{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 763,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001310615989515072,
      "grad_norm": 0.0791015625,
      "learning_rate": 9.98689384010485e-06,
      "loss": 1.2556,
      "step": 1
    },
    {
      "epoch": 0.002621231979030144,
      "grad_norm": 0.0810546875,
      "learning_rate": 9.973787680209699e-06,
      "loss": 1.3767,
      "step": 2
    },
    {
      "epoch": 0.003931847968545216,
      "grad_norm": 0.0830078125,
      "learning_rate": 9.960681520314549e-06,
      "loss": 1.4053,
      "step": 3
    },
    {
      "epoch": 0.005242463958060288,
      "grad_norm": 0.08251953125,
      "learning_rate": 9.947575360419398e-06,
      "loss": 1.3944,
      "step": 4
    },
    {
      "epoch": 0.00655307994757536,
      "grad_norm": 0.0849609375,
      "learning_rate": 9.934469200524247e-06,
      "loss": 1.3895,
      "step": 5
    },
    {
      "epoch": 0.007863695937090432,
      "grad_norm": 0.0859375,
      "learning_rate": 9.921363040629096e-06,
      "loss": 1.3428,
      "step": 6
    },
    {
      "epoch": 0.009174311926605505,
      "grad_norm": 0.09228515625,
      "learning_rate": 9.908256880733946e-06,
      "loss": 1.465,
      "step": 7
    },
    {
      "epoch": 0.010484927916120577,
      "grad_norm": 0.091796875,
      "learning_rate": 9.895150720838795e-06,
      "loss": 1.3829,
      "step": 8
    },
    {
      "epoch": 0.011795543905635648,
      "grad_norm": 0.09423828125,
      "learning_rate": 9.882044560943644e-06,
      "loss": 1.4298,
      "step": 9
    },
    {
      "epoch": 0.01310615989515072,
      "grad_norm": 0.0947265625,
      "learning_rate": 9.868938401048493e-06,
      "loss": 1.3555,
      "step": 10
    },
    {
      "epoch": 0.014416775884665793,
      "grad_norm": 0.103515625,
      "learning_rate": 9.855832241153342e-06,
      "loss": 1.4013,
      "step": 11
    },
    {
      "epoch": 0.015727391874180863,
      "grad_norm": 0.10546875,
      "learning_rate": 9.842726081258192e-06,
      "loss": 1.4454,
      "step": 12
    },
    {
      "epoch": 0.01703800786369594,
      "grad_norm": 0.111328125,
      "learning_rate": 9.829619921363041e-06,
      "loss": 1.4422,
      "step": 13
    },
    {
      "epoch": 0.01834862385321101,
      "grad_norm": 0.111328125,
      "learning_rate": 9.81651376146789e-06,
      "loss": 1.3862,
      "step": 14
    },
    {
      "epoch": 0.019659239842726082,
      "grad_norm": 0.1064453125,
      "learning_rate": 9.803407601572739e-06,
      "loss": 1.2946,
      "step": 15
    },
    {
      "epoch": 0.020969855832241154,
      "grad_norm": 0.12451171875,
      "learning_rate": 9.790301441677588e-06,
      "loss": 1.4944,
      "step": 16
    },
    {
      "epoch": 0.022280471821756225,
      "grad_norm": 0.1240234375,
      "learning_rate": 9.777195281782439e-06,
      "loss": 1.461,
      "step": 17
    },
    {
      "epoch": 0.023591087811271297,
      "grad_norm": 0.11669921875,
      "learning_rate": 9.764089121887287e-06,
      "loss": 1.303,
      "step": 18
    },
    {
      "epoch": 0.02490170380078637,
      "grad_norm": 0.1240234375,
      "learning_rate": 9.750982961992136e-06,
      "loss": 1.3031,
      "step": 19
    },
    {
      "epoch": 0.02621231979030144,
      "grad_norm": 0.138671875,
      "learning_rate": 9.737876802096987e-06,
      "loss": 1.4607,
