046295121e-05,
      "loss": 0.7873,
      "step": 8757
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0560699701309204,
      "learning_rate": 1.3024191473822111e-05,
      "loss": 0.6626,
      "step": 8758
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0552510023117065,
      "learning_rate": 1.300936275912098e-05,
      "loss": 0.8182,
      "step": 8759
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0969691276550293,
      "learning_rate": 1.2994541903531354e-05,
      "loss": 0.9077,
      "step": 8760
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.1323542594909668,
      "learning_rate": 1.2979728908392285e-05,
      "loss": 0.7804,
      "step": 8761
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.1235944032669067,
      "learning_rate": 1.296492377504196e-05,
      "loss": 0.8568,
      "step": 8762
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0323656797409058,
      "learning_rate": 1.2950126504817983e-05,
      "loss": 0.7398,
      "step": 8763
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.281171441078186,
      "learning_rate": 1.2935337099057243e-05,
      "loss": 0.7319,
      "step": 8764
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.104560136795044,
      "learning_rate": 1.2920555559095792e-05,
      "loss": 0.502,
      "step": 8765
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.056533932685852,
      "learning_rate": 1.290578188626912e-05,
      "loss": 0.8361,
      "step": 8766
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.1431658267974854,
      "learning_rate": 1.2891016081911945e-05,
      "loss": 0.7043,
      "step": 8767
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0929473638534546,
      "learning_rate": 1.2876258147358223e-05,
      "loss": 0.7285,
      "step": 8768
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.2290197610855103,
      "learning_rate": 1.286150808394131e-05,
      "loss": 0.5832,
      "step": 8769
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.132674217224121,
      "learning_rate": 1.2846765892993761e-05,
      "loss": 0.719,
      "step": 8770
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.072763442993164,
      "learning_rate": 1.2832031575847448e-05,
      "loss": 0.7739,
      "step": 8771
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0001726150512695,
      "learning_rate": 1.2817305133833524e-05,
      "loss": 0.6175,
      "step": 8772
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.9843897223472595,
      "learning_rate": 1.2802586568282482e-05,
      "loss": 0.7223,
      "step": 8773
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0138589143753052,
      "learning_rate": 1.2787875880524025e-05,
      "loss": 0.6755,
      "step": 8774
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.9594178199768066,
      "learning_rate": 1.2773173071887167e-05,
      "loss": 0.5791,
      "step": 8775
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.1643625497817993,
      "learning_rate": 1.275847814370027e-05,
      "loss": 0.7137,
      "step": 8776
    },
    {
      "epoch": 3.29,
      "grad_norm": 1.0750961303710938,
      "learning_rate": 1.2743791097290914e-05,
      "loss": 0.7487,
      "step": 8777
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.0453602075576782,
      "learning_rate": 1.272911193398596e-05,
      "loss": 0.6524,
      "step": 8778
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.133283019065857,
      "learning_rate": 1.2714440655111649e-05,
      "loss": 0.65,
      "step": 8779
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.0005536079406738,
      "learning_rate": 1.2699777261993407e-05,
      "loss": 0.6985,
      "step": 8780
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.1355688571929932,
      "learning_rate": 1.2685121755955998e-05,
      "loss": 0.6679,
      "step": 8781
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.118326187133789,
      "learning_rate": 1.267047413832344e-05,
      "loss": 0.6673,
      "step": 8782
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.031341314315796,
      "learning_rate": 1.265583441041911e-05,
      "loss": 0.767,
      "step": 8783
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.01856529712677,
      "learning_rate": 1.2641202573565591e-05,
      "loss": 0.7786,
      "step": 8784
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.0761728286743164,
      "learning_rate": 1.2626578629084784e-05,
      "loss": 0.6338,
      "step": 8785
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.0103627443313599,
      "learning_rate": 1.2611962578297931e-05,
      "loss": 0.5969,
      "step": 8786
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.9296401739120483,
      "learning_rate": 1.259735442252542e-05,
      "loss": 0.6567,
      "step": 8787
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.0846707820892334,
      "learning_rate": 1.2582754163087073e-05,
      "loss": 0.7595,
      "step": 8788
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.1962271928787231,
      "learning_rate": 1.2568161801301948e-05,
      "loss": 0.6924,
      "step": 8789
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.16973078250885,
      "learning_rate": 1.2553577338488376e-05,
      "loss": 0.899,
      "step": 8790
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.141801118850708,
      "learning_rate": 1.253900077596396e-05,
      "loss": 0.6669,
      "step": 8791
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.0035253763198853,
      "learning_rate": 1.2524432115045603e-05,
      "loss": 0.6542,
      "step": 8792
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.008530616760254,
      "learning_rate": 1.250987135704954e-05,
      "loss": 0.7201,
      "step": 8793
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.025848388671875,
      "learning_rate": 1.2495318503291242e-05,
      "loss": 0.7825,
      "step": 8794
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.1656174659729004,
      "learning_rate": 1.2480773555085434e-05,
      "loss": 0.7797,
      "step": 8795
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.9682310819625854,
      "learning_rate": 1.2466236513746255e-05,
      "loss": 0.8047,
      "step": 8796
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.0170296430587769,
      "learning_rate": 1.245170738058694e-05,
      "loss": 0.7845,
      "step": 8797
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.0576947927474976,
      "learning_rate": 1.2437186156920167e-05,
      "loss": 0.7072,
      "step": 8798
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.0234044790267944,
      "learning_rate": 1.2422672844057892e-05,
      "loss": 0.6913,
      "step": 8799
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.1672894954681396,
      "learning_rate": 1.2408167443311214e-05,
      "loss": 0.6537,
      "step": 8800
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.8794935345649719,
      "learning_rate": 1.2393669955990695e-05,
      "loss": 0.6329,
      "step": 8801
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.043129324913025,
      "learning_rate": 1.2379180383406064e-05,
      "loss": 0.8225,
      "step": 8802
    },
    {
      "epoch": 3.3,
      "grad_norm": 1.11471426486969,
      "learning_rate": 1.2364698726866363e-05,
      "loss": 0.7826,
      "step": 8803
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0735687017440796,
      "learning_rate": 1.2350224987679959e-05,
      "loss": 0.7282,
      "step": 8804
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0684387683868408,
      "learning_rate": 1.2335759167154448e-05,
      "loss": 0.7558,
      "step": 8805
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0510025024414062,
      "learning_rate": 1.2321301266596752e-05,
      "loss": 0.805,
      "step": 8806
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0960654020309448,
      "learning_rate": 1.2306851287313025e-05,
      "loss": 0.8148,
      "step": 8807
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.008629560470581,
      "learning_rate": 1.2292409230608793e-05,
      "loss": 0.8107,
      "step": 8808
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.2415306568145752,
      "learning_rate": 1.2277975097788785e-05,
      "loss": 0.7848,
      "step": 8809
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0439543724060059,
      "learning_rate": 1.2263548890157028e-05,
      "loss": 0.6609,
      "step": 8810
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0184595584869385,
      "learning_rate": 1.2249130609016879e-05,
      "loss": 0.6542,
      "step": 8811
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0871118307113647,
      "learning_rate": 1.223472025567095e-05,
      "loss": 0.7013,
      "step": 8812
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0364543199539185,
      "learning_rate": 1.222031783142109e-05,
      "loss": 0.6702,
      "step": 8813
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.9229085445404053,
      "learning_rate": 1.2205923337568526e-05,
      "loss": 0.614,
      "step": 8814
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.1470646858215332,
      "learning_rate": 1.2191536775413703e-05,
      "loss": 0.7004,
      "step": 8815
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.1500049829483032,
      "learning_rate": 1.2177158146256362e-05,
      "loss": 0.8315,
      "step": 8816
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.988416314125061,
      "learning_rate": 1.2162787451395507e-05,
      "loss": 0.7925,
      "step": 8817
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.9825804829597473,
      "learning_rate": 1.2148424692129489e-05,
      "loss": 0.6603,
      "step": 8818
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.1483981609344482,
      "learning_rate": 1.2134069869755893e-05,
      "loss": 0.7395,
      "step": 8819
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0763087272644043,
      "learning_rate": 1.211972298557157e-05,
      "loss": 0.7653,
      "step": 8820
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0980451107025146,
      "learning_rate": 1.2105384040872713e-05,
      "loss": 0.716,
      "step": 8821
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0419203042984009,
      "learning_rate": 1.2091053036954758e-05,
      "loss": 0.7936,
      "step": 8822
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.9901394248008728,
      "learning_rate": 1.20767299751124e-05,
      "loss": 0.6292,
      "step": 8823
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.3827600479125977,
      "learning_rate": 1.2062414856639681e-05,
      "loss": 0.7873,
      "step": 8824
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.9929487705230713,
      "learning_rate": 1.204810768282989e-05,
      "loss": 0.5475,
      "step": 8825
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.9862678647041321,
      "learning_rate": 1.203380845497558e-05,
      "loss": 0.6713,
      "step": 8826
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.092159390449524,
      "learning_rate": 1.2019517174368589e-05,
      "loss": 0.7058,
      "step": 8827
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.1625317335128784,
      "learning_rate": 1.2005233842300102e-05,
      "loss": 0.7345,
      "step": 8828
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.9911997318267822,
      "learning_rate": 1.1990958460060508e-05,
      "loss": 0.7823,
      "step": 8829
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.134833574295044,
      "learning_rate": 1.1976691028939501e-05,
      "loss": 0.7559,
      "step": 8830
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.1116830110549927,
      "learning_rate": 1.1962431550226105e-05,
      "loss": 0.8269,
      "step": 8831
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.9983357191085815,
      "learning_rate": 1.194818002520851e-05,
      "loss": 0.7501,
      "step": 8832
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.332642674446106,
      "learning_rate": 1.1933936455174311e-05,
      "loss": 0.7965,
      "step": 8833
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.1033611297607422,
      "learning_rate": 1.1919700841410353e-05,
      "loss": 0.7373,
      "step": 8834
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.9624487161636353,
      "learning_rate": 1.1905473185202688e-05,
      "loss": 0.6533,
      "step": 8835
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.0558937788009644,
      "learning_rate": 1.189125348783674e-05,
      "loss": 0.8103,
      "step": 8836
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.0096181631088257,
      "learning_rate": 1.1877041750597173e-05,
      "loss": 0.7347,
      "step": 8837
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.0504916906356812,
      "learning_rate": 1.1862837974767905e-05,
      "loss": 0.8081,
      "step": 8838
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.113984227180481,
      "learning_rate": 1.1848642161632228e-05,
      "loss": 0.8107,
      "step": 8839
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.0371501445770264,
      "learning_rate": 1.183445431247261e-05,
      "loss": 0.7102,
      "step": 8840
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.014523983001709,
      "learning_rate": 1.1820274428570855e-05,
      "loss": 0.6598,
      "step": 8841
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.2002360820770264,
      "learning_rate": 1.1806102511208017e-05,
      "loss": 0.812,
      "step": 8842
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.0149215459823608,
      "learning_rate": 1.1791938561664485e-05,
      "loss": 0.7072,
      "step": 8843
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.9724608659744263,
      "learning_rate": 1.1777782581219865e-05,
      "loss": 0.7378,
      "step": 8844
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.2411749362945557,
      "learning_rate": 1.176363457115306e-05,
      "loss": 0.6888,
      "step": 8845
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.074684500694275,
      "learning_rate": 1.1749494532742289e-05,
      "loss": 0.7973,
      "step": 8846
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.223426103591919,
      "learning_rate": 1.1735362467265031e-05,
      "loss": 0.8247,
      "step": 8847
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.093048095703125,
      "learning_rate": 1.1721238375997989e-05,
      "loss": 0.6482,
      "step": 8848
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.0123887062072754,
      "learning_rate": 1.1707122260217251e-05,
      "loss": 0.6629,
      "step": 8849
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.0067975521087646,
      "learning_rate": 1.1693014121198099e-05,
      "loss": 0.6467,
      "step": 8850
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.074493646621704,
      "learning_rate": 1.1678913960215133e-05,
      "loss": 0.6526,
      "step": 8851
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.1894903182983398,
      "learning_rate": 1.1664821778542202e-05,
      "loss": 0.6777,
      "step": 8852
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.191187858581543,
      "learning_rate": 1.1650737577452487e-05,
      "loss": 0.8576,
      "step": 8853
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.2908128499984741,
      "learning_rate": 1.1636661358218392e-05,
      "loss": 0.8005,
      "step": 8854
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.168337106704712,
      "learning_rate": 1.1622593122111624e-05,
      "loss": 0.8095,
      "step": 8855
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.0747157335281372,
      "learning_rate": 1.1608532870403188e-05,
      "loss": 0.765,
      "step": 8856
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.0196449756622314,
      "learning_rate": 1.1594480604363333e-05,
      "loss": 0.6492,
      "step": 8857
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.0936203002929688,
      "learning_rate": 1.158043632526159e-05,
      "loss": 0.7328,
      "step": 8858
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.9211946725845337,
      "learning_rate": 1.156640003436682e-05,
      "loss": 0.7009,
      "step": 8859
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.0295281410217285,
      "learning_rate": 1.1552371732947086e-05,
      "loss": 0.6633,
      "step": 8860
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.9417870044708252,
      "learning_rate": 1.1538351422269789e-05,
      "loss": 0.5197,
      "step": 8861
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.3714969158172607,
      "learning_rate": 1.1524339103601544e-05,
      "loss": 0.7637,
      "step": 8862
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.9123432636260986,
      "learning_rate": 1.1510334778208332e-05,
      "loss": 0.6447,
      "step": 8863
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.0048632621765137,
      "learning_rate": 1.1496338447355348e-05,
      "loss": 0.6153,
      "step": 8864
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.969779372215271,
      "learning_rate": 1.148235011230706e-05,
      "loss": 0.7262,
      "step": 8865
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.9770947098731995,
      "learning_rate": 1.1468369774327304e-05,
      "loss": 0.6826,
      "step": 8866
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.0299484729766846,
      "learning_rate": 1.1454397434679021e-05,
      "loss": 0.6028,
      "step": 8867
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.0243217945098877,
      "learning_rate": 1.1440433094624582e-05,
      "loss": 0.7492,
      "step": 8868
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.052586555480957,
      "learning_rate": 1.1426476755425642e-05,
      "loss": 0.7833,
      "step": 8869
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.004028081893921,
      "learning_rate": 1.1412528418342972e-05,
      "loss": 0.6822,
      "step": 8870
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.0427747964859009,
      "learning_rate": 1.1398588084636808e-05,
      "loss": 0.6697,
      "step": 8871
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.1805592775344849,
      "learning_rate": 1.1384655755566553e-05,
      "loss": 0.6819,
      "step": 8872
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.1632335186004639,
      "learning_rate": 1.1370731432390891e-05,
      "loss": 0.8799,
      "step": 8873
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.9442362785339355,
      "learning_rate": 1.1356815116367836e-05,
      "loss": 0.7294,
      "step": 8874
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.9278549551963806,
      "learning_rate": 1.134290680875465e-05,
      "loss": 0.753,
      "step": 8875
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.0882962942123413,
      "learning_rate": 1.132900651080785e-05,
      "loss": 0.7538,
      "step": 8876
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.1706678867340088,
      "learning_rate": 1.1315114223783241e-05,
      "loss": 0.7636,
      "step": 8877
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.1277543306350708,
      "learning_rate": 1.1301229948935955e-05,
      "loss": 0.9344,
      "step": 8878
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.0890122652053833,
      "learning_rate": 1.128735368752033e-05,
      "loss": 0.7803,
      "step": 8879
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.1197569370269775,
      "learning_rate": 1.1273485440789988e-05,
      "loss": 0.6195,
      "step": 8880
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.2603238821029663,
      "learning_rate": 1.1259625209997882e-05,
      "loss": 0.7189,
      "step": 8881
    },
    {
      "epoch": 3.33,
      "grad_norm": 1.1073569059371948,
      "learning_rate": 1.1245772996396198e-05,
      "loss": 0.842,
      "step": 8882
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.0943753719329834,
      "learning_rate": 1.123192880123638e-05,
      "loss": 0.7548,
      "step": 8883
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.1231410503387451,
      "learning_rate": 1.1218092625769205e-05,
      "loss": 0.716,
      "step": 8884
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.0403237342834473,
      "learning_rate": 1.1204264471244685e-05,
      "loss": 0.6161,
      "step": 8885
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.0028611421585083,
      "learning_rate": 1.119044433891211e-05,
      "loss": 0.6284,
      "step": 8886
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.096260666847229,
      "learning_rate": 1.1176632230020023e-05,
      "loss": 0.7704,
      "step": 8887
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.0440874099731445,
      "learning_rate": 1.1162828145816328e-05,
      "loss": 0.6694,
      "step": 8888
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.1612716913223267,
      "learning_rate": 1.1149032087548116e-05,
      "loss": 0.5851,
      "step": 8889
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.1269283294677734,
      "learning_rate": 1.1135244056461758e-05,
      "loss": 0.6953,
      "step": 8890
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.0996558666229248,
      "learning_rate": 1.1121464053802965e-05,
      "loss": 0.5868,
      "step": 8891
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.2263065576553345,
      "learning_rate": 1.1107692080816667e-05,
      "loss": 0.6976,
      "step": 8892
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.1737935543060303,
      "learning_rate": 1.1093928138747067e-05,
      "loss": 0.6385,
      "step": 8893
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.110512137413025,
      "learning_rate": 1.1080172228837682e-05,
      "loss": 0.7526,
      "step": 8894
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.1767126321792603,
      "learning_rate": 1.1066424352331285e-05,
      "loss": 0.7537,
      "step": 8895
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.092139720916748,
      "learning_rate": 1.1052684510469913e-05,
      "loss": 0.7483,
      "step": 8896
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.2095388174057007,
      "learning_rate": 1.1038952704494843e-05,
      "loss": 0.8828,
      "step": 8897
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.1247475147247314,
      "learning_rate": 1.1025228935646726e-05,
      "loss": 0.7375,
      "step": 8898
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.1491453647613525,
      "learning_rate": 1.1011513205165402e-05,
      "loss": 0.8298,
      "step": 8899
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.1012853384017944,
      "learning_rate": 1.0997805514289993e-05,
      "loss": 0.7041,
      "step": 8900
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.0686804056167603,
      "learning_rate": 1.0984105864258954e-05,
      "loss": 0.7074,
      "step": 8901
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.1544203758239746,
      "learning_rate": 1.0970414256309912e-05,
      "loss": 0.7759,
      "step": 8902
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.0651583671569824,
      "learning_rate": 1.0956730691679861e-05,
      "loss": 0.6691,
      "step": 8903
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.9959697127342224,
      "learning_rate": 1.0943055171605066e-05,
      "loss": 0.6938,
      "step": 8904
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.9243269562721252,
      "learning_rate": 1.0929387697320958e-05,
      "loss": 0.8292,
      "step": 8905
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.0533236265182495,
      "learning_rate": 1.0915728270062375e-05,
      "loss": 0.7243,
      "step": 8906
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.158239483833313,
      "learning_rate": 1.0902076891063339e-05,
      "loss": 0.7674,
      "step": 8907
    },
    {
      "epoch": 3.34,
      "grad_norm": 1.1329015493392944,
      "learning_rate": 1.0888433561557176e-05,
      "loss": 0.5877,
      "step": 8908
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.2614420652389526,
      "learning_rate": 1.087479828277651e-05,
      "loss": 0.8051,
      "step": 8909
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.9258426427841187,
      "learning_rate": 1.0861171055953202e-05,
      "loss": 0.6671,
      "step": 8910
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0364643335342407,
      "learning_rate": 1.0847551882318375e-05,
      "loss": 0.8509,
      "step": 8911
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.1250052452087402,
      "learning_rate": 1.0833940763102435e-05,
      "loss": 0.7567,
      "step": 8912
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.2252767086029053,
      "learning_rate": 1.0820337699535099e-05,
      "loss": 0.7743,
      "step": 8913
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.9477406740188599,
      "learning_rate": 1.080674269284535e-05,
      "loss": 0.5713,
      "step": 8914
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0420550107955933,
      "learning_rate": 1.0793155744261351e-05,
      "loss": 0.7066,
      "step": 8915
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.035784125328064,
      "learning_rate": 1.0779576855010665e-05,
      "loss": 0.797,
      "step": 8916
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.078521490097046,
      "learning_rate": 1.076600602632004e-05,
      "loss": 0.6794,
      "step": 8917
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.1002706289291382,
      "learning_rate": 1.0752443259415512e-05,
      "loss": 0.8392,
      "step": 8918
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0886036157608032,
      "learning_rate": 1.0738888555522442e-05,
      "loss": 0.6385,
      "step": 8919
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0478702783584595,
      "learning_rate": 1.0725341915865383e-05,
      "loss": 0.5914,
      "step": 8920
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0461649894714355,
      "learning_rate": 1.0711803341668225e-05,
      "loss": 0.6284,
      "step": 8921
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0445070266723633,
      "learning_rate": 1.0698272834154054e-05,
      "loss": 0.6722,
      "step": 8922
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.1056541204452515,
      "learning_rate": 1.068475039454534e-05,
      "loss": 0.729,
      "step": 8923
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0609325170516968,
      "learning_rate": 1.0671236024063714e-05,
      "loss": 0.7003,
      "step": 8924
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.207869291305542,
      "learning_rate": 1.0657729723930121e-05,
      "loss": 0.6888,
      "step": 8925
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.1574842929840088,
      "learning_rate": 1.0644231495364809e-05,
      "loss": 0.8159,
      "step": 8926
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0626403093338013,
      "learning_rate": 1.0630741339587257e-05,
      "loss": 0.7428,
      "step": 8927
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.994490921497345,
      "learning_rate": 1.0617259257816192e-05,
      "loss": 0.6311,
      "step": 8928
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.1082295179367065,
      "learning_rate": 1.0603785251269694e-05,
      "loss": 0.6435,
      "step": 8929
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0069910287857056,
      "learning_rate": 1.0590319321165032e-05,
      "loss": 0.6411,
      "step": 8930
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0726985931396484,
      "learning_rate": 1.057686146871879e-05,
      "loss": 0.8124,
      "step": 8931
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.1170225143432617,
      "learning_rate": 1.0563411695146785e-05,
      "loss": 0.7238,
      "step": 8932
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.177484393119812,
      "learning_rate": 1.0549970001664156e-05,
      "loss": 0.623,
      "step": 8933
    },
    {
      "epoch": 3.35,
      "grad_norm": 1.0865226984024048,
      "learning_rate": 1.0536536389485275e-05,
      "loss": 0.8288,
      "step": 8934
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.099410891532898,
      "learning_rate": 1.052311085982377e-05,
      "loss": 0.7251,
      "step": 8935
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0735993385314941,
      "learning_rate": 1.0509693413892619e-05,
      "loss": 0.6795,
      "step": 8936
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0707095861434937,
      "learning_rate": 1.0496284052903937e-05,
      "loss": 0.6455,
      "step": 8937
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0893478393554688,
      "learning_rate": 1.0482882778069225e-05,
      "loss": 0.8245,
      "step": 8938
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.9688610434532166,
      "learning_rate": 1.0469489590599257e-05,
      "loss": 0.7864,
      "step": 8939
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0062036514282227,
      "learning_rate": 1.0456104491703933e-05,
      "loss": 0.5583,
      "step": 8940
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.9604994654655457,
      "learning_rate": 1.0442727482592596e-05,
      "loss": 0.6722,
      "step": 8941
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.1830291748046875,
      "learning_rate": 1.0429358564473734e-05,
      "loss": 0.6991,
      "step": 8942
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0769587755203247,
      "learning_rate": 1.0415997738555205e-05,
      "loss": 0.7459,
      "step": 8943
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0572810173034668,
      "learning_rate": 1.0402645006044053e-05,
      "loss": 0.7157,
      "step": 8944
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.2629916667938232,
      "learning_rate": 1.0389300368146605e-05,
      "loss": 0.7532,
      "step": 8945
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.9848614931106567,
      "learning_rate": 1.0375963826068535e-05,
      "loss": 0.7855,
      "step": 8946
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0707390308380127,
      "learning_rate": 1.0362635381014641e-05,
      "loss": 0.6753,
      "step": 8947
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0985158681869507,
      "learning_rate": 1.0349315034189122e-05,
      "loss": 0.6238,
      "step": 8948
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.105000376701355,
      "learning_rate": 1.033600278679543e-05,
      "loss": 0.744,
      "step": 8949
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0651897192001343,
      "learning_rate": 1.0322698640036155e-05,
      "loss": 0.6418,
      "step": 8950
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0873266458511353,
      "learning_rate": 1.0309402595113338e-05,
      "loss": 0.8023,
      "step": 8951
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0663715600967407,
      "learning_rate": 1.0296114653228173e-05,
      "loss": 0.7891,
      "step": 8952
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0866620540618896,
      "learning_rate": 1.0282834815581122e-05,
      "loss": 0.7404,
      "step": 8953
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.1485662460327148,
      "learning_rate": 1.026956308337199e-05,
      "loss": 0.7067,
      "step": 8954
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.9453288316726685,
      "learning_rate": 1.0256299457799779e-05,
      "loss": 0.6182,
      "step": 8955
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.162636160850525,
      "learning_rate": 1.024304394006278e-05,
      "loss": 0.6877,
      "step": 8956
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.1744264364242554,
      "learning_rate": 1.0229796531358549e-05,
      "loss": 0.7317,
      "step": 8957
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.0897079706192017,
      "learning_rate": 1.0216557232883939e-05,
      "loss": 0.7147,
      "step": 8958
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.111498475074768,
      "learning_rate": 1.0203326045835027e-05,
      "loss": 0.8326,
      "step": 8959
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.054677128791809,
      "learning_rate": 1.0190102971407157e-05,
      "loss": 0.7831,
      "step": 8960
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.0102506875991821,
      "learning_rate": 1.0176888010795004e-05,
      "loss": 0.6957,
      "step": 8961
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.0889406204223633,
      "learning_rate": 1.0163681165192452e-05,
      "loss": 0.7556,
      "step": 8962
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.1253292560577393,
      "learning_rate": 1.0150482435792618e-05,
      "loss": 0.7233,
      "step": 8963
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.0858098268508911,
      "learning_rate": 1.0137291823788009e-05,
      "loss": 0.7389,
      "step": 8964
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.9969847798347473,
      "learning_rate": 1.0124109330370268e-05,
      "loss": 0.7528,
      "step": 8965
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.9610216617584229,
      "learning_rate": 1.0110934956730378e-05,
      "loss": 0.6857,
      "step": 8966
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.1233493089675903,
      "learning_rate": 1.0097768704058542e-05,
      "loss": 0.6794,
      "step": 8967
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.192576289176941,
      "learning_rate": 1.0084610573544295e-05,
      "loss": 0.737,
      "step": 8968
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.0832207202911377,
      "learning_rate": 1.0071460566376389e-05,
      "loss": 0.7136,
      "step": 8969
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.0673892498016357,
      "learning_rate": 1.005831868374283e-05,
      "loss": 0.6098,
      "step": 8970
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.1263158321380615,
      "learning_rate": 1.0045184926830942e-05,
      "loss": 0.7316,
      "step": 8971
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.027910590171814,