      "step": 20
    },
    {
      "epoch": 0.027522935779816515,
      "grad_norm": 0.140625,
      "learning_rate": 9.724770642201836e-06,
      "loss": 1.4104,
      "step": 21
    },
    {
      "epoch": 0.028833551769331587,
      "grad_norm": 0.1357421875,
      "learning_rate": 9.711664482306685e-06,
      "loss": 1.3664,
      "step": 22
    },
    {
      "epoch": 0.03014416775884666,
      "grad_norm": 0.142578125,
      "learning_rate": 9.698558322411533e-06,
      "loss": 1.3703,
      "step": 23
    },
    {
      "epoch": 0.03145478374836173,
      "grad_norm": 0.146484375,
      "learning_rate": 9.685452162516384e-06,
      "loss": 1.3818,
      "step": 24
    },
    {
      "epoch": 0.0327653997378768,
      "grad_norm": 0.1494140625,
      "learning_rate": 9.672346002621233e-06,
      "loss": 1.4122,
      "step": 25
    },
    {
      "epoch": 0.03407601572739188,
      "grad_norm": 0.15234375,
      "learning_rate": 9.659239842726082e-06,
      "loss": 1.3792,
      "step": 26
    },
    {
      "epoch": 0.035386631716906945,
      "grad_norm": 0.146484375,
      "learning_rate": 9.64613368283093e-06,
      "loss": 1.2953,
      "step": 27
    },
    {
      "epoch": 0.03669724770642202,
      "grad_norm": 0.162109375,
      "learning_rate": 9.633027522935781e-06,
      "loss": 1.2915,
      "step": 28
    },
    {
      "epoch": 0.03800786369593709,
      "grad_norm": 0.189453125,
      "learning_rate": 9.61992136304063e-06,
      "loss": 1.3475,
      "step": 29
    },
    {
      "epoch": 0.039318479685452164,
      "grad_norm": 0.1572265625,
      "learning_rate": 9.606815203145479e-06,
      "loss": 1.3437,
      "step": 30
    },
    {
      "epoch": 0.04062909567496723,
      "grad_norm": 0.16796875,
      "learning_rate": 9.59370904325033e-06,
      "loss": 1.364,
      "step": 31
    },
    {
      "epoch": 0.04193971166448231,
      "grad_norm": 0.1552734375,
      "learning_rate": 9.580602883355178e-06,
      "loss": 1.2421,
      "step": 32
    },
    {
      "epoch": 0.04325032765399738,
      "grad_norm": 0.166015625,
      "learning_rate": 9.567496723460027e-06,
      "loss": 1.2606,
      "step": 33
    },
    {
      "epoch": 0.04456094364351245,
      "grad_norm": 0.1640625,
      "learning_rate": 9.554390563564876e-06,
      "loss": 1.318,
      "step": 34
    },
    {
      "epoch": 0.045871559633027525,
      "grad_norm": 0.1875,
      "learning_rate": 9.541284403669727e-06,
      "loss": 1.4466,
      "step": 35
    },
    {
      "epoch": 0.047182175622542594,
      "grad_norm": 0.1923828125,
      "learning_rate": 9.528178243774576e-06,
      "loss": 1.4755,
      "step": 36
    },
    {
      "epoch": 0.04849279161205767,
      "grad_norm": 0.1708984375,
      "learning_rate": 9.515072083879425e-06,
      "loss": 1.2888,
      "step": 37
    },
    {
      "epoch": 0.04980340760157274,
      "grad_norm": 0.17578125,
      "learning_rate": 9.501965923984273e-06,
      "loss": 1.2959,
      "step": 38
    },
    {
      "epoch": 0.05111402359108781,
      "grad_norm": 0.18359375,
      "learning_rate": 9.488859764089122e-06,
      "loss": 1.3576,
      "step": 39
    },
    {
      "epoch": 0.05242463958060288,
      "grad_norm": 0.18359375,
      "learning_rate": 9.475753604193973e-06,
      "loss": 1.3013,
      "step": 40
    },
    {
      "epoch": 0.053735255570117955,
      "grad_norm": 0.1787109375,