      "learning_rate": 1.0032059296827278e-05,
      "loss": 0.6716,
      "step": 8972
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.0554802417755127,
      "learning_rate": 1.0018941794917636e-05,
      "loss": 0.7338,
      "step": 8973
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.0510958433151245,
      "learning_rate": 1.0005832422287154e-05,
      "loss": 0.6732,
      "step": 8974
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.1189073324203491,
      "learning_rate": 9.992731180120164e-06,
      "loss": 0.7633,
      "step": 8975
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.1378381252288818,
      "learning_rate": 9.979638069600295e-06,
      "loss": 0.8338,
      "step": 8976
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.9913272857666016,
      "learning_rate": 9.966553091910414e-06,
      "loss": 0.7242,
      "step": 8977
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.0632493495941162,
      "learning_rate": 9.953476248232707e-06,
      "loss": 0.7547,
      "step": 8978
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.1216795444488525,
      "learning_rate": 9.940407539748576e-06,
      "loss": 0.7466,
      "step": 8979
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.1259766817092896,
      "learning_rate": 9.927346967638695e-06,
      "loss": 0.8134,
      "step": 8980
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.1986085176467896,
      "learning_rate": 9.914294533083068e-06,
      "loss": 0.7784,
      "step": 8981
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.0748686790466309,
      "learning_rate": 9.901250237260818e-06,
      "loss": 0.7343,
      "step": 8982
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.226327896118164,
      "learning_rate": 9.888214081350477e-06,
      "loss": 0.8733,
      "step": 8983
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.0446569919586182,
      "learning_rate": 9.875186066529818e-06,
      "loss": 0.8145,
      "step": 8984
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.138670563697815,
      "learning_rate": 9.862166193975775e-06,
      "loss": 0.7362,
      "step": 8985
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.049315333366394,
      "learning_rate": 9.849154464864685e-06,
      "loss": 0.7394,
      "step": 8986
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.139749526977539,
      "learning_rate": 9.836150880372041e-06,
      "loss": 0.7171,
      "step": 8987
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0877528190612793,
      "learning_rate": 9.823155441672648e-06,
      "loss": 0.7254,
      "step": 8988
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0988537073135376,
      "learning_rate": 9.810168149940602e-06,
      "loss": 0.7079,
      "step": 8989
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.9446301460266113,
      "learning_rate": 9.797189006349206e-06,
      "loss": 0.7491,
      "step": 8990
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.9522596597671509,
      "learning_rate": 9.784218012071056e-06,
      "loss": 0.6174,
      "step": 8991
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.9943938851356506,
      "learning_rate": 9.771255168277994e-06,
      "loss": 0.7553,
      "step": 8992
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0373053550720215,
      "learning_rate": 9.75830047614117e-06,
      "loss": 0.7582,
      "step": 8993
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.9664321541786194,
      "learning_rate": 9.745353936830948e-06,
      "loss": 0.5379,
      "step": 8994
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0809634923934937,
      "learning_rate": 9.73241555151696e-06,
      "loss": 0.6452,
      "step": 8995
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.167545199394226,
      "learning_rate": 9.71948532136816e-06,
      "loss": 0.7563,
      "step": 8996
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.9973161220550537,
      "learning_rate": 9.70656324755268e-06,
      "loss": 0.7382,
      "step": 8997
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0919146537780762,
      "learning_rate": 9.693649331237964e-06,
      "loss": 0.7322,
      "step": 8998
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.1926342248916626,
      "learning_rate": 9.680743573590733e-06,
      "loss": 0.7376,
      "step": 8999
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0450118780136108,
      "learning_rate": 9.667845975776935e-06,
      "loss": 0.6526,
      "step": 9000
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0404491424560547,
      "learning_rate": 9.65495653896179e-06,
      "loss": 0.7118,
      "step": 9001
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0648921728134155,
      "learning_rate": 9.642075264309769e-06,
      "loss": 0.6688,
      "step": 9002
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.9782212376594543,
      "learning_rate": 9.629202152984662e-06,
      "loss": 0.7224,
      "step": 9003
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.9732612371444702,
      "learning_rate": 9.616337206149462e-06,
      "loss": 0.6654,
      "step": 9004
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.9922149777412415,
      "learning_rate": 9.603480424966427e-06,
      "loss": 0.6092,
      "step": 9005
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.1201145648956299,
      "learning_rate": 9.59063181059714e-06,
      "loss": 0.7985,
      "step": 9006
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.084450602531433,
      "learning_rate": 9.57779136420236e-06,
      "loss": 0.8062,
      "step": 9007
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0946884155273438,
      "learning_rate": 9.564959086942149e-06,
      "loss": 0.8033,
      "step": 9008
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0040627717971802,
      "learning_rate": 9.552134979975857e-06,
      "loss": 0.717,
      "step": 9009
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0643573999404907,
      "learning_rate": 9.539319044462058e-06,
      "loss": 0.725,
      "step": 9010
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.0367799997329712,
      "learning_rate": 9.526511281558593e-06,
      "loss": 0.7937,
      "step": 9011
    },
    {
      "epoch": 3.38,
      "grad_norm": 1.110681176185608,
      "learning_rate": 9.513711692422555e-06,
      "loss": 0.6485,
      "step": 9012
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.1017613410949707,
      "learning_rate": 9.500920278210357e-06,
      "loss": 0.8312,
      "step": 9013
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0429904460906982,
      "learning_rate": 9.488137040077606e-06,
      "loss": 0.7265,
      "step": 9014
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.9296382069587708,
      "learning_rate": 9.475361979179176e-06,
      "loss": 0.6809,
      "step": 9015
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0752742290496826,
      "learning_rate": 9.462595096669291e-06,
      "loss": 0.6391,
      "step": 9016
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0763835906982422,
      "learning_rate": 9.449836393701262e-06,
      "loss": 0.7939,
      "step": 9017
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.069398283958435,
      "learning_rate": 9.43708587142783e-06,
      "loss": 0.6665,
      "step": 9018
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0707080364227295,
      "learning_rate": 9.424343531000968e-06,
      "loss": 0.7012,
      "step": 9019
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.9251408576965332,
      "learning_rate": 9.411609373571783e-06,
      "loss": 0.7034,
      "step": 9020
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.07462739944458,
      "learning_rate": 9.398883400290804e-06,
      "loss": 0.6953,
      "step": 9021
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0360369682312012,
      "learning_rate": 9.386165612307718e-06,
      "loss": 0.7249,
      "step": 9022
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0224446058273315,
      "learning_rate": 9.373456010771509e-06,
      "loss": 0.7643,
      "step": 9023
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.9468536972999573,
      "learning_rate": 9.360754596830434e-06,
      "loss": 0.5593,
      "step": 9024
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.9730035066604614,
      "learning_rate": 9.348061371631978e-06,
      "loss": 0.7707,
      "step": 9025
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0437734127044678,
      "learning_rate": 9.33537633632292e-06,
      "loss": 0.6686,
      "step": 9026
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.970960795879364,
      "learning_rate": 9.322699492049247e-06,
      "loss": 0.6617,
      "step": 9027
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.118283987045288,
      "learning_rate": 9.310030839956274e-06,
      "loss": 0.7296,
      "step": 9028
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0573726892471313,
      "learning_rate": 9.297370381188531e-06,
      "loss": 0.6997,
      "step": 9029
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.089155673980713,
      "learning_rate": 9.284718116889802e-06,
      "loss": 0.7419,
      "step": 9030
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0954445600509644,
      "learning_rate": 9.272074048203182e-06,
      "loss": 0.6106,
      "step": 9031
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0231753587722778,
      "learning_rate": 9.259438176270962e-06,
      "loss": 0.8149,
      "step": 9032
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.1492067575454712,
      "learning_rate": 9.246810502234727e-06,
      "loss": 0.6538,
      "step": 9033
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0866496562957764,
      "learning_rate": 9.234191027235329e-06,
      "loss": 0.6536,
      "step": 9034
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.9750955700874329,
      "learning_rate": 9.221579752412856e-06,
      "loss": 0.693,
      "step": 9035
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0841495990753174,
      "learning_rate": 9.208976678906667e-06,
      "loss": 0.7087,
      "step": 9036
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.051024079322815,
      "learning_rate": 9.196381807855358e-06,
      "loss": 0.696,
      "step": 9037
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.0028351545333862,
      "learning_rate": 9.183795140396856e-06,
      "loss": 0.733,
      "step": 9038
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.1184941530227661,
      "learning_rate": 9.171216677668249e-06,
      "loss": 0.774,
      "step": 9039
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.9997143745422363,
      "learning_rate": 9.158646420805927e-06,
      "loss": 0.763,
      "step": 9040
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.2362152338027954,
      "learning_rate": 9.146084370945584e-06,
      "loss": 0.7059,
      "step": 9041
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.0616488456726074,
      "learning_rate": 9.13353052922209e-06,
      "loss": 0.714,
      "step": 9042
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.0774118900299072,
      "learning_rate": 9.120984896769625e-06,
      "loss": 0.806,
      "step": 9043
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.0442042350769043,
      "learning_rate": 9.108447474721627e-06,
      "loss": 0.7788,
      "step": 9044
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.190020203590393,
      "learning_rate": 9.095918264210779e-06,
      "loss": 0.8158,
      "step": 9045
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.0671321153640747,
      "learning_rate": 9.083397266369021e-06,
      "loss": 0.7236,
      "step": 9046
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.0428060293197632,
      "learning_rate": 9.070884482327524e-06,
      "loss": 0.6897,
      "step": 9047
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.027012586593628,
      "learning_rate": 9.058379913216807e-06,
      "loss": 0.7463,
      "step": 9048
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.9722789525985718,
      "learning_rate": 9.045883560166547e-06,
      "loss": 0.7299,
      "step": 9049
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.1363407373428345,
      "learning_rate": 9.033395424305702e-06,
      "loss": 0.7543,
      "step": 9050
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.9568233489990234,
      "learning_rate": 9.020915506762573e-06,
      "loss": 0.5894,
      "step": 9051
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.1371999979019165,
      "learning_rate": 9.008443808664568e-06,
      "loss": 0.6266,
      "step": 9052
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.9009671807289124,
      "learning_rate": 8.995980331138464e-06,
      "loss": 0.5027,
      "step": 9053
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.2190792560577393,
      "learning_rate": 8.983525075310317e-06,
      "loss": 0.844,
      "step": 9054
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.1420608758926392,
      "learning_rate": 8.971078042305314e-06,
      "loss": 0.7302,
      "step": 9055
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.109871745109558,
      "learning_rate": 8.958639233248023e-06,
      "loss": 0.7247,
      "step": 9056
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.1063612699508667,
      "learning_rate": 8.946208649262188e-06,
      "loss": 0.5798,
      "step": 9057
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.9917809963226318,
      "learning_rate": 8.933786291470858e-06,
      "loss": 0.6859,
      "step": 9058
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.1049385070800781,
      "learning_rate": 8.921372160996322e-06,
      "loss": 0.7067,
      "step": 9059
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.9979867339134216,
      "learning_rate": 8.908966258960138e-06,
      "loss": 0.7209,
      "step": 9060
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.097686529159546,
      "learning_rate": 8.89656858648309e-06,
      "loss": 0.7757,
      "step": 9061
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.2399580478668213,
      "learning_rate": 8.884179144685223e-06,
      "loss": 0.7626,
      "step": 9062
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.136938214302063,
      "learning_rate": 8.871797934685877e-06,
      "loss": 0.7034,
      "step": 9063
    },
    {
      "epoch": 3.4,
      "grad_norm": 1.1193691492080688,
      "learning_rate": 8.859424957603669e-06,
      "loss": 0.7238,
      "step": 9064
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.9170823693275452,
      "learning_rate": 8.847060214556325e-06,
      "loss": 0.6198,
      "step": 9065
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.177687168121338,
      "learning_rate": 8.834703706661008e-06,
      "loss": 0.8217,
      "step": 9066
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1676783561706543,
      "learning_rate": 8.822355435034047e-06,
      "loss": 0.8495,
      "step": 9067
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.0966275930404663,
      "learning_rate": 8.810015400790994e-06,
      "loss": 0.7493,
      "step": 9068
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1200814247131348,
      "learning_rate": 8.797683605046746e-06,
      "loss": 0.6732,
      "step": 9069
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1071935892105103,
      "learning_rate": 8.785360048915414e-06,
      "loss": 0.6973,
      "step": 9070
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.0973236560821533,
      "learning_rate": 8.773044733510338e-06,
      "loss": 0.6512,
      "step": 9071
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9079304337501526,
      "learning_rate": 8.760737659944129e-06,
      "loss": 0.6875,
      "step": 9072
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9435780048370361,
      "learning_rate": 8.748438829328686e-06,
      "loss": 0.5858,
      "step": 9073
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9567921161651611,
      "learning_rate": 8.736148242775144e-06,
      "loss": 0.7528,
      "step": 9074
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1331819295883179,
      "learning_rate": 8.723865901393847e-06,
      "loss": 0.7752,
      "step": 9075
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1733626127243042,
      "learning_rate": 8.711591806294483e-06,
      "loss": 0.6909,
      "step": 9076
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.0387272834777832,
      "learning_rate": 8.699325958585924e-06,
      "loss": 0.5803,
      "step": 9077
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9772284030914307,
      "learning_rate": 8.687068359376304e-06,
      "loss": 0.6603,
      "step": 9078
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1691044569015503,
      "learning_rate": 8.67481900977306e-06,
      "loss": 0.6968,
      "step": 9079
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.097934603691101,
      "learning_rate": 8.66257791088284e-06,
      "loss": 0.9158,
      "step": 9080
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1539489030838013,
      "learning_rate": 8.65034506381155e-06,
      "loss": 0.8023,
      "step": 9081
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1339150667190552,
      "learning_rate": 8.638120469664358e-06,
      "loss": 0.6782,
      "step": 9082
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1222469806671143,
      "learning_rate": 8.625904129545692e-06,
      "loss": 0.8694,
      "step": 9083
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9882358312606812,
      "learning_rate": 8.613696044559238e-06,
      "loss": 0.6959,
      "step": 9084
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1528429985046387,
      "learning_rate": 8.601496215807903e-06,
      "loss": 0.6603,
      "step": 9085
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.1927505731582642,
      "learning_rate": 8.589304644393925e-06,
      "loss": 0.5971,
      "step": 9086
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.076124906539917,
      "learning_rate": 8.57712133141867e-06,
      "loss": 0.8423,
      "step": 9087
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.0142133235931396,
      "learning_rate": 8.564946277982867e-06,
      "loss": 0.7821,
      "step": 9088
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9988782405853271,
      "learning_rate": 8.552779485186491e-06,
      "loss": 0.687,
      "step": 9089
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.0657709836959839,
      "learning_rate": 8.540620954128687e-06,
      "loss": 0.7444,
      "step": 9090
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9617489576339722,
      "learning_rate": 8.528470685907963e-06,
      "loss": 0.469,
      "step": 9091
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.110254168510437,
      "learning_rate": 8.516328681621988e-06,
      "loss": 0.7892,
      "step": 9092
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0816590785980225,
      "learning_rate": 8.504194942367726e-06,
      "loss": 0.7353,
      "step": 9093
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0882283449172974,
      "learning_rate": 8.492069469241426e-06,
      "loss": 0.7148,
      "step": 9094
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0106210708618164,
      "learning_rate": 8.479952263338509e-06,
      "loss": 0.5748,
      "step": 9095
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0457617044448853,
      "learning_rate": 8.467843325753755e-06,
      "loss": 0.7921,
      "step": 9096
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.117877721786499,
      "learning_rate": 8.455742657581067e-06,
      "loss": 0.6879,
      "step": 9097
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0676625967025757,
      "learning_rate": 8.443650259913715e-06,
      "loss": 0.8962,
      "step": 9098
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0059471130371094,
      "learning_rate": 8.431566133844204e-06,
      "loss": 0.7191,
      "step": 9099
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0120693445205688,
      "learning_rate": 8.419490280464193e-06,
      "loss": 0.7758,
      "step": 9100
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.204472303390503,
      "learning_rate": 8.40742270086473e-06,
      "loss": 0.7526,
      "step": 9101
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.062674641609192,
      "learning_rate": 8.395363396136025e-06,
      "loss": 0.81,
      "step": 9102
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0124889612197876,
      "learning_rate": 8.383312367367557e-06,
      "loss": 0.7251,
      "step": 9103
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.032525897026062,
      "learning_rate": 8.371269615648102e-06,
      "loss": 0.7441,
      "step": 9104
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.308310866355896,
      "learning_rate": 8.359235142065647e-06,
      "loss": 0.795,
      "step": 9105
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0728706121444702,
      "learning_rate": 8.34720894770742e-06,
      "loss": 0.7395,
      "step": 9106
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0160733461380005,
      "learning_rate": 8.335191033659907e-06,
      "loss": 0.8001,
      "step": 9107
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.1072288751602173,
      "learning_rate": 8.323181401008896e-06,
      "loss": 0.6637,
      "step": 9108
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.1431891918182373,
      "learning_rate": 8.311180050839385e-06,
      "loss": 0.6184,
      "step": 9109
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0665948390960693,
      "learning_rate": 8.299186984235585e-06,
      "loss": 0.7739,
      "step": 9110
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.9808893203735352,
      "learning_rate": 8.287202202281053e-06,
      "loss": 0.725,
      "step": 9111
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0655635595321655,
      "learning_rate": 8.275225706058532e-06,
      "loss": 0.6452,
      "step": 9112
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0207642316818237,
      "learning_rate": 8.263257496650001e-06,
      "loss": 0.6529,
      "step": 9113
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.2492492198944092,
      "learning_rate": 8.251297575136762e-06,
      "loss": 0.8524,
      "step": 9114
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.9219918251037598,
      "learning_rate": 8.239345942599296e-06,
      "loss": 0.643,
      "step": 9115
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.9330912828445435,
      "learning_rate": 8.227402600117396e-06,
      "loss": 0.6616,
      "step": 9116
    },
    {
      "epoch": 3.42,
      "grad_norm": 1.0108319520950317,
      "learning_rate": 8.215467548770028e-06,
      "loss": 0.6017,
      "step": 9117
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.1926735639572144,
      "learning_rate": 8.203540789635499e-06,
      "loss": 0.6724,
      "step": 9118
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.1337329149246216,
      "learning_rate": 8.191622323791315e-06,
      "loss": 0.65,
      "step": 9119
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.0990036725997925,
      "learning_rate": 8.179712152314224e-06,
      "loss": 0.6849,
      "step": 9120
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.053456425666809,
      "learning_rate": 8.167810276280286e-06,
      "loss": 0.7156,
      "step": 9121
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.0267279148101807,
      "learning_rate": 8.155916696764698e-06,
      "loss": 0.7274,
      "step": 9122
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.1953610181808472,
      "learning_rate": 8.144031414842013e-06,
      "loss": 0.7368,
      "step": 9123
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.0597689151763916,
      "learning_rate": 8.132154431586025e-06,
      "loss": 0.7286,
      "step": 9124
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.9678187370300293,
      "learning_rate": 8.120285748069734e-06,
      "loss": 0.7641,
      "step": 9125
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.127488613128662,
      "learning_rate": 8.108425365365402e-06,
      "loss": 0.6565,
      "step": 9126
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.0647679567337036,
      "learning_rate": 8.09657328454453e-06,
      "loss": 0.7675,
      "step": 9127
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.0030814409255981,
      "learning_rate": 8.084729506677924e-06,
      "loss": 0.565,
      "step": 9128
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.2076430320739746,
      "learning_rate": 8.072894032835587e-06,
      "loss": 0.6174,
      "step": 9129
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.0630863904953003,
      "learning_rate": 8.061066864086763e-06,
      "loss": 0.8243,
      "step": 9130
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.145521640777588,
      "learning_rate": 8.049248001500021e-06,
      "loss": 0.6198,
      "step": 9131
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.0241557359695435,
      "learning_rate": 8.03743744614307e-06,
      "loss": 0.8411,
      "step": 9132
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.052338719367981,
      "learning_rate": 8.025635199082959e-06,
      "loss": 0.6876,
      "step": 9133
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.2052361965179443,
      "learning_rate": 8.013841261385978e-06,
      "loss": 0.7624,
      "step": 9134
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.1443496942520142,
      "learning_rate": 8.002055634117578e-06,
      "loss": 0.8037,
      "step": 9135
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.1397665739059448,
      "learning_rate": 7.990278318342592e-06,
      "loss": 0.6938,
      "step": 9136
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.0048373937606812,
      "learning_rate": 7.978509315124982e-06,
      "loss": 0.5797,
      "step": 9137
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.1265645027160645,
      "learning_rate": 7.966748625528032e-06,
      "loss": 0.7616,
      "step": 9138
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.9839915037155151,
      "learning_rate": 7.954996250614255e-06,
      "loss": 0.8625,
      "step": 9139
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.1420049667358398,
      "learning_rate": 7.943252191445415e-06,
      "loss": 0.6819,
      "step": 9140
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.064190149307251,
      "learning_rate": 7.93151644908252e-06,
      "loss": 0.8381,
      "step": 9141
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.0163352489471436,
      "learning_rate": 7.91978902458581e-06,
      "loss": 0.6756,
      "step": 9142
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.2160987854003906,
      "learning_rate": 7.908069919014815e-06,
      "loss": 0.7625,
      "step": 9143
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.9988078474998474,
      "learning_rate": 7.896359133428278e-06,
      "loss": 0.5222,
      "step": 9144
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.1544286012649536,
      "learning_rate": 7.884656668884194e-06,
      "loss": 0.7392,
      "step": 9145
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0106085538864136,
      "learning_rate": 7.872962526439832e-06,
      "loss": 0.5909,
      "step": 9146
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0879319906234741,
      "learning_rate": 7.861276707151688e-06,
      "loss": 0.7571,
      "step": 9147
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0595837831497192,
      "learning_rate": 7.849599212075487e-06,
      "loss": 0.7661,
      "step": 9148
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.9230301380157471,
      "learning_rate": 7.837930042266262e-06,
      "loss": 0.716,
      "step": 9149
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.973720908164978,
      "learning_rate": 7.826269198778236e-06,
      "loss": 0.6282,
      "step": 9150
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.007436990737915,
      "learning_rate": 7.8146166826649e-06,
      "loss": 0.6932,
      "step": 9151
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.9404435753822327,
      "learning_rate": 7.802972494978988e-06,
      "loss": 0.6788,
      "step": 9152
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.1977880001068115,
      "learning_rate": 7.791336636772495e-06,
      "loss": 0.7648,
      "step": 9153
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0651112794876099,
      "learning_rate": 7.779709109096667e-06,
      "loss": 0.5129,
      "step": 9154
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.1280920505523682,
      "learning_rate": 7.768089913001941e-06,
      "loss": 0.5622,
      "step": 9155
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0237464904785156,
      "learning_rate": 7.7564790495381e-06,
      "loss": 0.6444,
      "step": 9156
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.9965409636497498,
      "learning_rate": 7.744876519754096e-06,
      "loss": 0.5191,
      "step": 9157
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0866248607635498,
      "learning_rate": 7.733282324698132e-06,
      "loss": 0.7608,
      "step": 9158
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.151962161064148,
      "learning_rate": 7.721696465417716e-06,
      "loss": 0.6955,
      "step": 9159
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.035443902015686,
      "learning_rate": 7.710118942959544e-06,
      "loss": 0.6529,
      "step": 9160
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0554676055908203,
      "learning_rate": 7.698549758369577e-06,
      "loss": 0.6236,
      "step": 9161
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0494537353515625,
      "learning_rate": 7.686988912693017e-06,
      "loss": 0.8637,
      "step": 9162
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0948110818862915,
      "learning_rate": 7.675436406974357e-06,
      "loss": 0.6915,
      "step": 9163
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.014312505722046,
      "learning_rate": 7.663892242257265e-06,
      "loss": 0.7038,
      "step": 9164
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.105259895324707,
      "learning_rate": 7.652356419584693e-06,
      "loss": 0.7668,
      "step": 9165
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.2151336669921875,
      "learning_rate": 7.64082893999889e-06,
      "loss": 0.7624,
      "step": 9166
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.3033804893493652,
      "learning_rate": 7.629309804541207e-06,
      "loss": 0.803,
      "step": 9167
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.0824614763259888,
      "learning_rate": 7.6177990142523914e-06,
      "loss": 0.8372,
      "step": 9168
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.036757469177246,
      "learning_rate": 7.6062965701723995e-06,
      "loss": 0.734,
      "step": 9169
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.9475414752960205,
      "learning_rate": 7.594802473340346e-06,
      "loss": 0.4939,
      "step": 9170
    },
    {
      "epoch": 3.44,
      "eval_loss": 1.0303876399993896,
      "eval_runtime": 17.7822,
      "eval_samples_per_second": 153.468,
      "eval_steps_per_second": 38.409,
      "step": 9170
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.9720313549041748,
      "learning_rate": 7.583316724794687e-06,
      "loss": 0.7098,
      "step": 9171
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.1041759252548218,
      "learning_rate": 7.571839325573149e-06,
      "loss": 0.7437,
      "step": 9172
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.1109431982040405,
      "learning_rate": 7.560370276712559e-06,
      "loss": 0.6277,
      "step": 9173
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0375548601150513,
      "learning_rate": 7.54890957924913e-06,
      "loss": 0.7086,
      "step": 9174
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.9868291616439819,
      "learning_rate": 7.53745723421827e-06,
      "loss": 0.8026,
      "step": 9175
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0683125257492065,
      "learning_rate": 7.526013242654617e-06,
      "loss": 0.7178,
      "step": 9176
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0639675855636597,
      "learning_rate": 7.514577605592099e-06,
      "loss": 0.6656,
      "step": 9177
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.2061840295791626,
      "learning_rate": 7.503150324063834e-06,
      "loss": 0.5706,
      "step": 9178
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.1063809394836426,
      "learning_rate": 7.491731399102231e-06,
      "loss": 0.7733,
      "step": 9179
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0708444118499756,
      "learning_rate": 7.480320831738885e-06,
      "loss": 0.9909,
      "step": 9180
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.9723497629165649,
      "learning_rate": 7.468918623004739e-06,
      "loss": 0.6657,
      "step": 9181
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.1326855421066284,
      "learning_rate": 7.4575247739298695e-06,
      "loss": 0.6794,
      "step": 9182
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0580499172210693,
      "learning_rate": 7.446139285543663e-06,
      "loss": 0.6849,
      "step": 9183
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.1659971475601196,
      "learning_rate": 7.434762158874742e-06,
      "loss": 0.6884,
      "step": 9184
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.9843112826347351,
      "learning_rate": 7.423393394950951e-06,
      "loss": 0.5587,
      "step": 9185
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0195828676223755,
      "learning_rate": 7.4120329947993894e-06,
      "loss": 0.7682,
      "step": 9186
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.2286359071731567,
      "learning_rate": 7.400680959446437e-06,
      "loss": 0.802,
      "step": 9187
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.1398859024047852,
      "learning_rate": 7.389337289917652e-06,
      "loss": 0.6445,
      "step": 9188
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.9962741732597351,
      "learning_rate": 7.378001987237893e-06,
      "loss": 0.6114,
      "step": 9189
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0173625946044922,
      "learning_rate": 7.366675052431205e-06,
      "loss": 0.8164,
      "step": 9190
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.1331255435943604,
      "learning_rate": 7.355356486520959e-06,
      "loss": 0.8575,
      "step": 9191
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0390856266021729,
      "learning_rate": 7.344046290529705e-06,
      "loss": 0.7736,
      "step": 9192
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0908116102218628,
      "learning_rate": 7.332744465479235e-06,
      "loss": 0.6188,
      "step": 9193
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0369235277175903,
      "learning_rate": 7.3214510123906345e-06,
      "loss": 0.6624,
      "step": 9194
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.0392862558364868,
      "learning_rate": 7.310165932284185e-06,
      "loss": 0.6258,
      "step": 9195
    },
    {
      "epoch": 3.45,
      "grad_norm": 1.1555900573730469,
      "learning_rate": 7.298889226179428e-06,
      "loss": 0.5147,
      "step": 9196
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.1685446500778198,
      "learning_rate": 7.2876208950951705e-06,
      "loss": 0.7301,
      "step": 9197
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.0557365417480469,
      "learning_rate": 7.276360940049432e-06,
      "loss": 0.8374,
      "step": 9198
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.1354906558990479,
      "learning_rate": 7.265109362059474e-06,
      "loss": 0.7082,
      "step": 9199
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.2236908674240112,
      "learning_rate": 7.253866162141809e-06,
      "loss": 0.7319,
      "step": 9200
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.0137429237365723,
      "learning_rate": 7.242631341312234e-06,
      "loss": 0.5561,
      "step": 9201
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.9280979633331299,
      "learning_rate": 7.231404900585714e-06,
      "loss": 0.707,
      "step": 9202
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.088822841644287,
      "learning_rate": 7.220186840976495e-06,
      "loss": 0.7325,
      "step": 9203
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.1555918455123901,
      "learning_rate": 7.208977163498121e-06,
      "loss": 0.872,
      "step": 9204
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.9516086578369141,
      "learning_rate": 7.197775869163226e-06,
      "loss": 0.7393,
      "step": 9205
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.0761804580688477,
      "learning_rate": 7.186582958983856e-06,
      "loss": 0.6505,
      "step": 9206
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.189380407333374,
      "learning_rate": 7.1753984339712254e-06,
      "loss": 0.7743,
      "step": 9207
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.2366197109222412,
      "learning_rate": 7.164222295135747e-06,
      "loss": 0.7277,
      "step": 9208
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.9702621102333069,
      "learning_rate": 7.1530545434871694e-06,
      "loss": 0.7271,
      "step": 9209
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.0080208778381348,
      "learning_rate": 7.141895180034408e-06,
      "loss": 0.7209,
      "step": 9210
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.0710352659225464,
      "learning_rate": 7.130744205785645e-06,
      "loss": 0.7182,
      "step": 9211
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.9884374737739563,
      "learning_rate": 7.1196016217483395e-06,
      "loss": 0.6439,
      "step": 9212
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.276144027709961,
      "learning_rate": 7.108467428929133e-06,
      "loss": 0.825,
      "step": 9213
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.0045021772384644,
      "learning_rate": 7.097341628333942e-06,
      "loss": 0.7838,
      "step": 9214
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.217192530632019,
      "learning_rate": 7.086224220967907e-06,
      "loss": 0.9372,
      "step": 9215
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.9985117316246033,
      "learning_rate": 7.075115207835436e-06,
      "loss": 0.7238,
      "step": 9216
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.0279706716537476,
      "learning_rate": 7.064014589940205e-06,
      "loss": 0.7444,
      "step": 9217
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.1445443630218506,
      "learning_rate": 7.0529223682850195e-06,
      "loss": 0.7798,
      "step": 9218
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.0769742727279663,
      "learning_rate": 7.041838543872048e-06,
      "loss": 0.8303,
      "step": 9219
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.067243218421936,
      "learning_rate": 7.030763117702632e-06,
      "loss": 0.6176,
      "step": 9220
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.1508725881576538,
      "learning_rate": 7.019696090777372e-06,
      "loss": 0.6969,
      "step": 9221
    },
    {
      "epoch": 3.46,
      "grad_norm": 1.2014031410217285,
      "learning_rate": 7.0086374640961235e-06,
      "loss": 0.6665,
      "step": 9222
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.173203945159912,
      "learning_rate": 6.997587238657976e-06,
      "loss": 0.8082,
      "step": 9223
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.100083351135254,
      "learning_rate": 6.986545415461232e-06,
      "loss": 0.6894,
      "step": 9224
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.200015902519226,
      "learning_rate": 6.9755119955034586e-06,
      "loss": 0.841,
      "step": 9225
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.0138157606124878,
      "learning_rate": 6.964486979781481e-06,
      "loss": 0.654,
      "step": 9226
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.120304822921753,
      "learning_rate": 6.953470369291348e-06,
      "loss": 0.7461,
      "step": 9227
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.1081653833389282,
      "learning_rate": 6.942462165028319e-06,
      "loss": 0.7204,
      "step": 9228
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.1913350820541382,
      "learning_rate": 6.931462367986963e-06,
      "loss": 0.7455,
      "step": 9229
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.208710789680481,
      "learning_rate": 6.92047097916102e-06,
      "loss": 0.7678,
      "step": 9230
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.9739933013916016,
      "learning_rate": 6.909487999543507e-06,
      "loss": 0.7025,
      "step": 9231
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.0176446437835693,
      "learning_rate": 6.8985134301266855e-06,
      "loss": 0.7432,
      "step": 9232
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.2036703824996948,
      "learning_rate": 6.887547271902028e-06,
      "loss": 0.7382,
      "step": 9233
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.0924941301345825,
      "learning_rate": 6.8765895258602775e-06,
      "loss": 0.6628,
      "step": 9234
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.2871224880218506,
      "learning_rate": 6.865640192991396e-06,
      "loss": 0.7473,
      "step": 9235
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.2030786275863647,
      "learning_rate": 6.854699274284604e-06,
      "loss": 0.8055,
      "step": 9236
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.0435148477554321,
      "learning_rate": 6.843766770728344e-06,
      "loss": 0.7143,
      "step": 9237
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.9861230850219727,
      "learning_rate": 6.832842683310292e-06,
      "loss": 0.7007,
      "step": 9238
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.125016450881958,
      "learning_rate": 6.821927013017426e-06,
      "loss": 0.7209,
      "step": 9239
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.08424973487854,
      "learning_rate": 6.811019760835846e-06,
      "loss": 0.7149,
      "step": 9240
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.0304640531539917,
      "learning_rate": 6.800120927751008e-06,
      "loss": 0.7551,
      "step": 9241
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.0916084051132202,
      "learning_rate": 6.789230514747569e-06,
      "loss": 0.7426,
      "step": 9242
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.9221816658973694,
      "learning_rate": 6.778348522809363e-06,
      "loss": 0.6587,
      "step": 9243
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.05246901512146,
      "learning_rate": 6.767474952919562e-06,
      "loss": 0.8001,
      "step": 9244
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.154273271560669,
      "learning_rate": 6.7566098060605095e-06,
      "loss": 0.8169,
      "step": 9245
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.201464056968689,
      "learning_rate": 6.745753083213835e-06,
      "loss": 0.756,
      "step": 9246
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.9706142544746399,
      "learning_rate": 6.734904785360363e-06,
      "loss": 0.689,
      "step": 9247
    },
    {
      "epoch": 3.47,
      "grad_norm": 1.1951816082000732,
      "learning_rate": 6.724064913480155e-06,
      "loss": 0.8248,
      "step": 9248
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.980755627155304,
      "learning_rate": 6.7132334685525935e-06,
      "loss": 0.6596,
      "step": 9249
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.0239224433898926,
      "learning_rate": 6.702410451556162e-06,
      "loss": 0.7235,
      "step": 9250
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.039171576499939,
      "learning_rate": 6.691595863468703e-06,
      "loss": 0.6682,
      "step": 9251
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.2829086780548096,
      "learning_rate": 6.680789705267276e-06,
      "loss": 0.762,
      "step": 9252
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.9197306632995605,
      "learning_rate": 6.669991977928103e-06,
      "loss": 0.6945,
      "step": 9253
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.9960047602653503,
      "learning_rate": 6.659202682426735e-06,
      "loss": 0.7764,
      "step": 9254
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.1266659498214722,
      "learning_rate": 6.648421819737905e-06,
      "loss": 0.8175,
      "step": 9255
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.1425597667694092,
      "learning_rate": 6.637649390835587e-06,
      "loss": 0.6931,
      "step": 9256
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.2252631187438965,
      "learning_rate": 6.626885396693061e-06,
      "loss": 0.6982,
      "step": 9257
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.0129222869873047,
      "learning_rate": 6.616129838282747e-06,
      "loss": 0.6916,
      "step": 9258
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.1560051441192627,
      "learning_rate": 6.60538271657638e-06,
      "loss": 0.7184,
      "step": 9259
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.0080316066741943,
      "learning_rate": 6.594644032544861e-06,
      "loss": 0.6439,
      "step": 9260
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.9577999114990234,
      "learning_rate": 6.583913787158402e-06,
      "loss": 0.7313,
      "step": 9261
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.0559805631637573,
      "learning_rate": 6.573191981386429e-06,
      "loss": 0.6372,
      "step": 9262
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.1412889957427979,
      "learning_rate": 6.562478616197554e-06,
      "loss": 0.7778,
      "step": 9263
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.9646375179290771,
      "learning_rate": 6.551773692559704e-06,
      "loss": 0.6954,
      "step": 9264
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.085612416267395,
      "learning_rate": 6.541077211440005e-06,
      "loss": 0.7333,
      "step": 9265
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.117197871208191,
      "learning_rate": 6.530389173804807e-06,
      "loss": 0.7139,
      "step": 9266
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.1950520277023315,
      "learning_rate": 6.519709580619726e-06,
      "loss": 0.7616,
      "step": 9267
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.1342331171035767,
      "learning_rate": 6.5090384328496125e-06,
      "loss": 0.8493,
      "step": 9268
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.9516593813896179,
      "learning_rate": 6.498375731458528e-06,
      "loss": 0.7098,
      "step": 9269
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.146816611289978,
      "learning_rate": 6.48772147740977e-06,
      "loss": 0.8917,
      "step": 9270
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.9969915151596069,
      "learning_rate": 6.477075671665933e-06,
      "loss": 0.6899,
      "step": 9271
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.1010537147521973,
      "learning_rate": 6.466438315188783e-06,
      "loss": 0.7704,
      "step": 9272
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.038866400718689,
      "learning_rate": 6.4558094089393285e-06,
      "loss": 0.8048,
      "step": 9273
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.0658985376358032,
      "learning_rate": 6.44518895387789e-06,
      "loss": 0.7212,
      "step": 9274
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.9857911467552185,
      "learning_rate": 6.4345769509638776e-06,
      "loss": 0.6485,
      "step": 9275
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.0056519508361816,
      "learning_rate": 6.4239734011560804e-06,
      "loss": 0.6198,
      "step": 9276
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.9862452745437622,
      "learning_rate": 6.413378305412465e-06,
      "loss": 0.7546,
      "step": 9277
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.230545997619629,
      "learning_rate": 6.402791664690244e-06,
      "loss": 0.7424,
      "step": 9278
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.998950719833374,
      "learning_rate": 6.392213479945852e-06,
      "loss": 0.6881,
      "step": 9279
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.0450317859649658,
      "learning_rate": 6.381643752134947e-06,
      "loss": 0.6857,
      "step": 9280
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.0637646913528442,
      "learning_rate": 6.371082482212476e-06,
      "loss": 0.6791,
      "step": 9281
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.9225853085517883,
      "learning_rate": 6.360529671132576e-06,
      "loss": 0.6718,
      "step": 9282
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.0557392835617065,
      "learning_rate": 6.349985319848628e-06,
      "loss": 0.7923,
      "step": 9283
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.1287504434585571,
      "learning_rate": 6.339449429313282e-06,
      "loss": 0.8728,
      "step": 9284
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.0938795804977417,
      "learning_rate": 6.328922000478354e-06,
      "loss": 0.8232,
      "step": 9285
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.9448834657669067,
      "learning_rate": 6.3184030342949506e-06,
      "loss": 0.6175,
      "step": 9286
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.2612577676773071,
      "learning_rate": 6.307892531713444e-06,
      "loss": 0.8046,
      "step": 9287
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.9884944558143616,
      "learning_rate": 6.2973904936833305e-06,
      "loss": 0.8525,
      "step": 9288
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.010074257850647,
      "learning_rate": 6.286896921153462e-06,
      "loss": 0.7999,
      "step": 9289
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.10933518409729,
      "learning_rate": 6.276411815071859e-06,
      "loss": 0.773,
      "step": 9290
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.9863457679748535,
      "learning_rate": 6.265935176385762e-06,
      "loss": 0.8442,
      "step": 9291
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.1050808429718018,
      "learning_rate": 6.255467006041715e-06,
      "loss": 0.7108,
      "step": 9292
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.132288932800293,
      "learning_rate": 6.2450073049854506e-06,
      "loss": 0.7992,
      "step": 9293
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.938675045967102,
      "learning_rate": 6.234556074161935e-06,
      "loss": 0.7356,
      "step": 9294
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.0119656324386597,
      "learning_rate": 6.224113314515367e-06,
      "loss": 0.5611,
      "step": 9295
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.070189356803894,
      "learning_rate": 6.213679026989216e-06,
      "loss": 0.6442,
      "step": 9296
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.2044641971588135,
      "learning_rate": 6.20325321252615e-06,
      "loss": 0.8608,
      "step": 9297
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.2190556526184082,
      "learning_rate": 6.192835872068059e-06,
      "loss": 0.6217,
      "step": 9298
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.9709049463272095,
      "learning_rate": 6.182427006556135e-06,
      "loss": 0.734,
      "step": 9299
    },
    {
      "epoch": 3.49,
      "grad_norm": 1.067549705505371,
      "learning_rate": 6.172026616930737e-06,
      "loss": 0.7862,
      "step": 9300
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.075640320777893,
      "learning_rate": 6.161634704131458e-06,
      "loss": 0.6628,
      "step": 9301
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.0709084272384644,
      "learning_rate": 6.151251269097202e-06,
      "loss": 0.8995,
      "step": 9302
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.0197151899337769,
      "learning_rate": 6.140876312766009e-06,
      "loss": 0.7361,
      "step": 9303
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.1653517484664917,
      "learning_rate": 6.130509836075215e-06,
      "loss": 0.6699,
      "step": 9304
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.054538607597351,
      "learning_rate": 6.1201518399613635e-06,
      "loss": 0.6776,
      "step": 9305
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.036051869392395,
      "learning_rate": 6.1098023253602586e-06,
      "loss": 0.718,
      "step": 9306
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.008304238319397,
      "learning_rate": 6.099461293206899e-06,
      "loss": 0.5714,
      "step": 9307
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.0663912296295166,
      "learning_rate": 6.089128744435546e-06,
      "loss": 0.8644,
      "step": 9308
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.1960687637329102,
      "learning_rate": 6.078804679979699e-06,
      "loss": 0.745,
      "step": 9309
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.1525475978851318,
      "learning_rate": 6.068489100772068e-06,
      "loss": 0.7509,
      "step": 9310
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.1699302196502686,
      "learning_rate": 6.058182007744584e-06,
      "loss": 0.6619,
      "step": 9311
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.1139271259307861,
      "learning_rate": 6.04788340182848e-06,
      "loss": 0.9148,
      "step": 9312
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.0354089736938477,
      "learning_rate": 6.037593283954168e-06,
      "loss": 0.6655,
      "step": 9313
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.113547444343567,
      "learning_rate": 6.027311655051271e-06,
      "loss": 0.9865,
      "step": 9314
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.1232999563217163,
      "learning_rate": 6.01703851604869e-06,
      "loss": 0.727,
      "step": 9315
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.0407660007476807,
      "learning_rate": 6.00677386787456e-06,
      "loss": 0.815,
      "step": 9316
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.0482016801834106,
      "learning_rate": 5.9965177114562175e-06,
      "loss": 0.6484,
      "step": 9317
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.1582766771316528,
      "learning_rate": 5.986270047720244e-06,
      "loss": 0.8498,
      "step": 9318
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.1951152086257935,
      "learning_rate": 5.976030877592498e-06,
      "loss": 0.7673,
      "step": 9319
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.172217607498169,
      "learning_rate": 5.965800201997973e-06,
      "loss": 0.7742,
      "step": 9320
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.0843828916549683,
      "learning_rate": 5.955578021860975e-06,
      "loss": 0.8112,
      "step": 9321
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.205381989479065,
      "learning_rate": 5.945364338105053e-06,
      "loss": 0.7688,
      "step": 9322
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.0953760147094727,
      "learning_rate": 5.935159151652902e-06,
      "loss": 0.5916,
      "step": 9323
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.1771008968353271,
      "learning_rate": 5.924962463426542e-06,
      "loss": 0.8288,
      "step": 9324
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.033271312713623,
      "learning_rate": 5.914774274347168e-06,
      "loss": 0.6563,
      "step": 9325
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.08896803855896,
      "learning_rate": 5.904594585335221e-06,
      "loss": 0.7908,
      "step": 9326
    },
    {
      "epoch": 3.5,
      "grad_norm": 1.1287399530410767,
      "learning_rate": 5.8944233973104e-06,
      "loss": 0.8376,
      "step": 9327
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.1184924840927124,
      "learning_rate": 5.884260711191591e-06,
      "loss": 0.7168,
      "step": 9328
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0064038038253784,
      "learning_rate": 5.87410652789695e-06,
      "loss": 0.7668,
      "step": 9329
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.1244049072265625,
      "learning_rate": 5.8639608483438305e-06,
      "loss": 0.7594,
      "step": 9330
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0200968980789185,
      "learning_rate": 5.853823673448877e-06,
      "loss": 0.711,
      "step": 9331
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.1480824947357178,
      "learning_rate": 5.8436950041278915e-06,
      "loss": 0.7852,
      "step": 9332
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.04997718334198,
      "learning_rate": 5.8335748412959305e-06,
      "loss": 0.7264,
      "step": 9333
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0580756664276123,
      "learning_rate": 5.823463185867329e-06,
      "loss": 0.4326,
      "step": 9334
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0175821781158447,
      "learning_rate": 5.813360038755611e-06,
      "loss": 0.7274,
      "step": 9335
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0774815082550049,
      "learning_rate": 5.803265400873514e-06,
      "loss": 0.7504,
      "step": 9336
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.1047577857971191,
      "learning_rate": 5.793179273133054e-06,
      "loss": 0.817,
      "step": 9337
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.13398277759552,
      "learning_rate": 5.783101656445455e-06,
      "loss": 0.6983,
      "step": 9338
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.2331796884536743,
      "learning_rate": 5.773032551721169e-06,
      "loss": 0.8807,
      "step": 9339
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.977189302444458,
      "learning_rate": 5.762971959869856e-06,
      "loss": 0.7333,
      "step": 9340
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0552687644958496,
      "learning_rate": 5.752919881800489e-06,
      "loss": 0.7035,
      "step": 9341
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.1275651454925537,
      "learning_rate": 5.742876318421176e-06,
      "loss": 0.8234,
      "step": 9342
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0209275484085083,
      "learning_rate": 5.732841270639288e-06,
      "loss": 0.7218,
      "step": 9343
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.1622731685638428,
      "learning_rate": 5.722814739361459e-06,
      "loss": 0.8217,
      "step": 9344
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0807585716247559,
      "learning_rate": 5.712796725493541e-06,
      "loss": 0.7675,
      "step": 9345
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.9779640436172485,
      "learning_rate": 5.702787229940554e-06,
      "loss": 0.6342,
      "step": 9346
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.1358660459518433,
      "learning_rate": 5.6927862536068635e-06,
      "loss": 0.7018,
      "step": 9347
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0226969718933105,
      "learning_rate": 5.682793797395958e-06,
      "loss": 0.5643,
      "step": 9348
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.121750831604004,
      "learning_rate": 5.672809862210604e-06,
      "loss": 0.6079,
      "step": 9349
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0560060739517212,
      "learning_rate": 5.662834448952803e-06,
      "loss": 0.6755,
      "step": 9350
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0009804964065552,
      "learning_rate": 5.652867558523778e-06,
      "loss": 0.5704,
      "step": 9351
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.1393840312957764,
      "learning_rate": 5.642909191823975e-06,
      "loss": 0.8372,
      "step": 9352
    },
    {
      "epoch": 3.51,
      "grad_norm": 1.0839788913726807,
      "learning_rate": 5.632959349753064e-06,
      "loss": 0.8299,
      "step": 9353
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.316831350326538,
      "learning_rate": 5.623018033210004e-06,
      "loss": 0.7045,
      "step": 9354
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.9953253269195557,
      "learning_rate": 5.613085243092875e-06,
      "loss": 0.6621,
      "step": 9355
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.073520541191101,
      "learning_rate": 5.6031609802990716e-06,
      "loss": 0.7864,
      "step": 9356
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.1314038038253784,
      "learning_rate": 5.593245245725232e-06,
      "loss": 0.7044,
      "step": 9357
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.1472229957580566,
      "learning_rate": 5.583338040267127e-06,
      "loss": 0.6193,
      "step": 9358
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.0609694719314575,
      "learning_rate": 5.573439364819855e-06,
      "loss": 0.7217,
      "step": 9359
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.9879776239395142,
      "learning_rate": 5.563549220277698e-06,
      "loss": 0.6442,
      "step": 9360
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.094910740852356,
      "learning_rate": 5.553667607534163e-06,
      "loss": 0.6857,
      "step": 9361
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.9375484585762024,
      "learning_rate": 5.5437945274820044e-06,
      "loss": 0.5475,
      "step": 9362
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.0470213890075684,
      "learning_rate": 5.5339299810132195e-06,
      "loss": 0.7947,
      "step": 9363
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.0474797487258911,
      "learning_rate": 5.524073969018984e-06,
      "loss": 0.6745,
      "step": 9364
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.049283504486084,
      "learning_rate": 5.514226492389729e-06,
      "loss": 0.7734,
      "step": 9365
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.1494179964065552,
      "learning_rate": 5.504387552015156e-06,
      "loss": 0.6897,
      "step": 9366
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.1172934770584106,
      "learning_rate": 5.494557148784119e-06,
      "loss": 0.6676,
      "step": 9367
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.1221845149993896,
      "learning_rate": 5.484735283584752e-06,
      "loss": 0.7182,
      "step": 9368
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.0873128175735474,
      "learning_rate": 5.474921957304424e-06,
      "loss": 0.6014,
      "step": 9369
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.1146165132522583,
      "learning_rate": 5.465117170829692e-06,
      "loss": 0.839,
      "step": 9370
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.0783514976501465,
      "learning_rate": 5.455320925046359e-06,
      "loss": 0.6387,
      "step": 9371
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.1522853374481201,
      "learning_rate": 5.445533220839483e-06,
      "loss": 0.7412,
      "step": 9372
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.453169345855713,
      "learning_rate": 5.435754059093312e-06,
      "loss": 0.6111,
      "step": 9373
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.0220494270324707,
      "learning_rate": 5.425983440691351e-06,
      "loss": 0.7081,
      "step": 9374
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.9741222858428955,
      "learning_rate": 5.416221366516283e-06,
      "loss": 0.6006,
      "step": 9375
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.0881893634796143,
      "learning_rate": 5.406467837450102e-06,
      "loss": 0.7714,
      "step": 9376
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.1242963075637817,
      "learning_rate": 5.396722854373959e-06,
      "loss": 0.6926,
      "step": 9377
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.104236364364624,
      "learning_rate": 5.3869864181682605e-06,
      "loss": 0.8587,
      "step": 9378
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.1668933629989624,
      "learning_rate": 5.377258529712648e-06,
      "loss": 0.7725,
      "step": 9379
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.146311640739441,
      "learning_rate": 5.367539189885973e-06,
      "loss": 0.5663,
      "step": 9380
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.235791563987732,
      "learning_rate": 5.357828399566312e-06,
      "loss": 0.6407,
      "step": 9381
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.135084867477417,
      "learning_rate": 5.348126159631006e-06,
      "loss": 0.6423,
      "step": 9382
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.086205244064331,
      "learning_rate": 5.338432470956589e-06,
      "loss": 0.7641,
      "step": 9383
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.0854499340057373,
      "learning_rate": 5.328747334418827e-06,
      "loss": 0.6943,
      "step": 9384
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.0190761089324951,
      "learning_rate": 5.319070750892696e-06,
      "loss": 0.6735,
      "step": 9385
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.9688328504562378,
      "learning_rate": 5.309402721252454e-06,
      "loss": 0.7624,
      "step": 9386
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.0789661407470703,
      "learning_rate": 5.2997432463715466e-06,
      "loss": 0.8296,
      "step": 9387
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.1297600269317627,
      "learning_rate": 5.29009232712262e-06,
      "loss": 0.6872,
      "step": 9388
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.062487244606018,
      "learning_rate": 5.280449964377643e-06,
      "loss": 0.5835,
      "step": 9389
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.9823678135871887,
      "learning_rate": 5.270816159007674e-06,
      "loss": 0.657,
      "step": 9390
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.9812545776367188,
      "learning_rate": 5.2611909118831185e-06,
      "loss": 0.6689,
      "step": 9391
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.2849899530410767,
      "learning_rate": 5.251574223873579e-06,
      "loss": 0.8093,
      "step": 9392
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.159959316253662,
      "learning_rate": 5.241966095847817e-06,
      "loss": 0.6084,
      "step": 9393
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.2984310388565063,
      "learning_rate": 5.232366528673915e-06,
      "loss": 0.7897,
      "step": 9394
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.065385341644287,
      "learning_rate": 5.222775523219125e-06,
      "loss": 0.6733,
      "step": 9395
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.0370830297470093,
      "learning_rate": 5.213193080349921e-06,
      "loss": 0.7598,
      "step": 9396
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.054455041885376,
      "learning_rate": 5.203619200932064e-06,
      "loss": 0.5935,
      "step": 9397
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.0969281196594238,
      "learning_rate": 5.1940538858304526e-06,
      "loss": 0.5973,
      "step": 9398
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.1610939502716064,
      "learning_rate": 5.1844971359093165e-06,
      "loss": 0.9037,
      "step": 9399
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.0251827239990234,
      "learning_rate": 5.174948952031988e-06,
      "loss": 0.5985,
      "step": 9400
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.0163699388504028,
      "learning_rate": 5.165409335061122e-06,
      "loss": 0.6815,
      "step": 9401
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.029495120048523,
      "learning_rate": 5.155878285858606e-06,
      "loss": 0.8049,
      "step": 9402
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.9699760675430298,
      "learning_rate": 5.146355805285452e-06,
      "loss": 0.6721,
      "step": 9403
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.1431121826171875,
      "learning_rate": 5.136841894201994e-06,
      "loss": 0.7457,
      "step": 9404
    },
    {
      "epoch": 3.53,
      "grad_norm": 1.0324819087982178,
      "learning_rate": 5.1273365534677654e-06,
      "loss": 0.6069,
      "step": 9405
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.2797703742980957,
      "learning_rate": 5.117839783941503e-06,
      "loss": 0.8221,
      "step": 9406
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0912635326385498,
      "learning_rate": 5.108351586481197e-06,
      "loss": 0.8013,
      "step": 9407
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.1131420135498047,
      "learning_rate": 5.098871961944063e-06,
      "loss": 0.7943,
      "step": 9408
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.1049731969833374,
      "learning_rate": 5.0894009111865035e-06,
      "loss": 0.7831,
      "step": 9409
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.075990080833435,
      "learning_rate": 5.079938435064191e-06,
      "loss": 0.6763,
      "step": 9410
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.1200283765792847,
      "learning_rate": 5.070484534432007e-06,
      "loss": 0.7034,
      "step": 9411
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0925379991531372,
      "learning_rate": 5.061039210144058e-06,
      "loss": 0.7187,
      "step": 9412
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0571510791778564,
      "learning_rate": 5.0516024630536614e-06,
      "loss": 0.6916,
      "step": 9413
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0604404211044312,
      "learning_rate": 5.042174294013402e-06,
      "loss": 0.5824,
      "step": 9414
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.1069512367248535,
      "learning_rate": 5.032754703875042e-06,
      "loss": 0.6802,
      "step": 9415
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.007451057434082,
      "learning_rate": 5.02334369348958e-06,
      "loss": 0.5663,
      "step": 9416
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.1277846097946167,
      "learning_rate": 5.013941263707267e-06,
      "loss": 0.6328,
      "step": 9417
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.9391193985939026,
      "learning_rate": 5.004547415377559e-06,
      "loss": 0.7896,
      "step": 9418
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0121228694915771,
      "learning_rate": 4.99516214934912e-06,
      "loss": 0.6141,
      "step": 9419
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0901838541030884,
      "learning_rate": 4.985785466469861e-06,
      "loss": 0.7417,
      "step": 9420
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0849426984786987,
      "learning_rate": 4.976417367586927e-06,
      "loss": 0.7905,
      "step": 9421
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0432554483413696,
      "learning_rate": 4.967057853546653e-06,
      "loss": 0.6997,
      "step": 9422
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0092642307281494,
      "learning_rate": 4.9577069251946274e-06,
      "loss": 0.7759,
      "step": 9423
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0631169080734253,
      "learning_rate": 4.948364583375675e-06,
      "loss": 0.7999,
      "step": 9424
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0133817195892334,
      "learning_rate": 4.939030828933766e-06,
      "loss": 0.7598,
      "step": 9425
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.139329195022583,
      "learning_rate": 4.929705662712192e-06,
      "loss": 0.6136,
      "step": 9426
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.2080692052841187,
      "learning_rate": 4.920389085553445e-06,
      "loss": 0.7006,
      "step": 9427
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.0923740863800049,
      "learning_rate": 4.9110810982991976e-06,
      "loss": 0.8602,
      "step": 9428
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.1765613555908203,
      "learning_rate": 4.901781701790387e-06,
      "loss": 0.7076,
      "step": 9429
    },
    {
      "epoch": 3.54,
      "grad_norm": 1.1101971864700317,
      "learning_rate": 4.892490896867141e-06,
      "loss": 0.5879,
      "step": 9430
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.9704740047454834,
      "learning_rate": 4.8832086843688564e-06,
      "loss": 0.6167,
      "step": 9431
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.1700632572174072,
      "learning_rate": 4.873935065134116e-06,
      "loss": 0.7172,
      "step": 9432
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.1385056972503662,
      "learning_rate": 4.864670040000729e-06,
      "loss": 0.6014,
      "step": 9433
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.9632239937782288,
      "learning_rate": 4.8554136098057704e-06,
      "loss": 0.6974,
      "step": 9434
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.1704237461090088,
      "learning_rate": 4.846165775385459e-06,
      "loss": 0.7572,
      "step": 9435
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0220500230789185,
      "learning_rate": 4.836926537575315e-06,
      "loss": 0.7115,
      "step": 9436
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0536317825317383,
      "learning_rate": 4.827695897210072e-06,
      "loss": 0.6511,
      "step": 9437
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.114321231842041,
      "learning_rate": 4.818473855123617e-06,
      "loss": 0.6086,
      "step": 9438
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0806936025619507,
      "learning_rate": 4.809260412149141e-06,
      "loss": 0.8077,
      "step": 9439
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.1288201808929443,
      "learning_rate": 4.800055569119022e-06,
      "loss": 0.6761,
      "step": 9440
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.2543715238571167,
      "learning_rate": 4.790859326864839e-06,
      "loss": 0.8337,
      "step": 9441
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0790655612945557,
      "learning_rate": 4.78167168621747e-06,
      "loss": 0.7256,
      "step": 9442
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.978199303150177,
      "learning_rate": 4.772492648006932e-06,
      "loss": 0.6966,
      "step": 9443
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0962568521499634,
      "learning_rate": 4.763322213062504e-06,
      "loss": 0.7192,
      "step": 9444
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0841591358184814,
      "learning_rate": 4.7541603822126805e-06,
      "loss": 0.7397,
      "step": 9445
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0186445713043213,
      "learning_rate": 4.745007156285197e-06,
      "loss": 0.6346,
      "step": 9446
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0834320783615112,
      "learning_rate": 4.735862536106972e-06,
      "loss": 0.8598,
      "step": 9447
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.1777695417404175,
      "learning_rate": 4.726726522504188e-06,
      "loss": 0.7792,
      "step": 9448
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.017266869544983,
      "learning_rate": 4.717599116302229e-06,
      "loss": 0.7455,
      "step": 9449
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0764591693878174,
      "learning_rate": 4.7084803183257035e-06,
      "loss": 0.7792,
      "step": 9450
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.9416154623031616,
      "learning_rate": 4.699370129398428e-06,
      "loss": 0.5732,
      "step": 9451
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.112833857536316,
      "learning_rate": 4.690268550343491e-06,
      "loss": 0.7053,
      "step": 9452
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.2027411460876465,
      "learning_rate": 4.681175581983144e-06,
      "loss": 0.6647,
      "step": 9453
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.9877023100852966,
      "learning_rate": 4.672091225138886e-06,
      "loss": 0.7609,
      "step": 9454
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0734071731567383,
      "learning_rate": 4.663015480631428e-06,
      "loss": 0.7893,
      "step": 9455
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.1049911975860596,
      "learning_rate": 4.653948349280734e-06,
      "loss": 0.7604,
      "step": 9456
    },
    {
      "epoch": 3.55,
      "grad_norm": 1.0704501867294312,
      "learning_rate": 4.644889831905974e-06,
      "loss": 0.7846,
      "step": 9457
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.074708104133606,
      "learning_rate": 4.635839929325492e-06,
      "loss": 0.808,
      "step": 9458
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1176865100860596,
      "learning_rate": 4.626798642356944e-06,
      "loss": 0.8615,
      "step": 9459
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.0474516153335571,
      "learning_rate": 4.617765971817123e-06,
      "loss": 0.6551,
      "step": 9460
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.0684980154037476,
      "learning_rate": 4.608741918522097e-06,
      "loss": 0.6976,
      "step": 9461
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.0753333568572998,
      "learning_rate": 4.599726483287136e-06,
      "loss": 0.7872,
      "step": 9462
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.9769883155822754,
      "learning_rate": 4.590719666926735e-06,
      "loss": 0.6956,
      "step": 9463
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.0145368576049805,
      "learning_rate": 4.581721470254608e-06,
      "loss": 0.6518,
      "step": 9464
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1712994575500488,
      "learning_rate": 4.572731894083682e-06,
      "loss": 0.7325,
      "step": 9465
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1277647018432617,
      "learning_rate": 4.56375093922613e-06,
      "loss": 0.7892,
      "step": 9466
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.0939749479293823,
      "learning_rate": 4.554778606493315e-06,
      "loss": 0.9303,
      "step": 9467
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.9853759407997131,
      "learning_rate": 4.545814896695844e-06,
      "loss": 0.6644,
      "step": 9468
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.2465983629226685,
      "learning_rate": 4.536859810643545e-06,
      "loss": 0.8834,
      "step": 9469
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.0126070976257324,
      "learning_rate": 4.527913349145441e-06,
      "loss": 0.7205,
      "step": 9470
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1108355522155762,
      "learning_rate": 4.518975513009793e-06,
      "loss": 0.8028,
      "step": 9471
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.2100151777267456,
      "learning_rate": 4.5100463030441355e-06,
      "loss": 0.7656,
      "step": 9472
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.0947849750518799,
      "learning_rate": 4.501125720055089e-06,
      "loss": 0.7739,
      "step": 9473
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.0809351205825806,
      "learning_rate": 4.492213764848652e-06,
      "loss": 0.7745,
      "step": 9474
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.0355212688446045,
      "learning_rate": 4.483310438229926e-06,
      "loss": 0.8224,
      "step": 9475
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1534838676452637,
      "learning_rate": 4.474415741003268e-06,
      "loss": 0.7819,
      "step": 9476
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1931309700012207,
      "learning_rate": 4.465529673972313e-06,
      "loss": 0.7259,
      "step": 9477
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.3788143396377563,
      "learning_rate": 4.456652237939829e-06,
      "loss": 0.8035,
      "step": 9478
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.187548041343689,
      "learning_rate": 4.447783433707842e-06,
      "loss": 0.6969,
      "step": 9479
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1373940706253052,
      "learning_rate": 4.43892326207761e-06,
      "loss": 0.8659,
      "step": 9480
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1595747470855713,
      "learning_rate": 4.430071723849594e-06,
      "loss": 0.8151,
      "step": 9481
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.0749086141586304,
      "learning_rate": 4.421228819823498e-06,
      "loss": 0.5164,
      "step": 9482
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1445127725601196,
      "learning_rate": 4.412394550798182e-06,
      "loss": 0.7336,
      "step": 9483
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.045810341835022,
      "learning_rate": 4.403568917571832e-06,
      "loss": 0.7837,
      "step": 9484
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.0188159942626953,
      "learning_rate": 4.394751920941753e-06,
      "loss": 0.6917,
      "step": 9485
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.0313225984573364,
      "learning_rate": 4.385943561704508e-06,
      "loss": 0.7087,
      "step": 9486
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.0635348558425903,
      "learning_rate": 4.3771438406559175e-06,
      "loss": 0.8137,
      "step": 9487
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.1369999647140503,
      "learning_rate": 4.368352758590966e-06,
      "loss": 0.7106,
      "step": 9488
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.0339741706848145,
      "learning_rate": 4.359570316303874e-06,
      "loss": 0.7545,
      "step": 9489
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.2778940200805664,
      "learning_rate": 4.3507965145880846e-06,
      "loss": 0.8184,
      "step": 9490
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.1356524229049683,
      "learning_rate": 4.342031354236265e-06,
      "loss": 0.679,
      "step": 9491
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.1680208444595337,
      "learning_rate": 4.333274836040302e-06,
      "loss": 0.7687,
      "step": 9492
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.1188141107559204,
      "learning_rate": 4.3245269607912866e-06,
      "loss": 0.8253,
      "step": 9493
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.9873570799827576,
      "learning_rate": 4.315787729279563e-06,
      "loss": 0.6649,
      "step": 9494
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.0524026155471802,
      "learning_rate": 4.307057142294646e-06,
      "loss": 0.7254,
      "step": 9495
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.0983495712280273,
      "learning_rate": 4.298335200625303e-06,
      "loss": 0.7548,
      "step": 9496
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.0813517570495605,
      "learning_rate": 4.289621905059516e-06,
      "loss": 0.8022,
      "step": 9497
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.1627041101455688,
      "learning_rate": 4.280917256384476e-06,
      "loss": 0.7343,
      "step": 9498
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.071501612663269,
      "learning_rate": 4.2722212553866105e-06,
      "loss": 0.7057,
      "step": 9499
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.101823091506958,
      "learning_rate": 4.263533902851535e-06,
      "loss": 0.7877,
      "step": 9500
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.0422240495681763,
      "learning_rate": 4.254855199564112e-06,
      "loss": 0.6636,
      "step": 9501
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.9442390203475952,
      "learning_rate": 4.2461851463084236e-06,
      "loss": 0.7734,
      "step": 9502
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.9244790077209473,
      "learning_rate": 4.237523743867744e-06,
      "loss": 0.7085,
      "step": 9503
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.114058494567871,
      "learning_rate": 4.2288709930246025e-06,
      "loss": 0.7525,
      "step": 9504
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.014620304107666,
      "learning_rate": 4.220226894560697e-06,
      "loss": 0.6177,
      "step": 9505
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.155929684638977,
      "learning_rate": 4.21159144925698e-06,
      "loss": 0.7547,
      "step": 9506
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.9988424777984619,
      "learning_rate": 4.202964657893649e-06,
      "loss": 0.7918,
      "step": 9507
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.5431848764419556,
      "learning_rate": 4.194346521250037e-06,
      "loss": 0.7887,
      "step": 9508
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.0519168376922607,
      "learning_rate": 4.1857370401047755e-06,
      "loss": 0.7142,
      "step": 9509
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.1756469011306763,
      "learning_rate": 4.1771362152356776e-06,
      "loss": 0.6357,
      "step": 9510
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.2993413209915161,
      "learning_rate": 4.168544047419753e-06,
      "loss": 0.8891,
      "step": 9511
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0030964612960815,
      "learning_rate": 4.159960537433305e-06,
      "loss": 0.73,
      "step": 9512
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.2626101970672607,
      "learning_rate": 4.151385686051768e-06,
      "loss": 0.7,
      "step": 9513
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0147868394851685,
      "learning_rate": 4.142819494049843e-06,
      "loss": 0.6691,
      "step": 9514
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0884225368499756,
      "learning_rate": 4.134261962201425e-06,
      "loss": 0.7698,
      "step": 9515
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.9450199604034424,
      "learning_rate": 4.125713091279659e-06,
      "loss": 0.7438,
      "step": 9516
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0475425720214844,
      "learning_rate": 4.117172882056875e-06,
      "loss": 0.7286,
      "step": 9517
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.9429231286048889,
      "learning_rate": 4.10864133530462e-06,
      "loss": 0.5811,
      "step": 9518
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0960689783096313,
      "learning_rate": 4.1001184517936996e-06,
      "loss": 0.7352,
      "step": 9519
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.2188198566436768,
      "learning_rate": 4.091604232294099e-06,
      "loss": 0.8286,
      "step": 9520
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.9997052550315857,
      "learning_rate": 4.083098677575003e-06,
      "loss": 0.7619,
      "step": 9521
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.1210790872573853,
      "learning_rate": 4.074601788404875e-06,
      "loss": 0.6359,
      "step": 9522
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.058549404144287,
      "learning_rate": 4.066113565551355e-06,
      "loss": 0.6137,
      "step": 9523
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.1089754104614258,
      "learning_rate": 4.057634009781286e-06,
      "loss": 0.7132,
      "step": 9524
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.046700119972229,
      "learning_rate": 4.049163121860755e-06,
      "loss": 0.8138,
      "step": 9525
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0299712419509888,
      "learning_rate": 4.040700902555073e-06,
      "loss": 0.7288,
      "step": 9526
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0212887525558472,
      "learning_rate": 4.032247352628748e-06,
      "loss": 0.6966,
      "step": 9527
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.145952820777893,
      "learning_rate": 4.023802472845495e-06,
      "loss": 0.9067,
      "step": 9528
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.9756731390953064,
      "learning_rate": 4.015366263968279e-06,
      "loss": 0.7298,
      "step": 9529
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.1324797868728638,
      "learning_rate": 4.006938726759268e-06,
      "loss": 0.7676,
      "step": 9530
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0332717895507812,
      "learning_rate": 3.99851986197981e-06,
      "loss": 0.7608,
      "step": 9531
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0036720037460327,
      "learning_rate": 3.9901096703905405e-06,
      "loss": 0.7003,
      "step": 9532
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0530592203140259,
      "learning_rate": 3.981708152751252e-06,
      "loss": 0.7386,
      "step": 9533
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.1703906059265137,
      "learning_rate": 3.97331530982098e-06,
      "loss": 0.7272,
      "step": 9534
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.0697475671768188,
      "learning_rate": 3.964931142357952e-06,
      "loss": 0.8147,
      "step": 9535
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.1214314699172974,
      "learning_rate": 3.956555651119664e-06,
      "loss": 0.6931,
      "step": 9536
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.106798529624939,
      "learning_rate": 3.948188836862776e-06,
      "loss": 0.7605,
      "step": 9537
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0015952587127686,
      "learning_rate": 3.939830700343172e-06,
      "loss": 0.6528,
      "step": 9538
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.1472705602645874,
      "learning_rate": 3.931481242315993e-06,
      "loss": 0.7548,
      "step": 9539
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.3280612230300903,
      "learning_rate": 3.923140463535524e-06,
      "loss": 0.6958,
      "step": 9540
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0619010925292969,
      "learning_rate": 3.914808364755318e-06,
      "loss": 0.6411,
      "step": 9541
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.174285650253296,
      "learning_rate": 3.906484946728184e-06,
      "loss": 0.7078,
      "step": 9542
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.144360899925232,
      "learning_rate": 3.89817021020602e-06,
      "loss": 0.6894,
      "step": 9543
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0451254844665527,
      "learning_rate": 3.8898641559400595e-06,
      "loss": 0.7953,
      "step": 9544
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.2547177076339722,
      "learning_rate": 3.881566784680712e-06,
      "loss": 0.8491,
      "step": 9545
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0960153341293335,
      "learning_rate": 3.873278097177557e-06,
      "loss": 0.7919,
      "step": 9546
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.137865662574768,
      "learning_rate": 3.864998094179484e-06,
      "loss": 0.7123,
      "step": 9547
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.1249855756759644,
      "learning_rate": 3.856726776434505e-06,
      "loss": 0.7122,
      "step": 9548
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.093711018562317,
      "learning_rate": 3.848464144689912e-06,
      "loss": 0.7635,
      "step": 9549
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0363121032714844,
      "learning_rate": 3.840210199692152e-06,
      "loss": 0.6806,
      "step": 9550
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.9843497276306152,
      "learning_rate": 3.8319649421869495e-06,
      "loss": 0.6816,
      "step": 9551
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0178742408752441,
      "learning_rate": 3.8237283729192445e-06,
      "loss": 0.6544,
      "step": 9552
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.159450888633728,
      "learning_rate": 3.815500492633095e-06,
      "loss": 0.7724,
      "step": 9553
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.180312156677246,
      "learning_rate": 3.8072813020719077e-06,
      "loss": 0.7019,
      "step": 9554
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.2326253652572632,
      "learning_rate": 3.7990708019781994e-06,
      "loss": 0.7871,
      "step": 9555
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.2589869499206543,
      "learning_rate": 3.7908689930937548e-06,
      "loss": 0.7656,
      "step": 9556
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.1132445335388184,
      "learning_rate": 3.782675876159558e-06,
      "loss": 0.8193,
      "step": 9557
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0724929571151733,
      "learning_rate": 3.774491451915829e-06,
      "loss": 0.6622,
      "step": 9558
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0773063898086548,
      "learning_rate": 3.766315721101965e-06,
      "loss": 0.8265,
      "step": 9559
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.066756248474121,
      "learning_rate": 3.758148684456597e-06,
      "loss": 0.6972,
      "step": 9560
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0886008739471436,
      "learning_rate": 3.7499903427175796e-06,
      "loss": 0.774,
      "step": 9561
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.3207203149795532,
      "learning_rate": 3.7418406966219675e-06,
      "loss": 0.7871,
      "step": 9562
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.1386826038360596,
      "learning_rate": 3.7336997469060276e-06,
      "loss": 0.6761,
      "step": 9563
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.9922816157341003,
      "learning_rate": 3.725567494305271e-06,
      "loss": 0.6297,
      "step": 9564
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.0554293394088745,
      "learning_rate": 3.717443939554388e-06,
      "loss": 0.7341,
      "step": 9565
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.1159563064575195,
      "learning_rate": 3.7093290833872698e-06,
      "loss": 0.7157,
      "step": 9566
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.9422329664230347,
      "learning_rate": 3.7012229265370957e-06,
      "loss": 0.747,
      "step": 9567
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.0567890405654907,
      "learning_rate": 3.6931254697361917e-06,
      "loss": 0.797,
      "step": 9568
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.0781539678573608,
      "learning_rate": 3.6850367137161058e-06,
      "loss": 0.7933,
      "step": 9569
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.1125067472457886,
      "learning_rate": 3.6769566592076087e-06,
      "loss": 0.6991,
      "step": 9570
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.1257882118225098,
      "learning_rate": 3.668885306940706e-06,
      "loss": 0.8072,
      "step": 9571
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.1065607070922852,
      "learning_rate": 3.6608226576446023e-06,
      "loss": 0.6327,
      "step": 9572
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.0312420129776,
      "learning_rate": 3.652768712047683e-06,
      "loss": 0.6657,
      "step": 9573
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.9149978756904602,
      "learning_rate": 3.644723470877609e-06,
      "loss": 0.6452,
      "step": 9574
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.0435965061187744,
      "learning_rate": 3.6366869348611887e-06,
      "loss": 0.6813,
      "step": 9575
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.0449802875518799,
      "learning_rate": 3.6286591047244965e-06,
      "loss": 0.7661,
      "step": 9576
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.1289963722229004,
      "learning_rate": 3.6206399811928306e-06,
      "loss": 0.7231,
      "step": 9577
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.961563229560852,
      "learning_rate": 3.6126295649906216e-06,
      "loss": 0.7454,
      "step": 9578
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.1868175268173218,
      "learning_rate": 3.6046278568415916e-06,
      "loss": 0.8665,
      "step": 9579
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.0859922170639038,
      "learning_rate": 3.5966348574686283e-06,
      "loss": 0.6864,
      "step": 9580
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.0696438550949097,
      "learning_rate": 3.588650567593899e-06,
      "loss": 0.741,
      "step": 9581
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.1079578399658203,
      "learning_rate": 3.5806749879387034e-06,
      "loss": 0.8848,
      "step": 9582
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.9780482649803162,
      "learning_rate": 3.5727081192235778e-06,
      "loss": 0.6834,
      "step": 9583
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.9938876032829285,
      "learning_rate": 3.5647499621683344e-06,
      "loss": 0.6882,
      "step": 9584
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.0841304063796997,
      "learning_rate": 3.5568005174918983e-06,
      "loss": 0.7525,
      "step": 9585
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.2428840398788452,
      "learning_rate": 3.5488597859124727e-06,
      "loss": 0.6329,
      "step": 9586
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.088956356048584,
      "learning_rate": 3.540927768147484e-06,
      "loss": 0.598,
      "step": 9587
    },
    {
      "epoch": 3.6,
      "grad_norm": 1.1099416017532349,
      "learning_rate": 3.533004464913492e-06,
      "loss": 0.6445,
      "step": 9588
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.2654850482940674,
      "learning_rate": 3.5250898769263686e-06,
      "loss": 0.8536,
      "step": 9589
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.0084056854248047,
      "learning_rate": 3.517184004901131e-06,
      "loss": 0.5685,
      "step": 9590
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.027130365371704,
      "learning_rate": 3.5092868495520292e-06,
      "loss": 0.5242,
      "step": 9591
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1001675128936768,
      "learning_rate": 3.501398411592538e-06,
      "loss": 0.8396,
      "step": 9592
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1401519775390625,
      "learning_rate": 3.493518691735331e-06,
      "loss": 0.6252,
      "step": 9593
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1631479263305664,
      "learning_rate": 3.4856476906922954e-06,
      "loss": 0.7256,
      "step": 9594
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.0147554874420166,
      "learning_rate": 3.4777854091745056e-06,
      "loss": 0.6519,
      "step": 9595
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.0893521308898926,
      "learning_rate": 3.4699318478923158e-06,
      "loss": 0.7738,
      "step": 9596
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.9949012994766235,
      "learning_rate": 3.4620870075552257e-06,
      "loss": 0.6352,
      "step": 9597
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.0625600814819336,
      "learning_rate": 3.454250888871979e-06,
      "loss": 0.7924,
      "step": 9598
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.2157313823699951,
      "learning_rate": 3.4464234925505213e-06,
      "loss": 0.7083,
      "step": 9599
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.0546051263809204,
      "learning_rate": 3.4386048192980305e-06,
      "loss": 0.6242,
      "step": 9600
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1151278018951416,
      "learning_rate": 3.4307948698208416e-06,
      "loss": 0.5664,
      "step": 9601
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1020997762680054,
      "learning_rate": 3.4229936448245793e-06,
      "loss": 0.656,
      "step": 9602
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.056983470916748,
      "learning_rate": 3.415201145014024e-06,
      "loss": 0.7267,
      "step": 9603
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1382547616958618,
      "learning_rate": 3.40741737109318e-06,
      "loss": 0.7283,
      "step": 9604
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.111603856086731,
      "learning_rate": 3.399642323765262e-06,
      "loss": 0.7204,
      "step": 9605
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.954963207244873,
      "learning_rate": 3.391876003732719e-06,
      "loss": 0.5688,
      "step": 9606
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1071138381958008,
      "learning_rate": 3.38411841169719e-06,
      "loss": 0.718,
      "step": 9607
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.0896059274673462,
      "learning_rate": 3.3763695483595037e-06,
      "loss": 0.8249,
      "step": 9608
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.9830428957939148,
      "learning_rate": 3.3686294144197662e-06,
      "loss": 0.6382,
      "step": 9609
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1327486038208008,
      "learning_rate": 3.3608980105772403e-06,
      "loss": 0.7794,
      "step": 9610
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1177420616149902,
      "learning_rate": 3.3531753375303897e-06,
      "loss": 0.7876,
      "step": 9611
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1888068914413452,
      "learning_rate": 3.3454613959769566e-06,
      "loss": 0.8254,
      "step": 9612
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.2450792789459229,
      "learning_rate": 3.337756186613816e-06,
      "loss": 0.7925,
      "step": 9613
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.1927120685577393,
      "learning_rate": 3.3300597101371125e-06,
      "loss": 0.731,
      "step": 9614
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.0863752365112305,
      "learning_rate": 3.322371967242155e-06,
      "loss": 0.6645,
      "step": 9615
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.142935872077942,
      "learning_rate": 3.314692958623522e-06,
      "loss": 0.8465,
      "step": 9616
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.9527440667152405,
      "learning_rate": 3.3070226849749366e-06,
      "loss": 0.6556,
      "step": 9617
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0439012050628662,
      "learning_rate": 3.299361146989377e-06,
      "loss": 0.6257,
      "step": 9618
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.9317110776901245,
      "learning_rate": 3.2917083453590347e-06,
      "loss": 0.6153,
      "step": 9619
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.9573559761047363,
      "learning_rate": 3.2840642807752677e-06,
      "loss": 0.8066,
      "step": 9620
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0099294185638428,
      "learning_rate": 3.2764289539286894e-06,
      "loss": 0.7432,
      "step": 9621
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.9000482559204102,
      "learning_rate": 3.2688023655091273e-06,
      "loss": 0.5393,
      "step": 9622
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.1031389236450195,
      "learning_rate": 3.261184516205551e-06,
      "loss": 0.7637,
      "step": 9623
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0536938905715942,
      "learning_rate": 3.2535754067062442e-06,
      "loss": 0.5692,
      "step": 9624
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0363940000534058,
      "learning_rate": 3.245975037698612e-06,
      "loss": 0.5792,
      "step": 9625
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.2607051134109497,
      "learning_rate": 3.2383834098693057e-06,
      "loss": 0.7855,
      "step": 9626
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.1446163654327393,
      "learning_rate": 3.2308005239042093e-06,
      "loss": 0.7893,
      "step": 9627
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0553935766220093,
      "learning_rate": 3.223226380488387e-06,
      "loss": 0.7906,
      "step": 9628
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0501688718795776,
      "learning_rate": 3.215660980306101e-06,
      "loss": 0.6669,
      "step": 9629
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.9783391952514648,
      "learning_rate": 3.2081043240408505e-06,
      "loss": 0.7427,
      "step": 9630
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0918532609939575,
      "learning_rate": 3.2005564123753438e-06,
      "loss": 0.483,
      "step": 9631
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.1988056898117065,
      "learning_rate": 3.193017245991492e-06,
      "loss": 0.7613,
      "step": 9632
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0268480777740479,
      "learning_rate": 3.1854868255704053e-06,
      "loss": 0.6312,
      "step": 9633
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.059897780418396,
      "learning_rate": 3.177965151792428e-06,
      "loss": 0.6416,
      "step": 9634
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.1083929538726807,
      "learning_rate": 3.1704522253370947e-06,
      "loss": 0.7882,
      "step": 9635
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.9673347473144531,
      "learning_rate": 3.1629480468831405e-06,
      "loss": 0.6576,
      "step": 9636
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.1394681930541992,
      "learning_rate": 3.1554526171085562e-06,
      "loss": 0.7843,
      "step": 9637
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0167393684387207,
      "learning_rate": 3.147965936690489e-06,
      "loss": 0.8226,
      "step": 9638
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0200989246368408,
      "learning_rate": 3.140488006305331e-06,
      "loss": 0.8262,
      "step": 9639
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.2328968048095703,
      "learning_rate": 3.1330188266286533e-06,
      "loss": 0.7328,
      "step": 9640
    },
    {
      "epoch": 3.62,
      "grad_norm": 1.0261443853378296,
      "learning_rate": 3.125558398335282e-06,
      "loss": 0.7733,
      "step": 9641
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0234793424606323,
      "learning_rate": 3.118106722099201e-06,
      "loss": 0.7147,
      "step": 9642
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.019017219543457,
      "learning_rate": 3.110663798593616e-06,
      "loss": 0.7381,
      "step": 9643
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0557971000671387,
      "learning_rate": 3.103229628491e-06,
      "loss": 0.5783,
      "step": 9644
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.1116085052490234,
      "learning_rate": 3.0958042124629383e-06,
      "loss": 0.6526,
      "step": 9645
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0065749883651733,
      "learning_rate": 3.088387551180305e-06,
      "loss": 0.6884,
      "step": 9646
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.024464726448059,
      "learning_rate": 3.080979645313142e-06,
      "loss": 0.7087,
      "step": 9647
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.9196699857711792,
      "learning_rate": 3.0735804955307146e-06,
      "loss": 0.3663,
      "step": 9648
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.1227691173553467,
      "learning_rate": 3.0661901025014984e-06,
      "loss": 0.7903,
      "step": 9649
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0344467163085938,
      "learning_rate": 3.0588084668931482e-06,
      "loss": 0.7109,
      "step": 9650
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0533043146133423,
      "learning_rate": 3.0514355893725975e-06,
      "loss": 0.6401,
      "step": 9651
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0511938333511353,
      "learning_rate": 3.044071470605914e-06,
      "loss": 0.667,
      "step": 9652
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0700619220733643,
      "learning_rate": 3.0367161112583973e-06,
      "loss": 0.7517,
      "step": 9653
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.2056522369384766,
      "learning_rate": 3.029369511994595e-06,
      "loss": 0.8362,
      "step": 9654
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.2480723857879639,
      "learning_rate": 3.0220316734781984e-06,
      "loss": 0.7755,
      "step": 9655
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.048323392868042,
      "learning_rate": 3.0147025963721433e-06,
      "loss": 0.574,
      "step": 9656
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0318301916122437,
      "learning_rate": 3.007382281338611e-06,
      "loss": 0.7015,
      "step": 9657
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0469781160354614,
      "learning_rate": 3.0000707290388952e-06,
      "loss": 0.7005,
      "step": 9658
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.122330904006958,
      "learning_rate": 2.9927679401335785e-06,
      "loss": 0.7054,
      "step": 9659
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0453784465789795,
      "learning_rate": 2.985473915282433e-06,
      "loss": 0.718,
      "step": 9660
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.2262742519378662,
      "learning_rate": 2.9781886551444095e-06,
      "loss": 0.6908,
      "step": 9661
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0093944072723389,
      "learning_rate": 2.9709121603777147e-06,
      "loss": 0.5727,
      "step": 9662
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.1152524948120117,
      "learning_rate": 2.9636444316397337e-06,
      "loss": 0.6938,
      "step": 9663
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.158189296722412,
      "learning_rate": 2.956385469587053e-06,
      "loss": 0.7293,
      "step": 9664
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.0365028381347656,
      "learning_rate": 2.9491352748754807e-06,
      "loss": 0.6045,
      "step": 9665
    },
    {
      "epoch": 3.63,
      "grad_norm": 1.1461862325668335,
      "learning_rate": 2.941893848160049e-06,
      "loss": 1.1584,
      "step": 9666
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.98199462890625,
      "learning_rate": 2.9346611900949563e-06,
      "loss": 0.5747,
      "step": 9667
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.026784062385559,
      "learning_rate": 2.9274373013336465e-06,
      "loss": 0.6845,
      "step": 9668
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.9717198014259338,
      "learning_rate": 2.9202221825287533e-06,
      "loss": 0.7112,
      "step": 9669
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0753607749938965,
      "learning_rate": 2.9130158343321224e-06,
      "loss": 0.784,
      "step": 9670
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0746945142745972,
      "learning_rate": 2.905818257394799e-06,
      "loss": 0.6653,
      "step": 9671
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.1293368339538574,
      "learning_rate": 2.898629452367052e-06,
      "loss": 0.693,
      "step": 9672
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.2853316068649292,
      "learning_rate": 2.891449419898362e-06,
      "loss": 0.7123,
      "step": 9673
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.1713504791259766,
      "learning_rate": 2.884278160637377e-06,
      "loss": 0.7017,
      "step": 9674
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.081799864768982,
      "learning_rate": 2.8771156752319894e-06,
      "loss": 0.6878,
      "step": 9675
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.9564076662063599,
      "learning_rate": 2.869961964329293e-06,
      "loss": 0.5719,
      "step": 9676
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.203377366065979,
      "learning_rate": 2.862817028575593e-06,
      "loss": 0.7408,
      "step": 9677
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0537962913513184,
      "learning_rate": 2.855680868616373e-06,
      "loss": 0.7921,
      "step": 9678
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0306793451309204,
      "learning_rate": 2.8485534850963723e-06,
      "loss": 0.7246,
      "step": 9679
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.174821376800537,
      "learning_rate": 2.841434878659488e-06,
      "loss": 0.807,
      "step": 9680
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.1522022485733032,
      "learning_rate": 2.834325049948838e-06,
      "loss": 0.7895,
      "step": 9681
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0366876125335693,
      "learning_rate": 2.8272239996067874e-06,
      "loss": 0.7425,
      "step": 9682
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.2727211713790894,
      "learning_rate": 2.8201317282748552e-06,
      "loss": 0.6738,
      "step": 9683
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.1075066328048706,
      "learning_rate": 2.813048236593785e-06,
      "loss": 0.7418,
      "step": 9684
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.9733402729034424,
      "learning_rate": 2.8059735252035313e-06,
      "loss": 0.7949,
      "step": 9685
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0424128770828247,
      "learning_rate": 2.798907594743272e-06,
      "loss": 0.7767,
      "step": 9686
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.9637560844421387,
      "learning_rate": 2.7918504458513516e-06,
      "loss": 0.6695,
      "step": 9687
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0607154369354248,
      "learning_rate": 2.7848020791653493e-06,
      "loss": 0.6256,
      "step": 9688
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0816208124160767,
      "learning_rate": 2.7777624953220672e-06,
      "loss": 0.7523,
      "step": 9689
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0990419387817383,
      "learning_rate": 2.7707316949574513e-06,
      "loss": 0.8434,
      "step": 9690
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.1480551958084106,
      "learning_rate": 2.763709678706716e-06,
      "loss": 0.8061,
      "step": 9691
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.022851586341858,
      "learning_rate": 2.7566964472042765e-06,
      "loss": 0.695,
      "step": 9692
    },
    {
      "epoch": 3.64,
      "grad_norm": 1.0613906383514404,
      "learning_rate": 2.7496920010837036e-06,
      "loss": 0.6685,
      "step": 9693
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.9984630346298218,
      "learning_rate": 2.742696340977846e-06,
      "loss": 0.7301,
      "step": 9694
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0692126750946045,
      "learning_rate": 2.735709467518699e-06,
      "loss": 0.7015,
      "step": 9695
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.1338649988174438,
      "learning_rate": 2.7287313813374906e-06,
      "loss": 0.6817,
      "step": 9696
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.1081340312957764,
      "learning_rate": 2.7217620830646495e-06,
      "loss": 0.7296,
      "step": 9697
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.949843168258667,
      "learning_rate": 2.714801573329828e-06,
      "loss": 0.8018,
      "step": 9698
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.086941123008728,
      "learning_rate": 2.7078498527618567e-06,
      "loss": 0.7828,
      "step": 9699
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0566928386688232,
      "learning_rate": 2.700906921988766e-06,
      "loss": 0.7788,
      "step": 9700
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0111572742462158,
      "learning_rate": 2.6939727816378436e-06,
      "loss": 0.6536,
      "step": 9701
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0589208602905273,
      "learning_rate": 2.687047432335532e-06,
      "loss": 0.7871,
      "step": 9702
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.1060186624526978,
      "learning_rate": 2.680130874707498e-06,
      "loss": 0.6066,
      "step": 9703
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.1846468448638916,
      "learning_rate": 2.67322310937862e-06,
      "loss": 0.7688,
      "step": 9704
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0330451726913452,
      "learning_rate": 2.6663241369729754e-06,
      "loss": 0.7115,
      "step": 9705
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0688211917877197,
      "learning_rate": 2.659433958113822e-06,
      "loss": 0.7065,
      "step": 9706
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0157527923583984,
      "learning_rate": 2.6525525734236944e-06,
      "loss": 0.6766,
      "step": 9707
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0492933988571167,
      "learning_rate": 2.645679983524252e-06,
      "loss": 0.806,
      "step": 9708
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0368744134902954,
      "learning_rate": 2.6388161890364082e-06,
      "loss": 0.7706,
      "step": 9709
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.1789920330047607,
      "learning_rate": 2.6319611905802565e-06,
      "loss": 0.8365,
      "step": 9710
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.060936450958252,
      "learning_rate": 2.6251149887751235e-06,
      "loss": 0.6692,
      "step": 9711
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0045349597930908,
      "learning_rate": 2.6182775842395146e-06,
      "loss": 0.6664,
      "step": 9712
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.1869513988494873,
      "learning_rate": 2.6114489775911356e-06,
      "loss": 0.7481,
      "step": 9713
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0766912698745728,
      "learning_rate": 2.604629169446948e-06,
      "loss": 0.8262,
      "step": 9714
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.2064628601074219,
      "learning_rate": 2.59781816042306e-06,
      "loss": 0.7259,
      "step": 9715
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0954605340957642,
      "learning_rate": 2.591015951134812e-06,
      "loss": 0.7259,
      "step": 9716
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.1014009714126587,
      "learning_rate": 2.584222542196746e-06,
      "loss": 0.7904,
      "step": 9717
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.997641384601593,
      "learning_rate": 2.577437934222604e-06,
      "loss": 0.6678,
      "step": 9718
    },
    {
      "epoch": 3.65,
      "grad_norm": 1.0851839780807495,
      "learning_rate": 2.5706621278253406e-06,
      "loss": 0.6571,
      "step": 9719
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.9889243245124817,
      "learning_rate": 2.5638951236170994e-06,
      "loss": 0.7722,
      "step": 9720
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.9658727049827576,
      "learning_rate": 2.5571369222092687e-06,
      "loss": 0.6727,
      "step": 9721
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.2902482748031616,
      "learning_rate": 2.550387524212394e-06,
      "loss": 0.6878,
      "step": 9722
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1079514026641846,
      "learning_rate": 2.543646930236232e-06,
      "loss": 0.735,
      "step": 9723
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.0050214529037476,
      "learning_rate": 2.536915140889795e-06,
      "loss": 0.489,
      "step": 9724
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1542370319366455,
      "learning_rate": 2.53019215678123e-06,
      "loss": 0.7447,
      "step": 9725
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.0960944890975952,
      "learning_rate": 2.5234779785179187e-06,
      "loss": 0.6575,
      "step": 9726
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1134847402572632,
      "learning_rate": 2.5167726067064968e-06,
      "loss": 0.6974,
      "step": 9727
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.9962137341499329,
      "learning_rate": 2.5100760419526915e-06,
      "loss": 0.8403,
      "step": 9728
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1506953239440918,
      "learning_rate": 2.503388284861541e-06,
      "loss": 0.7559,
      "step": 9729
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.0096590518951416,
      "learning_rate": 2.4967093360372285e-06,
      "loss": 0.7609,
      "step": 9730
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1617342233657837,
      "learning_rate": 2.49003919608316e-06,
      "loss": 0.7295,
      "step": 9731
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1123908758163452,
      "learning_rate": 2.483377865601966e-06,
      "loss": 0.5528,
      "step": 9732
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1398481130599976,
      "learning_rate": 2.476725345195419e-06,
      "loss": 0.6634,
      "step": 9733
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1603727340698242,
      "learning_rate": 2.470081635464594e-06,
      "loss": 0.8051,
      "step": 9734
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.0126755237579346,
      "learning_rate": 2.4634467370096558e-06,
      "loss": 0.7108,
      "step": 9735
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.089576005935669,
      "learning_rate": 2.456820650430058e-06,
      "loss": 0.6246,
      "step": 9736
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1837266683578491,
      "learning_rate": 2.4502033763244448e-06,
      "loss": 0.624,
      "step": 9737
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.176931381225586,
      "learning_rate": 2.4435949152906145e-06,
      "loss": 0.7875,
      "step": 9738
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.0680427551269531,
      "learning_rate": 2.4369952679256236e-06,
      "loss": 0.8083,
      "step": 9739
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.0503977537155151,
      "learning_rate": 2.430404434825717e-06,
      "loss": 0.7112,
      "step": 9740
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.9711684584617615,
      "learning_rate": 2.4238224165863187e-06,
      "loss": 0.6162,
      "step": 9741
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.010062336921692,
      "learning_rate": 2.4172492138020975e-06,
      "loss": 0.6872,
      "step": 9742
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1055881977081299,
      "learning_rate": 2.4106848270669e-06,
      "loss": 0.6422,
      "step": 9743
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.0001243352890015,
      "learning_rate": 2.404129256973786e-06,
      "loss": 0.5617,
      "step": 9744
    },
    {
      "epoch": 3.66,
      "grad_norm": 1.1521100997924805,
      "learning_rate": 2.3975825041149925e-06,
      "loss": 0.7939,
      "step": 9745
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.0689561367034912,
      "learning_rate": 2.3910445690820015e-06,
      "loss": 0.8219,
      "step": 9746
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.1006343364715576,
      "learning_rate": 2.3845154524654746e-06,
      "loss": 0.8064,
      "step": 9747
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.2550487518310547,
      "learning_rate": 2.3779951548552724e-06,
      "loss": 0.5829,
      "step": 9748
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.0450468063354492,
      "learning_rate": 2.37148367684048e-06,
      "loss": 0.6949,
      "step": 9749
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.2037488222122192,
      "learning_rate": 2.3649810190093712e-06,
      "loss": 0.6232,
      "step": 9750
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.4120844602584839,
      "learning_rate": 2.358487181949409e-06,
      "loss": 0.746,
      "step": 9751
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.1031715869903564,
      "learning_rate": 2.3520021662472914e-06,
      "loss": 0.7364,
      "step": 9752
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.261001467704773,
      "learning_rate": 2.3455259724888935e-06,
      "loss": 0.7576,
      "step": 9753
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.0343477725982666,
      "learning_rate": 2.339058601259314e-06,
      "loss": 0.7424,
      "step": 9754
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.056899070739746,
      "learning_rate": 2.3326000531428195e-06,
      "loss": 0.6693,
      "step": 9755
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.9946348071098328,
      "learning_rate": 2.3261503287229313e-06,
      "loss": 0.7847,
      "step": 9756
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.0172147750854492,
      "learning_rate": 2.3197094285823283e-06,
      "loss": 0.5536,
      "step": 9757
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.0063433647155762,
      "learning_rate": 2.3132773533029115e-06,
      "loss": 0.593,
      "step": 9758
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.2520272731781006,
      "learning_rate": 2.3068541034658054e-06,
      "loss": 0.841,
      "step": 9759
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.9585475325584412,
      "learning_rate": 2.300439679651267e-06,
      "loss": 0.728,
      "step": 9760
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.0840827226638794,
      "learning_rate": 2.294034082438834e-06,
      "loss": 0.7269,
      "step": 9761
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.0989664793014526,
      "learning_rate": 2.28763731240722e-06,
      "loss": 0.712,
      "step": 9762
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.0484001636505127,
      "learning_rate": 2.2812493701343305e-06,
      "loss": 0.6376,
      "step": 9763
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.2109870910644531,
      "learning_rate": 2.2748702561972813e-06,
      "loss": 0.6331,
      "step": 9764
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.0695486068725586,
      "learning_rate": 2.2684999711723665e-06,
      "loss": 0.6458,
      "step": 9765
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.1623964309692383,
      "learning_rate": 2.262138515635148e-06,
      "loss": 0.6697,
      "step": 9766
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.1807613372802734,
      "learning_rate": 2.255785890160311e-06,
      "loss": 0.7058,
      "step": 9767
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.176129937171936,
      "learning_rate": 2.2494420953217966e-06,
      "loss": 0.7181,
      "step": 9768
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.125860571861267,
      "learning_rate": 2.2431071316927344e-06,
      "loss": 0.6867,
      "step": 9769
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.066559910774231,
      "learning_rate": 2.2367809998454335e-06,
      "loss": 0.723,
      "step": 9770
    },
    {
      "epoch": 3.67,
      "grad_norm": 1.0393154621124268,
      "learning_rate": 2.230463700351426e-06,
      "loss": 0.636,
      "step": 9771
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.9594808220863342,
      "learning_rate": 2.2241552337814776e-06,
      "loss": 0.5267,
      "step": 9772
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.030861735343933,
      "learning_rate": 2.2178556007054872e-06,
      "loss": 0.7655,
      "step": 9773
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.0564464330673218,
      "learning_rate": 2.2115648016926005e-06,
      "loss": 0.8195,
      "step": 9774
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.05347740650177,
      "learning_rate": 2.2052828373111624e-06,
      "loss": 0.6505,
      "step": 9775
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.182479739189148,
      "learning_rate": 2.1990097081286966e-06,
      "loss": 0.586,
      "step": 9776
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.034058690071106,
      "learning_rate": 2.192745414711972e-06,
      "loss": 0.6954,
      "step": 9777
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.06488037109375,
      "learning_rate": 2.1864899576269027e-06,
      "loss": 0.6989,
      "step": 9778
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.050141453742981,
      "learning_rate": 2.1802433374386588e-06,
      "loss": 0.7977,
      "step": 9779
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.0775222778320312,
      "learning_rate": 2.174005554711567e-06,
      "loss": 0.7656,
      "step": 9780
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.1563800573349,
      "learning_rate": 2.167776610009187e-06,
      "loss": 0.6748,
      "step": 9781
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.9925518035888672,
      "learning_rate": 2.1615565038942576e-06,
      "loss": 0.6546,
      "step": 9782
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.0256890058517456,
      "learning_rate": 2.155345236928741e-06,
      "loss": 0.7076,
      "step": 9783
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.2011566162109375,
      "learning_rate": 2.1491428096737987e-06,
      "loss": 0.7071,
      "step": 9784
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.0400447845458984,
      "learning_rate": 2.142949222689761e-06,
      "loss": 0.8383,
      "step": 9785
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.009543776512146,
      "learning_rate": 2.1367644765362014e-06,
      "loss": 0.623,
      "step": 9786
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.0223863124847412,
      "learning_rate": 2.130588571771874e-06,
      "loss": 0.7282,
      "step": 9787
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.3726271390914917,
      "learning_rate": 2.1244215089547324e-06,
      "loss": 0.7702,
      "step": 9788
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.0192673206329346,
      "learning_rate": 2.1182632886419306e-06,
      "loss": 0.7814,
      "step": 9789
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.0487014055252075,
      "learning_rate": 2.112113911389835e-06,
      "loss": 0.7729,
      "step": 9790
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.9788785576820374,
      "learning_rate": 2.1059733777540225e-06,
      "loss": 0.8045,
      "step": 9791
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.122594952583313,
      "learning_rate": 2.099841688289228e-06,
      "loss": 0.7047,
      "step": 9792
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.1299375295639038,
      "learning_rate": 2.0937188435494303e-06,
      "loss": 0.6296,
      "step": 9793
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.05686354637146,
      "learning_rate": 2.0876048440877983e-06,
      "loss": 0.7389,
      "step": 9794
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.2119766473770142,
      "learning_rate": 2.0814996904566785e-06,
      "loss": 0.5981,
      "step": 9795
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.003216028213501,
      "learning_rate": 2.0754033832076524e-06,
      "loss": 0.6262,
      "step": 9796
    },
    {
      "epoch": 3.68,
      "grad_norm": 1.047455072402954,
      "learning_rate": 2.0693159228914905e-06,
      "loss": 0.8586,
      "step": 9797
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.9886437654495239,
      "learning_rate": 2.0632373100581527e-06,
      "loss": 0.6327,
      "step": 9798
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.225093960762024,
      "learning_rate": 2.0571675452567995e-06,
      "loss": 0.7913,
      "step": 9799
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.1020965576171875,
      "learning_rate": 2.051106629035804e-06,
      "loss": 0.6468,
      "step": 9800
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.1290167570114136,
      "learning_rate": 2.0450545619427497e-06,
      "loss": 0.7041,
      "step": 9801
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.9512486457824707,
      "learning_rate": 2.0390113445243887e-06,
      "loss": 0.64,
      "step": 9802
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.0297565460205078,
      "learning_rate": 2.032976977326706e-06,
      "loss": 0.7239,
      "step": 9803
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.101706624031067,
      "learning_rate": 2.0269514608948657e-06,
      "loss": 0.6604,
      "step": 9804
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.0368657112121582,
      "learning_rate": 2.0209347957732328e-06,
      "loss": 0.7448,
      "step": 9805
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.1157509088516235,
      "learning_rate": 2.014926982505383e-06,
      "loss": 0.6636,
      "step": 9806
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.0000903606414795,
      "learning_rate": 2.0089280216341044e-06,
      "loss": 0.7488,
      "step": 9807
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.113412618637085,
      "learning_rate": 2.0029379137013417e-06,
      "loss": 0.7886,
      "step": 9808
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.3572226762771606,
      "learning_rate": 1.9969566592482837e-06,
      "loss": 0.6789,
      "step": 9809
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.022551417350769,
      "learning_rate": 1.9909842588153094e-06,
      "loss": 0.7222,
      "step": 9810
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.076275110244751,
      "learning_rate": 1.9850207129419652e-06,
      "loss": 0.716,
      "step": 9811
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.9679408073425293,
      "learning_rate": 1.979066022167042e-06,
      "loss": 0.6808,
      "step": 9812
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.085477352142334,
      "learning_rate": 1.973120187028521e-06,
      "loss": 0.8114,
      "step": 9813
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.0572433471679688,
      "learning_rate": 1.967183208063561e-06,
      "loss": 0.6458,
      "step": 9814
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.0551872253417969,
      "learning_rate": 1.9612550858085334e-06,
      "loss": 0.8278,
      "step": 9815
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.154021143913269,
      "learning_rate": 1.9553358207990202e-06,
      "loss": 0.734,
      "step": 9816
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.0145132541656494,
      "learning_rate": 1.949425413569794e-06,
      "loss": 0.7317,
      "step": 9817
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.088685154914856,
      "learning_rate": 1.943523864654806e-06,
      "loss": 0.9291,
      "step": 9818
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.2380071878433228,
      "learning_rate": 1.937631174587251e-06,
      "loss": 0.7687,
      "step": 9819
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.016995906829834,
      "learning_rate": 1.9317473438994926e-06,
      "loss": 0.7977,
      "step": 9820
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.9818235635757446,
      "learning_rate": 1.9258723731230943e-06,
      "loss": 0.8036,
      "step": 9821
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.215066909790039,
      "learning_rate": 1.9200062627888425e-06,
      "loss": 0.8425,
      "step": 9822
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.9307699203491211,
      "learning_rate": 1.9141490134266914e-06,
      "loss": 0.5827,
      "step": 9823
    },
    {
      "epoch": 3.69,
      "grad_norm": 1.0436304807662964,
      "learning_rate": 1.9083006255658176e-06,
      "loss": 0.8036,
      "step": 9824
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.119469404220581,
      "learning_rate": 1.902461099734587e-06,
      "loss": 0.6502,
      "step": 9825
    },
    {
      "epoch": 3.7,
      "eval_loss": 1.0305216312408447,
      "eval_runtime": 17.7871,
      "eval_samples_per_second": 153.426,
      "eval_steps_per_second": 38.399,
      "step": 9825
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.0616732835769653,
      "learning_rate": 1.896630436460567e-06,
      "loss": 0.7554,
      "step": 9826
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.9883622527122498,
      "learning_rate": 1.8908086362705357e-06,
      "loss": 0.7012,
      "step": 9827
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.1083694696426392,
      "learning_rate": 1.8849956996904393e-06,
      "loss": 0.6807,
      "step": 9828
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.163120985031128,
      "learning_rate": 1.879191627245469e-06,
      "loss": 0.6681,
      "step": 9829
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.43439781665802,
      "learning_rate": 1.8733964194599719e-06,
      "loss": 0.6516,
      "step": 9830
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.174665927886963,
      "learning_rate": 1.8676100768575068e-06,
      "loss": 1.0256,
      "step": 9831
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.2440253496170044,
      "learning_rate": 1.8618325999608554e-06,
      "loss": 0.5819,
      "step": 9832
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.031128168106079,
      "learning_rate": 1.8560639892919784e-06,
      "loss": 0.6346,
      "step": 9833
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.12943434715271,
      "learning_rate": 1.8503042453720255e-06,
      "loss": 0.7526,
      "step": 9834
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.2058182954788208,
      "learning_rate": 1.8445533687213467e-06,
      "loss": 0.7516,
      "step": 9835
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.9837791919708252,
      "learning_rate": 1.838811359859538e-06,
      "loss": 0.5414,
      "step": 9836
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.0371630191802979,
      "learning_rate": 1.8330782193053286e-06,
      "loss": 0.829,
      "step": 9837
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.1439690589904785,
      "learning_rate": 1.827353947576671e-06,
      "loss": 0.8186,
      "step": 9838
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.1212270259857178,
      "learning_rate": 1.8216385451907624e-06,
      "loss": 0.686,
      "step": 9839
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.181603193283081,
      "learning_rate": 1.8159320126639012e-06,
      "loss": 0.8001,
      "step": 9840
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.0484421253204346,
      "learning_rate": 1.8102343505116637e-06,
      "loss": 0.808,
      "step": 9841
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.9749422669410706,
      "learning_rate": 1.8045455592488157e-06,
      "loss": 0.6415,
      "step": 9842
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.0475901365280151,
      "learning_rate": 1.798865639389291e-06,
      "loss": 0.7469,
      "step": 9843
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.062718152999878,
      "learning_rate": 1.7931945914462456e-06,
      "loss": 0.7571,
      "step": 9844
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.0805530548095703,
      "learning_rate": 1.787532415932025e-06,
      "loss": 0.7172,
      "step": 9845
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.1658021211624146,
      "learning_rate": 1.7818791133581647e-06,
      "loss": 0.7899,
      "step": 9846
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.8577368855476379,
      "learning_rate": 1.7762346842354227e-06,
      "loss": 0.6481,
      "step": 9847
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.091615915298462,
      "learning_rate": 1.7705991290737357e-06,
      "loss": 0.8293,
      "step": 9848
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.1081324815750122,
      "learning_rate": 1.7649724483822406e-06,
      "loss": 0.7051,
      "step": 9849
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.0685514211654663,
      "learning_rate": 1.7593546426692753e-06,
      "loss": 0.6811,
      "step": 9850
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.055387258529663,
      "learning_rate": 1.7537457124423895e-06,
      "loss": 0.6174,
      "step": 9851
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.3414644002914429,
      "learning_rate": 1.748145658208311e-06,
      "loss": 0.865,
      "step": 9852
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.004217505455017,
      "learning_rate": 1.7425544804729577e-06,
      "loss": 0.7416,
      "step": 9853
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.87591552734375,
      "learning_rate": 1.7369721797414918e-06,
      "loss": 0.4698,
      "step": 9854
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.1836233139038086,
      "learning_rate": 1.731398756518221e-06,
      "loss": 0.7765,
      "step": 9855
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.1792932748794556,
      "learning_rate": 1.725834211306665e-06,
      "loss": 0.6316,
      "step": 9856
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0358437299728394,
      "learning_rate": 1.720278544609577e-06,
      "loss": 0.6868,
      "step": 9857
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.1626344919204712,
      "learning_rate": 1.7147317569288558e-06,
      "loss": 0.714,
      "step": 9858
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0566418170928955,
      "learning_rate": 1.7091938487656333e-06,
      "loss": 0.8074,
      "step": 9859
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0773680210113525,
      "learning_rate": 1.703664820620221e-06,
      "loss": 0.7237,
      "step": 9860
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0311118364334106,
      "learning_rate": 1.6981446729921525e-06,
      "loss": 0.7166,
      "step": 9861
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0613900423049927,
      "learning_rate": 1.6926334063801174e-06,
      "loss": 0.6788,
      "step": 9862
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.075157642364502,
      "learning_rate": 1.68713102128204e-06,
      "loss": 0.8654,
      "step": 9863
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0478132963180542,
      "learning_rate": 1.6816375181950338e-06,
      "loss": 0.7228,
      "step": 9864
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.047194242477417,
      "learning_rate": 1.6761528976154017e-06,
      "loss": 0.7765,
      "step": 9865
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0373905897140503,
      "learning_rate": 1.6706771600386362e-06,
      "loss": 0.7134,
      "step": 9866
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.026845932006836,
      "learning_rate": 1.6652103059594636e-06,
      "loss": 0.7014,
      "step": 9867
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.073614478111267,
      "learning_rate": 1.659752335871756e-06,
      "loss": 0.7248,
      "step": 9868
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0079059600830078,
      "learning_rate": 1.6543032502686406e-06,
      "loss": 0.752,
      "step": 9869
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0139681100845337,
      "learning_rate": 1.6488630496423795e-06,
      "loss": 0.6033,
      "step": 9870
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.9866427183151245,
      "learning_rate": 1.6434317344844907e-06,
      "loss": 0.6985,
      "step": 9871
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0392417907714844,
      "learning_rate": 1.6380093052856483e-06,
      "loss": 0.7892,
      "step": 9872
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.2023638486862183,
      "learning_rate": 1.6325957625357269e-06,
      "loss": 0.877,
      "step": 9873
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.0018748044967651,
      "learning_rate": 1.6271911067238466e-06,
      "loss": 0.7182,
      "step": 9874
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.1125253438949585,
      "learning_rate": 1.62179533833825e-06,
      "loss": 0.6679,
      "step": 9875
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.1425950527191162,
      "learning_rate": 1.616408457866425e-06,
      "loss": 0.8733,
      "step": 9876
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.1069751977920532,
      "learning_rate": 1.6110304657950715e-06,
      "loss": 0.8228,
      "step": 9877
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.9907463788986206,
      "learning_rate": 1.6056613626100115e-06,
      "loss": 0.6237,
      "step": 9878
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.3451051712036133,
      "learning_rate": 1.600301148796357e-06,
      "loss": 0.7693,
      "step": 9879
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.9863184094429016,
      "learning_rate": 1.5949498248383543e-06,
      "loss": 0.6678,
      "step": 9880
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0487309694290161,
      "learning_rate": 1.589607391219461e-06,
      "loss": 0.6823,
      "step": 9881
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0370200872421265,
      "learning_rate": 1.5842738484223463e-06,
      "loss": 0.7194,
      "step": 9882
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.1077667474746704,
      "learning_rate": 1.5789491969288584e-06,
      "loss": 0.7825,
      "step": 9883
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.003806233406067,
      "learning_rate": 1.5736334372200567e-06,
      "loss": 0.8864,
      "step": 9884
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.033172845840454,
      "learning_rate": 1.5683265697761685e-06,
      "loss": 0.6308,
      "step": 9885
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0874158143997192,
      "learning_rate": 1.5630285950766653e-06,
      "loss": 0.8135,
      "step": 9886
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0353286266326904,
      "learning_rate": 1.5577395136001982e-06,
      "loss": 0.7031,
      "step": 9887
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.1358810663223267,
      "learning_rate": 1.5524593258245623e-06,
      "loss": 0.7479,
      "step": 9888
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0765749216079712,
      "learning_rate": 1.5471880322268318e-06,
      "loss": 0.8234,
      "step": 9889
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.2230089902877808,
      "learning_rate": 1.5419256332832255e-06,
      "loss": 0.8136,
      "step": 9890
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0148102045059204,
      "learning_rate": 1.5366721294691633e-06,
      "loss": 0.7216,
      "step": 9891
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0898252725601196,
      "learning_rate": 1.5314275212592766e-06,
      "loss": 0.7496,
      "step": 9892
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.073722243309021,
      "learning_rate": 1.5261918091273863e-06,
      "loss": 0.7853,
      "step": 9893
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0187586545944214,
      "learning_rate": 1.5209649935465142e-06,
      "loss": 0.7287,
      "step": 9894
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.1332582235336304,
      "learning_rate": 1.5157470749888602e-06,
      "loss": 0.7144,
      "step": 9895
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.2716000080108643,
      "learning_rate": 1.5105380539258473e-06,
      "loss": 0.8695,
      "step": 9896
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0987699031829834,
      "learning_rate": 1.5053379308280767e-06,
      "loss": 0.8239,
      "step": 9897
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0365173816680908,
      "learning_rate": 1.5001467061653396e-06,
      "loss": 0.7576,
      "step": 9898
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0827094316482544,
      "learning_rate": 1.4949643804066493e-06,
      "loss": 0.8439,
      "step": 9899
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0519158840179443,
      "learning_rate": 1.4897909540201983e-06,
      "loss": 0.6548,
      "step": 9900
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.940477728843689,
      "learning_rate": 1.4846264274733567e-06,
      "loss": 0.6815,
      "step": 9901
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.0034369230270386,
      "learning_rate": 1.4794708012327408e-06,
      "loss": 0.6825,
      "step": 9902
    },
    {
      "epoch": 3.72,
      "grad_norm": 1.1498050689697266,
      "learning_rate": 1.4743240757641107e-06,
      "loss": 0.6989,
      "step": 9903
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.1237034797668457,
      "learning_rate": 1.4691862515324507e-06,
      "loss": 0.6337,
      "step": 9904
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.9891869425773621,
      "learning_rate": 1.4640573290019221e-06,
      "loss": 0.7337,
      "step": 9905
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0898044109344482,
      "learning_rate": 1.4589373086359103e-06,
      "loss": 0.5933,
      "step": 9906
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0346795320510864,
      "learning_rate": 1.4538261908969785e-06,
      "loss": 0.8823,
      "step": 9907
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.9823195338249207,
      "learning_rate": 1.4487239762468795e-06,
      "loss": 0.7665,
      "step": 9908
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.9464462399482727,
      "learning_rate": 1.4436306651465892e-06,
      "loss": 0.5784,
      "step": 9909
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.011700987815857,
      "learning_rate": 1.4385462580562281e-06,
      "loss": 0.6923,
      "step": 9910
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0684466361999512,
      "learning_rate": 1.4334707554351511e-06,
      "loss": 0.7418,
      "step": 9911
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.9975201487541199,
      "learning_rate": 1.428404157741925e-06,
      "loss": 0.7402,
      "step": 9912
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0804088115692139,
      "learning_rate": 1.423346465434261e-06,
      "loss": 0.8455,
      "step": 9913
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0910948514938354,
      "learning_rate": 1.4182976789691161e-06,
      "loss": 0.7915,
      "step": 9914
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0968644618988037,
      "learning_rate": 1.413257798802592e-06,
      "loss": 0.7913,
      "step": 9915
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0338205099105835,
      "learning_rate": 1.4082268253900465e-06,
      "loss": 0.5617,
      "step": 9916
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0938935279846191,
      "learning_rate": 1.403204759185972e-06,
      "loss": 0.727,
      "step": 9917
    },
    {
      "epoch": 3.73,
      "grad_norm": Infinity,
      "learning_rate": 1.403204759185972e-06,
      "loss": 0.7515,
      "step": 9918
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0785713195800781,
      "learning_rate": 1.3981916006440943e-06,
      "loss": 0.672,
      "step": 9919
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.2548428773880005,
      "learning_rate": 1.3931873502173399e-06,
      "loss": 0.7444,
      "step": 9920
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.9652256369590759,
      "learning_rate": 1.3881920083577805e-06,
      "loss": 0.6885,
      "step": 9921
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.1881190538406372,
      "learning_rate": 1.3832055755167328e-06,
      "loss": 0.6526,
      "step": 9922
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.2623116970062256,
      "learning_rate": 1.3782280521447032e-06,
      "loss": 0.7004,
      "step": 9923
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0638874769210815,
      "learning_rate": 1.3732594386913655e-06,
      "loss": 0.7044,
      "step": 9924
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.9064394235610962,
      "learning_rate": 1.3682997356056272e-06,
      "loss": 0.5762,
      "step": 9925
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.9041042327880859,
      "learning_rate": 1.3633489433355406e-06,
      "loss": 0.6643,
      "step": 9926
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0490760803222656,
      "learning_rate": 1.358407062328404e-06,
      "loss": 0.6434,
      "step": 9927
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0887682437896729,
      "learning_rate": 1.3534740930306933e-06,
      "loss": 0.6957,
      "step": 9928
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.0943554639816284,
      "learning_rate": 1.348550035888052e-06,
      "loss": 0.6722,
      "step": 9929
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.1318111419677734,
      "learning_rate": 1.3436348913453578e-06,
      "loss": 0.7196,
      "step": 9930
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.2489521503448486,
      "learning_rate": 1.3387286598466442e-06,
      "loss": 0.744,
      "step": 9931
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.0668063163757324,
      "learning_rate": 1.333831341835201e-06,
      "loss": 0.6967,
      "step": 9932
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.1455309391021729,
      "learning_rate": 1.3289429377534302e-06,
      "loss": 0.6601,
      "step": 9933
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.1283488273620605,
      "learning_rate": 1.3240634480430004e-06,
      "loss": 0.6763,
      "step": 9934
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.2648805379867554,
      "learning_rate": 1.3191928731447366e-06,
      "loss": 0.7691,
      "step": 9935
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.953333854675293,
      "learning_rate": 1.3143312134986651e-06,
      "loss": 0.6682,
      "step": 9936
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.112151026725769,
      "learning_rate": 1.309478469544012e-06,
      "loss": 0.738,
      "step": 9937
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.061240315437317,
      "learning_rate": 1.3046346417191934e-06,
      "loss": 0.6828,
      "step": 9938
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.3843162059783936,
      "learning_rate": 1.299799730461826e-06,
      "loss": 0.8102,
      "step": 9939
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.2334059476852417,
      "learning_rate": 1.2949737362087156e-06,
      "loss": 0.7554,
      "step": 9940
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.0488616228103638,
      "learning_rate": 1.290156659395858e-06,
      "loss": 0.5982,
      "step": 9941
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.2437196969985962,
      "learning_rate": 1.2853485004584498e-06,
      "loss": 0.8281,
      "step": 9942
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.0276274681091309,
      "learning_rate": 1.2805492598308988e-06,
      "loss": 0.732,
      "step": 9943
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.1245017051696777,
      "learning_rate": 1.2757589379467583e-06,
      "loss": 0.8529,
      "step": 9944
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.2529852390289307,
      "learning_rate": 1.2709775352388264e-06,
      "loss": 0.8017,
      "step": 9945
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.0180026292800903,
      "learning_rate": 1.266205052139069e-06,
      "loss": 0.6793,
      "step": 9946
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.0648458003997803,
      "learning_rate": 1.261441489078663e-06,
      "loss": 0.7155,
      "step": 9947
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.0922154188156128,
      "learning_rate": 1.2566868464879533e-06,
      "loss": 0.6721,
      "step": 9948
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.0671385526657104,
      "learning_rate": 1.2519411247965186e-06,
      "loss": 0.6979,
      "step": 9949
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.9098845720291138,
      "learning_rate": 1.2472043244330821e-06,
      "loss": 0.5204,
      "step": 9950
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.0716118812561035,
      "learning_rate": 1.2424764458255911e-06,
      "loss": 0.7402,
      "step": 9951
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.9720531105995178,
      "learning_rate": 1.2377574894012034e-06,
      "loss": 0.7371,
      "step": 9952
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.1392651796340942,
      "learning_rate": 1.2330474555862336e-06,
      "loss": 0.6893,
      "step": 9953
    },
    {
      "epoch": 3.74,
      "grad_norm": 1.0593737363815308,
      "learning_rate": 1.2283463448062082e-06,
      "loss": 0.8411,
      "step": 9954
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.9101545214653015,
      "learning_rate": 1.223654157485854e-06,
      "loss": 0.6983,
      "step": 9955
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.1066205501556396,
      "learning_rate": 1.2189708940490652e-06,
      "loss": 0.7262,
      "step": 9956
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.964721143245697,
      "learning_rate": 1.2142965549189477e-06,
      "loss": 0.664,
      "step": 9957
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.150708556175232,
      "learning_rate": 1.2096311405178418e-06,
      "loss": 0.7718,
      "step": 9958
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.1257002353668213,
      "learning_rate": 1.2049746512671988e-06,
      "loss": 0.84,
      "step": 9959
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.9935750365257263,
      "learning_rate": 1.200327087587716e-06,
      "loss": 0.698,
      "step": 9960
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.9364817142486572,
      "learning_rate": 1.1956884498992904e-06,
      "loss": 0.7762,
      "step": 9961
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.2192072868347168,
      "learning_rate": 1.1910587386209649e-06,
      "loss": 0.6588,
      "step": 9962
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.149577260017395,
      "learning_rate": 1.186437954171049e-06,
      "loss": 0.698,
      "step": 9963
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.9917477965354919,
      "learning_rate": 1.1818260969669758e-06,
      "loss": 0.7615,
      "step": 9964
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.049358606338501,
      "learning_rate": 1.177223167425401e-06,
      "loss": 0.6534,
      "step": 9965
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.115843653678894,
      "learning_rate": 1.1726291659621801e-06,
      "loss": 0.8461,
      "step": 9966
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.0291237831115723,
      "learning_rate": 1.1680440929923708e-06,
      "loss": 0.7658,
      "step": 9967
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.0941379070281982,
      "learning_rate": 1.1634679489301747e-06,
      "loss": 0.6997,
      "step": 9968
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.1664254665374756,
      "learning_rate": 1.158900734189039e-06,
      "loss": 0.8589,
      "step": 9969
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.0937172174453735,
      "learning_rate": 1.1543424491816e-06,
      "loss": 0.7246,
      "step": 9970
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.2768316268920898,
      "learning_rate": 1.149793094319651e-06,
      "loss": 0.8031,
      "step": 9971
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.024124264717102,
      "learning_rate": 1.1452526700141964e-06,
      "loss": 0.6788,
      "step": 9972
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.0567468404769897,
      "learning_rate": 1.1407211766754633e-06,
      "loss": 0.766,
      "step": 9973
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.0547617673873901,
      "learning_rate": 1.1361986147128357e-06,
      "loss": 0.8209,
      "step": 9974
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.0394614934921265,
      "learning_rate": 1.1316849845348865e-06,
      "loss": 0.5793,
      "step": 9975
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.245154857635498,
      "learning_rate": 1.1271802865494009e-06,
      "loss": 0.8175,
      "step": 9976
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.2085556983947754,
      "learning_rate": 1.1226845211633752e-06,
      "loss": 0.8839,
      "step": 9977
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.115144968032837,
      "learning_rate": 1.1181976887829626e-06,
      "loss": 0.682,
      "step": 9978
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.0983599424362183,
      "learning_rate": 1.1137197898135054e-06,
      "loss": 0.6947,
      "step": 9979
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.059057354927063,
      "learning_rate": 1.1092508246595912e-06,
      "loss": 0.7148,
      "step": 9980
    },
    {
      "epoch": 3.75,
      "grad_norm": 1.0948294401168823,
      "learning_rate": 1.1047907937249414e-06,
      "loss": 0.6809,
      "step": 9981
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.9911684393882751,
      "learning_rate": 1.1003396974124891e-06,
      "loss": 0.6423,
      "step": 9982
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.0742179155349731,
      "learning_rate": 1.0958975361243795e-06,
      "loss": 0.7205,
      "step": 9983
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.1334620714187622,
      "learning_rate": 1.091464310261947e-06,
      "loss": 0.6938,
      "step": 9984
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.2088441848754883,
      "learning_rate": 1.0870400202256824e-06,
      "loss": 0.6655,
      "step": 9985
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.9839671850204468,
      "learning_rate": 1.0826246664153106e-06,
      "loss": 0.6495,
      "step": 9986
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.1840052604675293,
      "learning_rate": 1.0782182492297343e-06,
      "loss": 0.774,
      "step": 9987
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.0975052118301392,
      "learning_rate": 1.0738207690670577e-06,
      "loss": 0.6852,
      "step": 9988
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.1456769704818726,
      "learning_rate": 1.0694322263245293e-06,
      "loss": 0.6926,
      "step": 9989
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.2086966037750244,
      "learning_rate": 1.0650526213986878e-06,
      "loss": 0.8795,
      "step": 9990
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.0199048519134521,
      "learning_rate": 1.06068195468515e-06,
      "loss": 0.6598,
      "step": 9991
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.1541393995285034,
      "learning_rate": 1.0563202265787997e-06,
      "loss": 0.7129,
      "step": 9992
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.0221788883209229,
      "learning_rate": 1.051967437473722e-06,
      "loss": 0.8571,
      "step": 9993
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.227849006652832,
      "learning_rate": 1.0476235877631246e-06,
      "loss": 0.6363,
      "step": 9994
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.9209998846054077,
      "learning_rate": 1.0432886778394824e-06,
      "loss": 0.6312,
      "step": 9995
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.1917728185653687,
      "learning_rate": 1.0389627080944153e-06,
      "loss": 0.7491,
      "step": 9996
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.1708908081054688,
      "learning_rate": 1.0346456789187441e-06,
      "loss": 0.8331,
      "step": 9997
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.223755955696106,
      "learning_rate": 1.0303375907025015e-06,
      "loss": 0.7869,
      "step": 9998
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.125267744064331,
      "learning_rate": 1.026038443834887e-06,
      "loss": 0.7261,
      "step": 9999
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.2152901887893677,
      "learning_rate": 1.021748238704323e-06,
      "loss": 0.7418,
      "step": 10000
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.087896704673767,
      "learning_rate": 1.017466975698378e-06,
      "loss": 0.623,
      "step": 10001
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.1164805889129639,
      "learning_rate": 1.013194655203853e-06,
      "loss": 0.657,
      "step": 10002
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.1882754564285278,
      "learning_rate": 1.0089312776067395e-06,
      "loss": 0.7242,
      "step": 10003
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.0324262380599976,
      "learning_rate": 1.004676843292196e-06,
      "loss": 0.7467,
      "step": 10004
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.9625290036201477,
      "learning_rate": 1.000431352644582e-06,
      "loss": 0.7747,
      "step": 10005
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.20183265209198,
      "learning_rate": 9.96194806047468e-07,
      "loss": 0.6989,
      "step": 10006
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.1936566829681396,
      "learning_rate": 9.919672038835925e-07,
      "loss": 0.6617,
      "step": 10007
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.0127053260803223,
      "learning_rate": 9.877485465349058e-07,
      "loss": 0.7343,
      "step": 10008
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.0498379468917847,
      "learning_rate": 9.835388343825248e-07,
      "loss": 0.6492,
      "step": 10009
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.0718302726745605,
      "learning_rate": 9.793380678067898e-07,
      "loss": 0.715,
      "step": 10010
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.956362247467041,
      "learning_rate": 9.751462471871864e-07,
      "loss": 0.7218,
      "step": 10011
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.9879927635192871,
      "learning_rate": 9.70963372902467e-07,
      "loss": 0.6621,
      "step": 10012
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.1577439308166504,
      "learning_rate": 9.667894453304959e-07,
      "loss": 0.7443,
      "step": 10013
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.103188395500183,
      "learning_rate": 9.626244648483606e-07,
      "loss": 0.8165,
      "step": 10014
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.1314506530761719,
      "learning_rate": 9.584684318323712e-07,
      "loss": 0.7864,
      "step": 10015
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.9815592765808105,
      "learning_rate": 9.543213466579827e-07,
      "loss": 0.7375,
      "step": 10016
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.1377731561660767,
      "learning_rate": 9.501832096998619e-07,
      "loss": 0.8261,
      "step": 10017
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.268371820449829,
      "learning_rate": 9.460540213318769e-07,
      "loss": 0.75,
      "step": 10018
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.0274016857147217,
      "learning_rate": 9.419337819270735e-07,
      "loss": 0.5949,
      "step": 10019
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.1986650228500366,
      "learning_rate": 9.378224918576872e-07,
      "loss": 0.7825,
      "step": 10020
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.001662254333496,
      "learning_rate": 9.337201514951321e-07,
      "loss": 0.7181,
      "step": 10021
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.1871719360351562,
      "learning_rate": 9.296267612100673e-07,
      "loss": 0.5746,
      "step": 10022
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.2384448051452637,
      "learning_rate": 9.255423213722858e-07,
      "loss": 0.8447,
      "step": 10023
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.106985330581665,
      "learning_rate": 9.214668323507925e-07,
      "loss": 0.7419,
      "step": 10024
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.0780296325683594,
      "learning_rate": 9.174002945137927e-07,
      "loss": 0.8055,
      "step": 10025
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.189407229423523,
      "learning_rate": 9.133427082286705e-07,
      "loss": 0.8046,
      "step": 10026
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.1356770992279053,
      "learning_rate": 9.092940738620104e-07,
      "loss": 0.8084,
      "step": 10027
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.116146445274353,
      "learning_rate": 9.052543917795863e-07,
      "loss": 0.7904,
      "step": 10028
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.9942764043807983,
      "learning_rate": 9.012236623463399e-07,
      "loss": 0.7401,
      "step": 10029
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.9996702671051025,
      "learning_rate": 8.972018859264463e-07,
      "loss": 0.7593,
      "step": 10030
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.9933799505233765,
      "learning_rate": 8.931890628832485e-07,
      "loss": 0.6084,
      "step": 10031
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.1011178493499756,
      "learning_rate": 8.891851935792673e-07,
      "loss": 0.7103,
      "step": 10032
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.1249346733093262,
      "learning_rate": 8.851902783762466e-07,
      "loss": 0.8594,
      "step": 10033
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.0535365343093872,
      "learning_rate": 8.812043176351092e-07,
      "loss": 0.6709,
      "step": 10034
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.9636355638504028,
      "learning_rate": 8.772273117159335e-07,
      "loss": 0.6045,
      "step": 10035
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.9659979939460754,
      "learning_rate": 8.732592609780432e-07,
      "loss": 0.7391,
      "step": 10036
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.0363208055496216,
      "learning_rate": 8.693001657799294e-07,
      "loss": 0.6457,
      "step": 10037
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.9615933299064636,
      "learning_rate": 8.653500264792724e-07,
      "loss": 0.7554,
      "step": 10038
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.9478326439857483,
      "learning_rate": 8.614088434329315e-07,
      "loss": 0.5248,
      "step": 10039
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.050695538520813,
      "learning_rate": 8.574766169969883e-07,
      "loss": 0.7534,
      "step": 10040
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.1325639486312866,
      "learning_rate": 8.535533475266921e-07,
      "loss": 0.7275,
      "step": 10041
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.1841737031936646,
      "learning_rate": 8.496390353764705e-07,
      "loss": 0.6622,
      "step": 10042
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.1159855127334595,
      "learning_rate": 8.45733680899985e-07,
      "loss": 0.7222,
      "step": 10043
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.3739477396011353,
      "learning_rate": 8.418372844500532e-07,
      "loss": 0.7407,
      "step": 10044
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.415931224822998,
      "learning_rate": 8.379498463786828e-07,
      "loss": 0.5787,
      "step": 10045
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.1157495975494385,
      "learning_rate": 8.340713670370925e-07,
      "loss": 0.9139,
      "step": 10046
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.133553147315979,
      "learning_rate": 8.3020184677568e-07,
      "loss": 0.8239,
      "step": 10047
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.2630914449691772,
      "learning_rate": 8.263412859440434e-07,
      "loss": 0.844,
      "step": 10048
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.157052993774414,
      "learning_rate": 8.224896848909369e-07,
      "loss": 0.6671,
      "step": 10049
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.0761855840682983,
      "learning_rate": 8.18647043964349e-07,
      "loss": 0.635,
      "step": 10050
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.0302183628082275,
      "learning_rate": 8.148133635114352e-07,
      "loss": 0.6293,
      "step": 10051
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.0393892526626587,
      "learning_rate": 8.109886438785519e-07,
      "loss": 0.6753,
      "step": 10052
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.0503435134887695,
      "learning_rate": 8.071728854112337e-07,
      "loss": 0.7328,
      "step": 10053
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.1270707845687866,
      "learning_rate": 8.033660884542272e-07,
      "loss": 0.687,
      "step": 10054
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.0003979206085205,
      "learning_rate": 7.995682533514459e-07,
      "loss": 0.6378,
      "step": 10055
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.9889844655990601,
      "learning_rate": 7.957793804459824e-07,
      "loss": 0.7286,
      "step": 10056
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.0894112586975098,
      "learning_rate": 7.919994700801737e-07,
      "loss": 0.718,
      "step": 10057
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.1481521129608154,
      "learning_rate": 7.882285225955022e-07,
      "loss": 0.5825,
      "step": 10058
    },
    {
      "epoch": 3.78,
      "grad_norm": 1.0614179372787476,
      "learning_rate": 7.844665383326399e-07,
      "loss": 0.6882,
      "step": 10059
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.9681883454322815,
      "learning_rate": 7.807135176314706e-07,
      "loss": 0.6247,
      "step": 10060
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.1403526067733765,
      "learning_rate": 7.769694608310562e-07,
      "loss": 0.6777,
      "step": 10061
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.0537508726119995,
      "learning_rate": 7.732343682696486e-07,
      "loss": 0.6247,
      "step": 10062
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.1085972785949707,
      "learning_rate": 7.695082402847109e-07,
      "loss": 0.7137,
      "step": 10063
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.0717341899871826,
      "learning_rate": 7.657910772128407e-07,
      "loss": 0.7903,
      "step": 10064
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.260422706604004,
      "learning_rate": 7.620828793899026e-07,
      "loss": 0.6602,
      "step": 10065
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.1068841218948364,
      "learning_rate": 7.583836471508732e-07,
      "loss": 0.7863,
      "step": 10066
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.9649311304092407,
      "learning_rate": 7.546933808299961e-07,
      "loss": 0.6674,
      "step": 10067
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.2655513286590576,
      "learning_rate": 7.51012080760638e-07,
      "loss": 0.7631,
      "step": 10068
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.047822117805481,
      "learning_rate": 7.473397472753885e-07,
      "loss": 0.7722,
      "step": 10069
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.9567108154296875,
      "learning_rate": 7.436763807060376e-07,
      "loss": 0.7279,
      "step": 10070
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.255197525024414,
      "learning_rate": 7.400219813835208e-07,
      "loss": 0.8808,
      "step": 10071
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.9817781448364258,
      "learning_rate": 7.363765496380182e-07,
      "loss": 0.6891,
      "step": 10072
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.1552553176879883,
      "learning_rate": 7.327400857988776e-07,
      "loss": 0.8054,
      "step": 10073
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.1271185874938965,
      "learning_rate": 7.291125901946027e-07,
      "loss": 0.7244,
      "step": 10074
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.152830719947815,
      "learning_rate": 7.254940631529427e-07,
      "loss": 0.7119,
      "step": 10075
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.0397251844406128,
      "learning_rate": 7.218845050008027e-07,
      "loss": 0.6893,
      "step": 10076
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.0502945184707642,
      "learning_rate": 7.182839160642774e-07,
      "loss": 0.7748,
      "step": 10077
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.1092222929000854,
      "learning_rate": 7.146922966686843e-07,
      "loss": 0.7577,
      "step": 10078
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.1920450925827026,
      "learning_rate": 7.11109647138486e-07,
      "loss": 0.749,
      "step": 10079
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.0592926740646362,
      "learning_rate": 7.075359677973569e-07,
      "loss": 0.7384,
      "step": 10080
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.075227975845337,
      "learning_rate": 7.039712589681502e-07,
      "loss": 0.6723,
      "step": 10081
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.186279296875,
      "learning_rate": 7.004155209729413e-07,
      "loss": 0.8135,
      "step": 10082
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.2132257223129272,
      "learning_rate": 6.96868754132951e-07,
      "loss": 0.7268,
      "step": 10083
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.0569440126419067,
      "learning_rate": 6.93330958768612e-07,
      "loss": 0.7183,
      "step": 10084
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.0761044025421143,
      "learning_rate": 6.898021351995576e-07,
      "loss": 0.8194,
      "step": 10085
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.1437177658081055,
      "learning_rate": 6.862822837445881e-07,
      "loss": 0.6958,
      "step": 10086
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.0625132322311401,
      "learning_rate": 6.827714047216938e-07,
      "loss": 0.7898,
      "step": 10087
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.171156406402588,
      "learning_rate": 6.792694984480652e-07,
      "loss": 0.6345,
      "step": 10088
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.1108416318893433,
      "learning_rate": 6.757765652401049e-07,
      "loss": 0.6583,
      "step": 10089
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.3225679397583008,
      "learning_rate": 6.722926054133494e-07,
      "loss": 0.8477,
      "step": 10090
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.078497290611267,
      "learning_rate": 6.688176192825579e-07,
      "loss": 0.6964,
      "step": 10091
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.0054478645324707,
      "learning_rate": 6.653516071616906e-07,
      "loss": 0.617,
      "step": 10092
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.0786943435668945,
      "learning_rate": 6.618945693638745e-07,
      "loss": 0.7445,
      "step": 10093
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.9267763495445251,
      "learning_rate": 6.584465062014267e-07,
      "loss": 0.5596,
      "step": 10094
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.9944291710853577,
      "learning_rate": 6.550074179858756e-07,
      "loss": 0.6828,
      "step": 10095
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.180585265159607,
      "learning_rate": 6.515773050279061e-07,
      "loss": 0.8386,
      "step": 10096
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.210412621498108,
      "learning_rate": 6.481561676374149e-07,
      "loss": 0.6693,
      "step": 10097
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.1794562339782715,
      "learning_rate": 6.447440061234988e-07,
      "loss": 0.7499,
      "step": 10098
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.9560544490814209,
      "learning_rate": 6.413408207944005e-07,
      "loss": 0.9088,
      "step": 10099
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.1450140476226807,
      "learning_rate": 6.379466119576072e-07,
      "loss": 0.7339,
      "step": 10100
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.9087095260620117,
      "learning_rate": 6.345613799197292e-07,
      "loss": 0.7166,
      "step": 10101
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.0168064832687378,
      "learning_rate": 6.31185124986644e-07,
      "loss": 0.8164,
      "step": 10102
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.2922720909118652,
      "learning_rate": 6.278178474633522e-07,
      "loss": 0.6751,
      "step": 10103
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.1179317235946655,
      "learning_rate": 6.24459547654066e-07,
      "loss": 0.6749,
      "step": 10104
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.0544607639312744,
      "learning_rate": 6.211102258622092e-07,
      "loss": 0.6137,
      "step": 10105
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.1071523427963257,
      "learning_rate": 6.177698823903622e-07,
      "loss": 0.6523,
      "step": 10106
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.0634361505508423,
      "learning_rate": 6.144385175402944e-07,
      "loss": 0.8883,
      "step": 10107
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.9624271392822266,
      "learning_rate": 6.111161316129987e-07,
      "loss": 0.5516,
      "step": 10108
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.1100654602050781,
      "learning_rate": 6.078027249086126e-07,
      "loss": 0.843,
      "step": 10109
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.381820797920227,
      "learning_rate": 6.044982977264968e-07,
      "loss": 0.792,
      "step": 10110
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.124940037727356,
      "learning_rate": 6.012028503651901e-07,
      "loss": 0.7614,
      "step": 10111
    },
    {
      "epoch": 3.8,
      "grad_norm": 1.2110086679458618,
      "learning_rate": 5.979163831223989e-07,
      "loss": 0.7346,
      "step": 10112
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.9798184633255005,
      "learning_rate": 5.946388962950522e-07,
      "loss": 0.7623,
      "step": 10113
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.9927152395248413,
      "learning_rate": 5.913703901792578e-07,
      "loss": 0.6763,
      "step": 10114
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.9982891082763672,
      "learning_rate": 5.881108650702905e-07,
      "loss": 0.7705,
      "step": 10115
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.1986182928085327,
      "learning_rate": 5.84860321262648e-07,
      "loss": 0.5801,
      "step": 10116
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.3353630304336548,
      "learning_rate": 5.816187590499733e-07,
      "loss": 0.719,
      "step": 10117
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.9897480607032776,
      "learning_rate": 5.783861787251543e-07,
      "loss": 0.6961,
      "step": 10118
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.136168360710144,
      "learning_rate": 5.75162580580213e-07,
      "loss": 0.7519,
      "step": 10119
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.1490932703018188,
      "learning_rate": 5.719479649063831e-07,
      "loss": 0.8223,
      "step": 10120
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.061972975730896,
      "learning_rate": 5.687423319941098e-07,
      "loss": 0.6704,
      "step": 10121
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.2456351518630981,
      "learning_rate": 5.65545682132973e-07,
      "loss": 0.7968,
      "step": 10122
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.099271535873413,
      "learning_rate": 5.623580156117969e-07,
      "loss": 0.8464,
      "step": 10123
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.178529977798462,
      "learning_rate": 5.591793327185624e-07,
      "loss": 0.6787,
      "step": 10124
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.0939611196517944,
      "learning_rate": 5.560096337404397e-07,
      "loss": 0.5433,
      "step": 10125
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.1290323734283447,
      "learning_rate": 5.528489189637886e-07,
      "loss": 0.7263,
      "step": 10126
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.3324412107467651,
      "learning_rate": 5.496971886741698e-07,
      "loss": 0.6884,
      "step": 10127
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.134002923965454,
      "learning_rate": 5.46554443156333e-07,
      "loss": 0.7677,
      "step": 10128
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.117805004119873,
      "learning_rate": 5.434206826941956e-07,
      "loss": 0.729,
      "step": 10129
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.215340495109558,
      "learning_rate": 5.402959075708758e-07,
      "loss": 0.6857,
      "step": 10130
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.1562162637710571,
      "learning_rate": 5.371801180686808e-07,
      "loss": 0.7374,
      "step": 10131
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.0804635286331177,
      "learning_rate": 5.340733144691079e-07,
      "loss": 0.7373,
      "step": 10132
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.9803819060325623,
      "learning_rate": 5.309754970528324e-07,
      "loss": 0.7191,
      "step": 10133
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.1646614074707031,
      "learning_rate": 5.278866660997306e-07,
      "loss": 0.8296,
      "step": 10134
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.9612960815429688,
      "learning_rate": 5.24806821888868e-07,
      "loss": 0.6124,
      "step": 10135
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.2418568134307861,
      "learning_rate": 5.217359646984666e-07,
      "loss": 0.7091,
      "step": 10136
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.0681583881378174,
      "learning_rate": 5.186740948059821e-07,
      "loss": 0.6827,
      "step": 10137
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.0703145265579224,
      "learning_rate": 5.156212124880378e-07,
      "loss": 0.7256,
      "step": 10138
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0681973695755005,
      "learning_rate": 5.125773180204352e-07,
      "loss": 0.725,
      "step": 10139
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0785722732543945,
      "learning_rate": 5.095424116781767e-07,
      "loss": 0.7386,
      "step": 10140
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.9934455752372742,
      "learning_rate": 5.065164937354428e-07,
      "loss": 0.6322,
      "step": 10141
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0095465183258057,
      "learning_rate": 5.034995644656038e-07,
      "loss": 0.4397,
      "step": 10142
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0020241737365723,
      "learning_rate": 5.004916241412527e-07,
      "loss": 0.7513,
      "step": 10143
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0242481231689453,
      "learning_rate": 4.974926730341057e-07,
      "loss": 0.7462,
      "step": 10144
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.1092456579208374,
      "learning_rate": 4.945027114151124e-07,
      "loss": 0.7306,
      "step": 10145
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.226828932762146,
      "learning_rate": 4.915217395544125e-07,
      "loss": 0.6395,
      "step": 10146
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.9237508773803711,
      "learning_rate": 4.885497577212905e-07,
      "loss": 0.6469,
      "step": 10147
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.1169458627700806,
      "learning_rate": 4.855867661842762e-07,
      "loss": 0.7881,
      "step": 10148
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.9912639260292053,
      "learning_rate": 4.826327652110552e-07,
      "loss": 0.7744,
      "step": 10149
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.1114553213119507,
      "learning_rate": 4.796877550684919e-07,
      "loss": 0.7971,
      "step": 10150
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0113972425460815,
      "learning_rate": 4.7675173602265144e-07,
      "loss": 0.6636,
      "step": 10151
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.108242154121399,
      "learning_rate": 4.738247083387992e-07,
      "loss": 0.7234,
      "step": 10152
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0709055662155151,
      "learning_rate": 4.709066722813682e-07,
      "loss": 0.7459,
      "step": 10153
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0955545902252197,
      "learning_rate": 4.6799762811398086e-07,
      "loss": 0.8383,
      "step": 10154
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0357388257980347,
      "learning_rate": 4.6509757609946025e-07,
      "loss": 0.6799,
      "step": 10155
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.1340432167053223,
      "learning_rate": 4.622065164998191e-07,
      "loss": 0.6477,
      "step": 10156
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0476224422454834,
      "learning_rate": 4.593244495762261e-07,
      "loss": 0.6442,
      "step": 10157
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.1167852878570557,
      "learning_rate": 4.5645137558907314e-07,
      "loss": 0.7255,
      "step": 10158
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.9429742693901062,
      "learning_rate": 4.535872947979303e-07,
      "loss": 0.5973,
      "step": 10159
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.1268279552459717,
      "learning_rate": 4.507322074615461e-07,
      "loss": 0.7282,
      "step": 10160
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.1311348676681519,
      "learning_rate": 4.478861138378476e-07,
      "loss": 0.8302,
      "step": 10161
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0726604461669922,
      "learning_rate": 4.450490141839847e-07,
      "loss": 0.6861,
      "step": 10162
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.0780662298202515,
      "learning_rate": 4.422209087562634e-07,
      "loss": 0.7843,
      "step": 10163
    },
    {
      "epoch": 3.82,
      "grad_norm": 1.1214207410812378,
      "learning_rate": 4.3940179781019055e-07,
      "loss": 0.7607,
      "step": 10164
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.0474151372909546,
      "learning_rate": 4.3659168160045115e-07,
      "loss": 0.7545,
      "step": 10165
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.1370128393173218,
      "learning_rate": 4.3379056038093114e-07,
      "loss": 0.737,
      "step": 10166
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.1345137357711792,
      "learning_rate": 4.3099843440468356e-07,
      "loss": 0.7327,
      "step": 10167
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.0273923873901367,
      "learning_rate": 4.282153039239845e-07,
      "loss": 0.8406,
      "step": 10168
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.9391980171203613,
      "learning_rate": 4.254411691902438e-07,
      "loss": 0.6662,
      "step": 10169
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.1485267877578735,
      "learning_rate": 4.226760304541166e-07,
      "loss": 0.7608,
      "step": 10170
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.0384386777877808,
      "learning_rate": 4.1991988796539207e-07,
      "loss": 0.7001,
      "step": 10171
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.1486417055130005,
      "learning_rate": 4.1717274197309306e-07,
      "loss": 0.7722,
      "step": 10172
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.167599081993103,
      "learning_rate": 4.14434592725399e-07,
      "loss": 0.7492,
      "step": 10173
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.033146619796753,
      "learning_rate": 4.117054404696785e-07,
      "loss": 0.5247,
      "step": 10174
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.1659486293792725,
      "learning_rate": 4.0898528545252334e-07,
      "loss": 0.6462,
      "step": 10175
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.0949732065200806,
      "learning_rate": 4.06274127919648e-07,
      "loss": 0.7884,
      "step": 10176
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.1966544389724731,
      "learning_rate": 4.0357196811601217e-07,
      "loss": 0.76,
      "step": 10177
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.1527520418167114,
      "learning_rate": 4.008788062857316e-07,
      "loss": 0.8314,
      "step": 10178
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.0645240545272827,
      "learning_rate": 3.9819464267213393e-07,
      "loss": 0.7442,
      "step": 10179
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.0930896997451782,
      "learning_rate": 3.955194775176918e-07,
      "loss": 0.7168,
      "step": 10180
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.3560266494750977,
      "learning_rate": 3.9285331106411194e-07,
      "loss": 0.6589,
      "step": 10181
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.006644606590271,
      "learning_rate": 3.901961435522683e-07,
      "loss": 0.8134,
      "step": 10182
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.2472214698791504,
      "learning_rate": 3.875479752222022e-07,
      "loss": 0.7072,
      "step": 10183
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.9534869194030762,
      "learning_rate": 3.849088063131778e-07,
      "loss": 0.5931,
      "step": 10184
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.0042247772216797,
      "learning_rate": 3.8227863706362666e-07,
      "loss": 0.7421,
      "step": 10185
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.320602297782898,
      "learning_rate": 3.796574677111586e-07,
      "loss": 0.6721,
      "step": 10186
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.0391452312469482,
      "learning_rate": 3.770452984925954e-07,
      "loss": 0.7816,
      "step": 10187
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.9227380156517029,
      "learning_rate": 3.74442129643926e-07,
      "loss": 0.6156,
      "step": 10188
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.0802727937698364,
      "learning_rate": 3.7184796140032895e-07,
      "loss": 0.7806,
      "step": 10189
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.1780742406845093,
      "learning_rate": 3.6926279399617235e-07,
      "loss": 0.9998,
      "step": 10190
    },
    {
      "epoch": 3.83,
      "grad_norm": 1.1809314489364624,
      "learning_rate": 3.6668662766502495e-07,
      "loss": 0.7414,
      "step": 10191
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.1165690422058105,
      "learning_rate": 3.641194626396116e-07,
      "loss": 0.6648,
      "step": 10192
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.1865373849868774,
      "learning_rate": 3.615612991518691e-07,
      "loss": 0.6771,
      "step": 10193
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.9549691677093506,
      "learning_rate": 3.590121374329125e-07,
      "loss": 0.6824,
      "step": 10194
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.066297173500061,
      "learning_rate": 3.564719777130465e-07,
      "loss": 0.6949,
      "step": 10195
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.0364919900894165,
      "learning_rate": 3.5394082022174313e-07,
      "loss": 0.7116,
      "step": 10196
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.9976560473442078,
      "learning_rate": 3.5141866518770825e-07,
      "loss": 0.6577,
      "step": 10197
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.1061553955078125,
      "learning_rate": 3.4890551283878194e-07,
      "loss": 0.796,
      "step": 10198
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.1360870599746704,
      "learning_rate": 3.464013634020158e-07,
      "loss": 0.8133,
      "step": 10199
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.9274671673774719,
      "learning_rate": 3.439062171036511e-07,
      "loss": 0.7211,
      "step": 10200
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.0013492107391357,
      "learning_rate": 3.414200741691187e-07,
      "loss": 0.6565,
      "step": 10201
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.0790327787399292,
      "learning_rate": 3.389429348230055e-07,
      "loss": 0.7722,
      "step": 10202
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.0717904567718506,
      "learning_rate": 3.364747992891215e-07,
      "loss": 0.8373,
      "step": 10203
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.2947211265563965,
      "learning_rate": 3.340156677904438e-07,
      "loss": 0.7712,
      "step": 10204
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.0043305158615112,
      "learning_rate": 3.315655405491502e-07,
      "loss": 0.7439,
      "step": 10205
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.9222303032875061,
      "learning_rate": 3.29124417786586e-07,
      "loss": 0.5299,
      "step": 10206
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.035064458847046,
      "learning_rate": 3.2669229972329684e-07,
      "loss": 0.6043,
      "step": 10207
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.1197693347930908,
      "learning_rate": 3.2426918657900704e-07,
      "loss": 0.7642,
      "step": 10208
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.1531667709350586,
      "learning_rate": 3.218550785726304e-07,
      "loss": 0.8072,
      "step": 10209
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.1412495374679565,
      "learning_rate": 3.194499759222813e-07,
      "loss": 0.6041,
      "step": 10210
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.9792382121086121,
      "learning_rate": 3.170538788452193e-07,
      "loss": 0.6598,
      "step": 10211
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.2381423711776733,
      "learning_rate": 3.14666787557949e-07,
      "loss": 0.7815,
      "step": 10212
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.9339486956596375,
      "learning_rate": 3.1228870227612004e-07,
      "loss": 0.7263,
      "step": 10213
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.1330033540725708,
      "learning_rate": 3.099196232145718e-07,
      "loss": 0.8084,
      "step": 10214
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.0861912965774536,
      "learning_rate": 3.0755955058734407e-07,
      "loss": 0.7309,
      "step": 10215
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.0379889011383057,
      "learning_rate": 3.0520848460765527e-07,
      "loss": 0.64,
      "step": 10216
    },
    {
      "epoch": 3.84,
      "grad_norm": 1.0546818971633911,
      "learning_rate": 3.028664254879021e-07,
      "loss": 0.5524,
      "step": 10217
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.991212785243988,
      "learning_rate": 3.00533373439682e-07,
      "loss": 0.6809,
      "step": 10218
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.0082114934921265,
      "learning_rate": 2.982093286737819e-07,
      "loss": 0.7526,
      "step": 10219
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.1457719802856445,
      "learning_rate": 2.9589429140015614e-07,
      "loss": 0.5842,
      "step": 10220
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.0047378540039062,
      "learning_rate": 2.9358826182795953e-07,
      "loss": 0.5792,
      "step": 10221
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.9974850416183472,
      "learning_rate": 2.9129124016552544e-07,
      "loss": 0.6634,
      "step": 10222
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.4352089166641235,
      "learning_rate": 2.890032266203879e-07,
      "loss": 0.7492,
      "step": 10223
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.0173624753952026,
      "learning_rate": 2.8672422139923715e-07,
      "loss": 0.8202,
      "step": 10224
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.2964680194854736,
      "learning_rate": 2.8445422470797513e-07,
      "loss": 0.6294,
      "step": 10225
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.1161788702011108,
      "learning_rate": 2.821932367516933e-07,
      "loss": 0.8585,
      "step": 10226
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.011875867843628,
      "learning_rate": 2.7994125773465053e-07,
      "loss": 0.9305,
      "step": 10227
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.1668351888656616,
      "learning_rate": 2.776982878602952e-07,
      "loss": 0.6589,
      "step": 10228
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.0270957946777344,
      "learning_rate": 2.754643273312874e-07,
      "loss": 0.7205,
      "step": 10229
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.2454346418380737,
      "learning_rate": 2.732393763494323e-07,
      "loss": 0.806,
      "step": 10230
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.2036510705947876,
      "learning_rate": 2.710234351157359e-07,
      "loss": 0.7665,
      "step": 10231
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.2232073545455933,
      "learning_rate": 2.688165038304269e-07,
      "loss": 0.8363,
      "step": 10232
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.194474220275879,
      "learning_rate": 2.666185826928569e-07,
      "loss": 0.7197,
      "step": 10233
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.2119613885879517,
      "learning_rate": 2.644296719016115e-07,
      "loss": 0.6024,
      "step": 10234
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.0143555402755737,
      "learning_rate": 2.622497716544436e-07,
      "loss": 0.7342,
      "step": 10235
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.0222737789154053,
      "learning_rate": 2.600788821483069e-07,
      "loss": 0.7917,
      "step": 10236
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.0858430862426758,
      "learning_rate": 2.579170035792999e-07,
      "loss": 0.688,
      "step": 10237
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.0466959476470947,
      "learning_rate": 2.5576413614276653e-07,
      "loss": 0.7364,
      "step": 10238
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.9896711707115173,
      "learning_rate": 2.536202800331844e-07,
      "loss": 0.8807,
      "step": 10239
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.9940568208694458,
      "learning_rate": 2.5148543544425416e-07,
      "loss": 0.6304,
      "step": 10240
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.9926055669784546,
      "learning_rate": 2.4935960256883275e-07,
      "loss": 0.6711,
      "step": 10241
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.158828616142273,
      "learning_rate": 2.472427815989886e-07,
      "loss": 0.7608,
      "step": 10242
    },
    {
      "epoch": 3.85,
      "grad_norm": 1.2101527452468872,
      "learning_rate": 2.4513497272596887e-07,
      "loss": 0.8092,
      "step": 10243
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.1259400844573975,
      "learning_rate": 2.430361761401767e-07,
      "loss": 0.673,
      "step": 10244
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.127925992012024,
      "learning_rate": 2.409463920312605e-07,
      "loss": 0.6926,
      "step": 10245
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.0365463495254517,
      "learning_rate": 2.3886562058799133e-07,
      "loss": 0.7988,
      "step": 10246
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.107488989830017,
      "learning_rate": 2.367938619983745e-07,
      "loss": 0.7363,
      "step": 10247
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.1583000421524048,
      "learning_rate": 2.3473111644957135e-07,
      "loss": 0.6751,
      "step": 10248
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.104158878326416,
      "learning_rate": 2.3267738412794393e-07,
      "loss": 0.7982,
      "step": 10249
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.063612937927246,
      "learning_rate": 2.3063266521903271e-07,
      "loss": 0.7256,
      "step": 10250
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.9509289264678955,
      "learning_rate": 2.2859695990755658e-07,
      "loss": 0.6807,
      "step": 10251
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.9802016615867615,
      "learning_rate": 2.2657026837745733e-07,
      "loss": 0.7315,
      "step": 10252
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.0145543813705444,
      "learning_rate": 2.2455259081181068e-07,
      "loss": 0.6952,
      "step": 10253
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.0961731672286987,
      "learning_rate": 2.225439273929153e-07,
      "loss": 0.6098,
      "step": 10254
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.1338154077529907,
      "learning_rate": 2.2054427830223713e-07,
      "loss": 0.7888,
      "step": 10255
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.1103328466415405,
      "learning_rate": 2.1855364372043163e-07,
      "loss": 0.7788,
      "step": 10256
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.1680703163146973,
      "learning_rate": 2.1657202382734387e-07,
      "loss": 0.7121,
      "step": 10257
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.1330238580703735,
      "learning_rate": 2.145994188019973e-07,
      "loss": 0.7926,
      "step": 10258
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.102016568183899,
      "learning_rate": 2.1263582882261602e-07,
      "loss": 0.864,
      "step": 10259
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.1528857946395874,
      "learning_rate": 2.1068125406659145e-07,
      "loss": 0.7396,
      "step": 10260
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.0586588382720947,
      "learning_rate": 2.0873569471050458e-07,
      "loss": 0.6752,
      "step": 10261
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.9400776028633118,
      "learning_rate": 2.0679915093013703e-07,
      "loss": 0.6994,
      "step": 10262
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.0464847087860107,
      "learning_rate": 2.048716229004377e-07,
      "loss": 0.6697,
      "step": 10263
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.0557441711425781,
      "learning_rate": 2.0295311079554513e-07,
      "loss": 0.77,
      "step": 10264
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.200779676437378,
      "learning_rate": 2.010436147887984e-07,
      "loss": 0.6971,
      "step": 10265
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.968054473400116,
      "learning_rate": 1.991431350526929e-07,
      "loss": 0.6722,
      "step": 10266
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.0753424167633057,
      "learning_rate": 1.9725167175893567e-07,
      "loss": 0.7262,
      "step": 10267
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.994718611240387,
      "learning_rate": 1.9536922507841228e-07,
      "loss": 0.6985,
      "step": 10268
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.059556245803833,
      "learning_rate": 1.9349579518117555e-07,
      "loss": 0.6565,
      "step": 10269
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.102897047996521,
      "learning_rate": 1.9163138223650122e-07,
      "loss": 0.7052,
      "step": 10270
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.1199229955673218,
      "learning_rate": 1.8977598641282125e-07,
      "loss": 0.6092,
      "step": 10271
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.2251956462860107,
      "learning_rate": 1.8792960787774593e-07,
      "loss": 0.876,
      "step": 10272
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.0963590145111084,
      "learning_rate": 1.8609224679809743e-07,
      "loss": 0.8218,
      "step": 10273
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.2151145935058594,
      "learning_rate": 1.8426390333987632e-07,
      "loss": 0.8191,
      "step": 10274
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.0078610181808472,
      "learning_rate": 1.824445776682504e-07,
      "loss": 0.7281,
      "step": 10275
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.1063811779022217,
      "learning_rate": 1.8063426994759937e-07,
      "loss": 0.7435,
      "step": 10276
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.0754696130752563,
      "learning_rate": 1.7883298034145902e-07,
      "loss": 0.6787,
      "step": 10277
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.2878892421722412,
      "learning_rate": 1.7704070901257696e-07,
      "loss": 0.6559,
      "step": 10278
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.0053775310516357,
      "learning_rate": 1.7525745612286814e-07,
      "loss": 0.5906,
      "step": 10279
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.3063461780548096,
      "learning_rate": 1.73483221833437e-07,
      "loss": 0.8742,
      "step": 10280
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.1032742261886597,
      "learning_rate": 1.7171800630458868e-07,
      "loss": 0.7476,
      "step": 10281
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.11796236038208,
      "learning_rate": 1.6996180969579555e-07,
      "loss": 0.6904,
      "step": 10282
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.1269010305404663,
      "learning_rate": 1.6821463216570854e-07,
      "loss": 0.7188,
      "step": 10283
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.997366189956665,
      "learning_rate": 1.6647647387219023e-07,
      "loss": 0.5546,
      "step": 10284
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.148486614227295,
      "learning_rate": 1.6474733497225947e-07,
      "loss": 0.7628,
      "step": 10285
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.2299525737762451,
      "learning_rate": 1.6302721562214684e-07,
      "loss": 0.7923,
      "step": 10286
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.0600361824035645,
      "learning_rate": 1.6131611597723918e-07,
      "loss": 0.6856,
      "step": 10287
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.050676703453064,
      "learning_rate": 1.596140361921461e-07,
      "loss": 0.7628,
      "step": 10288
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.207611322402954,
      "learning_rate": 1.5792097642063354e-07,
      "loss": 0.6813,
      "step": 10289
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.0785123109817505,
      "learning_rate": 1.5623693681564577e-07,
      "loss": 0.6953,
      "step": 10290
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.094940423965454,
      "learning_rate": 1.5456191752933891e-07,
      "loss": 0.6755,
      "step": 10291
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.105302095413208,
      "learning_rate": 1.5289591871304744e-07,
      "loss": 0.6624,
      "step": 10292
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.228195071220398,
      "learning_rate": 1.5123894051727316e-07,
      "loss": 0.7273,
      "step": 10293
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.1043932437896729,
      "learning_rate": 1.4959098309170749e-07,
      "loss": 0.8344,
      "step": 10294
    },
    {
      "epoch": 3.87,
      "grad_norm": 1.1355969905853271,
      "learning_rate": 1.4795204658525353e-07,
      "loss": 0.8917,
      "step": 10295
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.9850760102272034,
      "learning_rate": 1.463221311459817e-07,
      "loss": 0.777,
      "step": 10296
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.1100587844848633,
      "learning_rate": 1.4470123692111869e-07,
      "loss": 0.845,
      "step": 10297
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.1376063823699951,
      "learning_rate": 1.4308936405712513e-07,
      "loss": 0.7928,
      "step": 10298
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.1133650541305542,
      "learning_rate": 1.4148651269962897e-07,
      "loss": 0.8205,
      "step": 10299
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.1653835773468018,
      "learning_rate": 1.398926829934144e-07,
      "loss": 0.8226,
      "step": 10300
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.090388536453247,
      "learning_rate": 1.3830787508249955e-07,
      "loss": 0.482,
      "step": 10301
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.2912877798080444,
      "learning_rate": 1.3673208911004765e-07,
      "loss": 0.7953,
      "step": 10302
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.0894230604171753,
      "learning_rate": 1.3516532521842263e-07,
      "loss": 0.6995,
      "step": 10303
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.077724814414978,
      "learning_rate": 1.3360758354918902e-07,
      "loss": 0.6704,
      "step": 10304
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.095837950706482,
      "learning_rate": 1.3205886424305646e-07,
      "loss": 0.6781,
      "step": 10305
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.1681841611862183,
      "learning_rate": 1.305191674399686e-07,
      "loss": 0.7088,
      "step": 10306
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.9835118055343628,
      "learning_rate": 1.2898849327900308e-07,
      "loss": 0.6012,
      "step": 10307
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.2376967668533325,
      "learning_rate": 1.2746684189846036e-07,
      "loss": 0.7213,
      "step": 10308
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.0905146598815918,
      "learning_rate": 1.2595421343581936e-07,
      "loss": 0.8111,
      "step": 10309
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.2606335878372192,
      "learning_rate": 1.2445060802771525e-07,
      "loss": 0.7741,
      "step": 10310
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.1465544700622559,
      "learning_rate": 1.229560258100171e-07,
      "loss": 0.8711,
      "step": 10311
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.2311286926269531,
      "learning_rate": 1.2147046691773911e-07,
      "loss": 0.7694,
      "step": 10312
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.016685128211975,
      "learning_rate": 1.199939314850851e-07,
      "loss": 0.7116,
      "step": 10313
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.0993316173553467,
      "learning_rate": 1.1852641964547051e-07,
      "loss": 0.717,
      "step": 10314
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.0414283275604248,
      "learning_rate": 1.1706793153145601e-07,
      "loss": 0.6909,
      "step": 10315
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.0833755731582642,
      "learning_rate": 1.1561846727482506e-07,
      "loss": 0.7905,
      "step": 10316
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.0231516361236572,
      "learning_rate": 1.1417802700652846e-07,
      "loss": 0.8382,
      "step": 10317
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.165995717048645,
      "learning_rate": 1.1274661085668437e-07,
      "loss": 0.7248,
      "step": 10318
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.0700451135635376,
      "learning_rate": 1.1132421895462264e-07,
      "loss": 0.6769,
      "step": 10319
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.24215829372406,
      "learning_rate": 1.0991085142886271e-07,
      "loss": 0.7443,
      "step": 10320
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.071614146232605,
      "learning_rate": 1.085065084070691e-07,
      "loss": 0.6101,
      "step": 10321
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.0156344175338745,
      "learning_rate": 1.0711119001612924e-07,
      "loss": 0.6923,
      "step": 10322
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.144028663635254,
      "learning_rate": 1.0572489638210892e-07,
      "loss": 0.6586,
      "step": 10323
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.022682785987854,
      "learning_rate": 1.043476276302302e-07,
      "loss": 0.6714,
      "step": 10324
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.0869050025939941,
      "learning_rate": 1.029793838849491e-07,
      "loss": 0.6297,
      "step": 10325
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.029304027557373,
      "learning_rate": 1.0162016526985563e-07,
      "loss": 0.861,
      "step": 10326
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.0667486190795898,
      "learning_rate": 1.0026997190776267e-07,
      "loss": 0.7638,
      "step": 10327
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.1721434593200684,
      "learning_rate": 9.89288039206504e-08,
      "loss": 0.739,
      "step": 10328
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.0044025182724,
      "learning_rate": 9.75966614296775e-08,
      "loss": 0.6764,
      "step": 10329
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.9981142282485962,
      "learning_rate": 9.627354455520321e-08,
      "loss": 0.6777,
      "step": 10330
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.1405082941055298,
      "learning_rate": 9.495945341675416e-08,
      "loss": 0.745,
      "step": 10331
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.0364693403244019,
      "learning_rate": 9.365438813306871e-08,
      "loss": 0.5952,
      "step": 10332
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.0092532634735107,
      "learning_rate": 9.235834882203031e-08,
      "loss": 0.6523,
      "step": 10333
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.1862736940383911,
      "learning_rate": 9.107133560074533e-08,
      "loss": 0.7349,
      "step": 10334
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.1217505931854248,
      "learning_rate": 8.979334858547627e-08,
      "loss": 0.7939,
      "step": 10335
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.0616002082824707,
      "learning_rate": 8.852438789169748e-08,
      "loss": 0.7334,
      "step": 10336
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.0876346826553345,
      "learning_rate": 8.726445363403945e-08,
      "loss": 0.7852,
      "step": 10337
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.1245304346084595,
      "learning_rate": 8.601354592633337e-08,
      "loss": 0.7742,
      "step": 10338
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.009469747543335,
      "learning_rate": 8.477166488158883e-08,
      "loss": 0.7959,
      "step": 10339
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.1147099733352661,
      "learning_rate": 8.353881061200497e-08,
      "loss": 0.6548,
      "step": 10340
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.2045879364013672,
      "learning_rate": 8.231498322897046e-08,
      "loss": 0.8655,
      "step": 10341
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.2170072793960571,
      "learning_rate": 8.110018284304133e-08,
      "loss": 0.7722,
      "step": 10342
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.137194037437439,
      "learning_rate": 7.98944095639631e-08,
      "loss": 0.791,
      "step": 10343
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.1293737888336182,
      "learning_rate": 7.869766350069308e-08,
      "loss": 0.7866,
      "step": 10344
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.0234087705612183,
      "learning_rate": 7.750994476132256e-08,
      "loss": 0.6933,
      "step": 10345
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.3251829147338867,
      "learning_rate": 7.633125345317682e-08,
      "loss": 0.8388,
      "step": 10346
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.072607159614563,
      "learning_rate": 7.516158968273734e-08,
      "loss": 0.8231,
      "step": 10347
    },
    {
      "epoch": 3.89,
      "grad_norm": 1.0104254484176636,
      "learning_rate": 7.400095355567515e-08,
      "loss": 0.6915,
      "step": 10348
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0770972967147827,
      "learning_rate": 7.284934517685083e-08,
      "loss": 0.8134,
      "step": 10349
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.1194143295288086,
      "learning_rate": 7.170676465029225e-08,
      "loss": 0.5891,
      "step": 10350
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.1037037372589111,
      "learning_rate": 7.057321207923905e-08,
      "loss": 0.5927,
      "step": 10351
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.005163311958313,
      "learning_rate": 6.944868756609824e-08,
      "loss": 0.6615,
      "step": 10352
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.1407287120819092,
      "learning_rate": 6.833319121246629e-08,
      "loss": 0.6878,
      "step": 10353
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.005780577659607,
      "learning_rate": 6.722672311910704e-08,
      "loss": 0.6374,
      "step": 10354
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0661780834197998,
      "learning_rate": 6.612928338600721e-08,
      "loss": 0.6672,
      "step": 10355
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0074044466018677,
      "learning_rate": 6.504087211229859e-08,
      "loss": 0.6394,
      "step": 10356
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0175762176513672,
      "learning_rate": 6.396148939632473e-08,
      "loss": 0.6373,
      "step": 10357
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.2253483533859253,
      "learning_rate": 6.289113533558544e-08,
      "loss": 0.617,
      "step": 10358
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0005747079849243,
      "learning_rate": 6.182981002679223e-08,
      "loss": 0.6712,
      "step": 10359
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0758814811706543,
      "learning_rate": 6.077751356582395e-08,
      "loss": 0.7089,
      "step": 10360
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0941797494888306,
      "learning_rate": 5.973424604774902e-08,
      "loss": 0.8444,
      "step": 10361
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0083016157150269,
      "learning_rate": 5.870000756683647e-08,
      "loss": 0.6643,
      "step": 10362
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.9881961345672607,
      "learning_rate": 5.767479821650046e-08,
      "loss": 0.7092,
      "step": 10363
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.1549128293991089,
      "learning_rate": 5.66586180893891e-08,
      "loss": 0.8437,
      "step": 10364
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0680090188980103,
      "learning_rate": 5.565146727728454e-08,
      "loss": 0.6498,
      "step": 10365
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.2584865093231201,
      "learning_rate": 5.465334587118065e-08,
      "loss": 0.7876,
      "step": 10366
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.2348891496658325,
      "learning_rate": 5.366425396126085e-08,
      "loss": 0.7512,
      "step": 10367
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0702674388885498,
      "learning_rate": 5.268419163688698e-08,
      "loss": 0.7843,
      "step": 10368
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.964288055896759,
      "learning_rate": 5.171315898659934e-08,
      "loss": 0.6978,
      "step": 10369
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0477534532546997,
      "learning_rate": 5.075115609811665e-08,
      "loss": 0.6807,
      "step": 10370
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.022194743156433,
      "learning_rate": 4.979818305834716e-08,
      "loss": 0.627,
      "step": 10371
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.1440361738204956,
      "learning_rate": 4.885423995341088e-08,
      "loss": 0.6837,
      "step": 10372
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.1446566581726074,
      "learning_rate": 4.7919326868561823e-08,
      "loss": 0.7622,
      "step": 10373
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0857774019241333,
      "learning_rate": 4.699344388827687e-08,
      "loss": 0.7284,
      "step": 10374
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0691699981689453,
      "learning_rate": 4.6076591096200213e-08,
      "loss": 0.7183,
      "step": 10375
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0890984535217285,
      "learning_rate": 4.5168768575176715e-08,
      "loss": 0.7093,
      "step": 10376
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.1601163148880005,
      "learning_rate": 4.4269976407196324e-08,
      "loss": 0.8735,
      "step": 10377
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0781034231185913,
      "learning_rate": 4.338021467348297e-08,
      "loss": 0.8226,
      "step": 10378
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.1257914304733276,
      "learning_rate": 4.2499483454416785e-08,
      "loss": 0.8641,
      "step": 10379
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0694079399108887,
      "learning_rate": 4.1627782829567476e-08,
      "loss": 0.7099,
      "step": 10380
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.1362286806106567,
      "learning_rate": 4.076511287767204e-08,
      "loss": 0.7249,
      "step": 10381
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.9777843952178955,
      "learning_rate": 3.991147367669035e-08,
      "loss": 0.7448,
      "step": 10382
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.1467918157577515,
      "learning_rate": 3.9066865303738486e-08,
      "loss": 0.6955,
      "step": 10383
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.1310486793518066,
      "learning_rate": 3.823128783511099e-08,
      "loss": 0.7772,
      "step": 10384
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.2427482604980469,
      "learning_rate": 3.7404741346303005e-08,
      "loss": 0.7194,
      "step": 10385
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.258570671081543,
      "learning_rate": 3.6587225911999254e-08,
      "loss": 0.8477,
      "step": 10386
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0018856525421143,
      "learning_rate": 3.5778741606040664e-08,
      "loss": 0.6433,
      "step": 10387
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.1217608451843262,
      "learning_rate": 3.497928850146881e-08,
      "loss": 0.6273,
      "step": 10388
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.9166359305381775,
      "learning_rate": 3.4188866670537e-08,
      "loss": 0.6236,
      "step": 10389
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0344411134719849,
      "learning_rate": 3.340747618462148e-08,
      "loss": 0.6665,
      "step": 10390
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.1618742942810059,
      "learning_rate": 3.263511711433242e-08,
      "loss": 0.8308,
      "step": 10391
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.1930203437805176,
      "learning_rate": 3.187178952945846e-08,
      "loss": 0.6346,
      "step": 10392
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0410478115081787,
      "learning_rate": 3.111749349893334e-08,
      "loss": 0.6746,
      "step": 10393
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0664647817611694,
      "learning_rate": 3.037222909093584e-08,
      "loss": 0.6596,
      "step": 10394
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.080114722251892,
      "learning_rate": 2.9635996372778807e-08,
      "loss": 0.684,
      "step": 10395
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0304405689239502,
      "learning_rate": 2.8908795410964585e-08,
      "loss": 0.7623,
      "step": 10396
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0488518476486206,
      "learning_rate": 2.819062627121838e-08,
      "loss": 0.7142,
      "step": 10397
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0898771286010742,
      "learning_rate": 2.7481489018410523e-08,
      "loss": 0.6602,
      "step": 10398
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.0864341259002686,
      "learning_rate": 2.6781383716600885e-08,
      "loss": 0.7361,
      "step": 10399
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.064661979675293,
      "learning_rate": 2.6090310429049968e-08,
      "loss": 0.7083,
      "step": 10400
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0142401456832886,
      "learning_rate": 2.540826921818562e-08,
      "loss": 0.6698,
      "step": 10401
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0332825183868408,
      "learning_rate": 2.473526014563632e-08,
      "loss": 0.7379,
      "step": 10402
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.9789639711380005,
      "learning_rate": 2.4071283272186773e-08,
      "loss": 0.7171,
      "step": 10403
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.005342960357666,
      "learning_rate": 2.341633865784454e-08,
      "loss": 0.6826,
      "step": 10404
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.2225888967514038,
      "learning_rate": 2.27704263617734e-08,
      "loss": 0.6967,
      "step": 10405
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0611093044281006,
      "learning_rate": 2.2133546442315578e-08,
      "loss": 0.6109,
      "step": 10406
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0371791124343872,
      "learning_rate": 2.1505698957025033e-08,
      "loss": 0.8175,
      "step": 10407
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0663065910339355,
      "learning_rate": 2.088688396261196e-08,
      "loss": 0.6707,
      "step": 10408
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0942575931549072,
      "learning_rate": 2.0277101514987184e-08,
      "loss": 0.7583,
      "step": 10409
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0970426797866821,
      "learning_rate": 1.967635166925108e-08,
      "loss": 0.7347,
      "step": 10410
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.124154806137085,
      "learning_rate": 1.9084634479660247e-08,
      "loss": 0.6359,
      "step": 10411
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0846364498138428,
      "learning_rate": 1.850194999968302e-08,
      "loss": 0.725,
      "step": 10412
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.1189419031143188,
      "learning_rate": 1.7928298281966183e-08,
      "loss": 0.7703,
      "step": 10413
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0126746892929077,
      "learning_rate": 1.736367937833494e-08,
      "loss": 0.6381,
      "step": 10414
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0252676010131836,
      "learning_rate": 1.6808093339781838e-08,
      "loss": 0.6599,
      "step": 10415
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.205735683441162,
      "learning_rate": 1.6261540216522264e-08,
      "loss": 0.8681,
      "step": 10416
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.022552728652954,
      "learning_rate": 1.5724020057927834e-08,
      "loss": 0.7699,
      "step": 10417
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.1499346494674683,
      "learning_rate": 1.5195532912559706e-08,
      "loss": 0.7251,
      "step": 10418
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.1192296743392944,
      "learning_rate": 1.4676078828157468e-08,
      "loss": 0.6778,
      "step": 10419
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.117783546447754,
      "learning_rate": 1.4165657851661351e-08,
      "loss": 0.5979,
      "step": 10420
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0307210683822632,
      "learning_rate": 1.3664270029167814e-08,
      "loss": 0.7157,
      "step": 10421
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.1085056066513062,
      "learning_rate": 1.3171915405996161e-08,
      "loss": 0.708,
      "step": 10422
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.2808337211608887,
      "learning_rate": 1.2688594026621925e-08,
      "loss": 0.686,
      "step": 10423
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.1182042360305786,
      "learning_rate": 1.2214305934699078e-08,
      "loss": 0.8056,
      "step": 10424
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.0577539205551147,
      "learning_rate": 1.1749051173082226e-08,
      "loss": 0.6434,
      "step": 10425
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.235945463180542,
      "learning_rate": 1.1292829783815517e-08,
      "loss": 0.6979,
      "step": 10426
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.1472970247268677,
      "learning_rate": 1.0845641808099327e-08,
      "loss": 0.7722,
      "step": 10427
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.0047780275344849,
      "learning_rate": 1.0407487286345774e-08,
      "loss": 0.7189,
      "step": 10428
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.058333396911621,
      "learning_rate": 9.978366258134309e-09,
      "loss": 0.7531,
      "step": 10429
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.170149326324463,
      "learning_rate": 9.558278762245022e-09,
      "loss": 0.6062,
      "step": 10430
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.9996271133422852,
      "learning_rate": 9.147224836614233e-09,
      "loss": 0.851,
      "step": 10431
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.1697818040847778,
      "learning_rate": 8.745204518378902e-09,
      "loss": 0.582,
      "step": 10432
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.0871305465698242,
      "learning_rate": 8.352217843865528e-09,
      "loss": 0.7915,
      "step": 10433
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.0466443300247192,
      "learning_rate": 7.968264848590145e-09,
      "loss": 0.6706,
      "step": 10434
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.0232700109481812,
      "learning_rate": 7.593345567213917e-09,
      "loss": 0.7941,
      "step": 10435
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.9262697100639343,
      "learning_rate": 7.22746003363195e-09,
      "loss": 0.6898,
      "step": 10436
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.113515019416809,
      "learning_rate": 6.8706082808955855e-09,
      "loss": 0.8549,
      "step": 10437
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.177722692489624,
      "learning_rate": 6.522790341234597e-09,
      "loss": 0.8178,
      "step": 10438
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.152925968170166,
      "learning_rate": 6.184006246079399e-09,
      "loss": 0.7865,
      "step": 10439
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.0532795190811157,
      "learning_rate": 5.854256026027738e-09,
      "loss": 0.7181,
      "step": 10440
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.1392796039581299,
      "learning_rate": 5.533539710889102e-09,
      "loss": 0.7507,
      "step": 10441
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.1688835620880127,
      "learning_rate": 5.221857329618107e-09,
      "loss": 0.8085,
      "step": 10442
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.1555466651916504,
      "learning_rate": 4.919208910392215e-09,
      "loss": 0.7079,
      "step": 10443
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.1976960897445679,
      "learning_rate": 4.625594480545115e-09,
      "loss": 0.8281,
      "step": 10444
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.9713913202285767,
      "learning_rate": 4.341014066600035e-09,
      "loss": 0.8622,
      "step": 10445
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.047964334487915,
      "learning_rate": 4.065467694269742e-09,
      "loss": 0.7357,
      "step": 10446
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.163371205329895,
      "learning_rate": 3.7989553884454354e-09,
      "loss": 0.824,
      "step": 10447
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.1952106952667236,
      "learning_rate": 3.5414771732189544e-09,
      "loss": 0.6064,
      "step": 10448
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.0980684757232666,
      "learning_rate": 3.2930330718383696e-09,
      "loss": 0.6395,
      "step": 10449
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.0460476875305176,
      "learning_rate": 3.053623106741288e-09,
      "loss": 0.7636,
      "step": 10450
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.0712224245071411,
      "learning_rate": 2.8232472995770585e-09,
      "loss": 0.8345,
      "step": 10451
    },
    {
      "epoch": 3.93,
      "grad_norm": 1.1951779127120972,
      "learning_rate": 2.6019056711512614e-09,
      "loss": 0.746,
      "step": 10452
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.125878930091858,
      "learning_rate": 2.3895982414590123e-09,
      "loss": 0.7096,
      "step": 10453
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.1865314245224,
      "learning_rate": 2.1863250296849656e-09,
      "loss": 0.7116,
      "step": 10454
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.0893054008483887,
      "learning_rate": 1.9920860541811084e-09,
      "loss": 0.7622,
      "step": 10455
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.9922797083854675,
      "learning_rate": 1.8068813325111677e-09,
      "loss": 0.7209,
      "step": 10456
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.035309910774231,
      "learning_rate": 1.6307108813951032e-09,
      "loss": 0.6607,
      "step": 10457
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.975841760635376,
      "learning_rate": 1.4635747167646153e-09,
      "loss": 0.7024,
      "step": 10458
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.1933470964431763,
      "learning_rate": 1.3054728537076344e-09,
      "loss": 0.6538,
      "step": 10459
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.046862006187439,
      "learning_rate": 1.156405306501629e-09,
      "loss": 0.6269,
      "step": 10460
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.0369027853012085,
      "learning_rate": 1.0163720886247064e-09,
      "loss": 0.8598,
      "step": 10461
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.1073603630065918,
      "learning_rate": 8.853732127334091e-10,
      "loss": 0.75,
      "step": 10462
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.0886046886444092,
      "learning_rate": 7.634086906516125e-10,
      "loss": 0.7341,
      "step": 10463
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.0717804431915283,
      "learning_rate": 6.50478533403831e-10,
      "loss": 0.5664,
      "step": 10464
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.115136981010437,
      "learning_rate": 5.465827511819122e-10,
      "loss": 0.7369,
      "step": 10465
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.1457079648971558,
      "learning_rate": 4.5172135338944487e-10,
      "loss": 0.6697,
      "step": 10466
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.1076148748397827,
      "learning_rate": 3.658943485862487e-10,
      "loss": 0.8078,
      "step": 10467
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.9490792155265808,
      "learning_rate": 2.8910174452168036e-10,
      "loss": 0.6723,
      "step": 10468
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.9843056201934814,
      "learning_rate": 2.21343548145736e-10,
      "loss": 0.6527,
      "step": 10469
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.0199030637741089,
      "learning_rate": 1.6261976557574442e-10,
      "loss": 0.7553,
      "step": 10470
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.1680927276611328,
      "learning_rate": 1.1293040211857176e-10,
      "loss": 0.8834,
      "step": 10471
    },
    {
      "epoch": 3.94,
      "grad_norm": 1.0654876232147217,
      "learning_rate": 7.227546224841675e-11,
      "loss": 0.9042,
      "step": 10472
    }
  ],
  "logging_steps": 1,
  "max_steps": 10472,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 2618,
  "total_flos": 8.128681149485548e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}