      "learning_rate": 9.462647444298822e-06,
      "loss": 1.2869,
      "step": 41
    },
    {
      "epoch": 0.05504587155963303,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.44954128440367e-06,
      "loss": 1.222,
      "step": 42
    },
    {
      "epoch": 0.0563564875491481,
      "grad_norm": 0.205078125,
      "learning_rate": 9.43643512450852e-06,
      "loss": 1.4095,
      "step": 43
    },
    {
      "epoch": 0.057667103538663174,
      "grad_norm": 0.1787109375,
      "learning_rate": 9.423328964613368e-06,
      "loss": 1.2354,
      "step": 44
    },
    {
      "epoch": 0.05897771952817824,
      "grad_norm": 0.19140625,
      "learning_rate": 9.410222804718219e-06,
      "loss": 1.3337,
      "step": 45
    },
    {
      "epoch": 0.06028833551769332,
      "grad_norm": 0.1962890625,
      "learning_rate": 9.397116644823068e-06,
      "loss": 1.3496,
      "step": 46
    },
    {
      "epoch": 0.061598951507208385,
      "grad_norm": 0.1826171875,
      "learning_rate": 9.384010484927917e-06,
      "loss": 1.2765,
      "step": 47
    },
    {
      "epoch": 0.06290956749672345,
      "grad_norm": 0.185546875,
      "learning_rate": 9.370904325032766e-06,
      "loss": 1.3376,
      "step": 48
    },
    {
      "epoch": 0.06422018348623854,
      "grad_norm": 0.189453125,
      "learning_rate": 9.357798165137616e-06,
      "loss": 1.2688,
      "step": 49
    },
    {
      "epoch": 0.0655307994757536,
      "grad_norm": 0.1875,
      "learning_rate": 9.344692005242465e-06,
      "loss": 1.3008,
      "step": 50
    },
    {
      "epoch": 0.06684141546526867,
      "grad_norm": 0.1904296875,
      "learning_rate": 9.331585845347314e-06,
      "loss": 1.2723,
      "step": 51
    },
    {
      "epoch": 0.06815203145478375,
      "grad_norm": 0.1767578125,
      "learning_rate": 9.318479685452163e-06,
      "loss": 1.2026,
      "step": 52
    },
    {
      "epoch": 0.06946264744429882,
      "grad_norm": 0.1787109375,
      "learning_rate": 9.305373525557012e-06,
      "loss": 1.231,
      "step": 53
    },
    {
      "epoch": 0.07077326343381389,
      "grad_norm": 0.1884765625,
      "learning_rate": 9.292267365661862e-06,
      "loss": 1.2807,
      "step": 54
    },
    {
      "epoch": 0.07208387942332896,
      "grad_norm": 0.197265625,
      "learning_rate": 9.279161205766711e-06,
      "loss": 1.3617,
      "step": 55
    },
    {
      "epoch": 0.07339449541284404,
      "grad_norm": 0.171875,
      "learning_rate": 9.26605504587156e-06,
      "loss": 1.1847,
      "step": 56
    },
    {
      "epoch": 0.07470511140235911,
      "grad_norm": 0.203125,
      "learning_rate": 9.252948885976409e-06,
      "loss": 1.4314,
      "step": 57
    },
    {
      "epoch": 0.07601572739187418,
      "grad_norm": 0.189453125,
      "learning_rate": 9.239842726081258e-06,
      "loss": 1.2855,
      "step": 58
    },
    {
      "epoch": 0.07732634338138926,
      "grad_norm": 0.173828125,
      "learning_rate": 9.226736566186108e-06,
      "loss": 1.2622,
      "step": 59
    },
    {
      "epoch": 0.07863695937090433,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.213630406290957e-06,
      "loss": 1.2148,
      "step": 60
    },
    {
      "epoch": 0.0799475753604194,
      "grad_norm": 0.185546875,
      "learning_rate": 9.200524246395806e-06,
      "loss": 1.2849,
      "step": 61
    },
    {
      "epoch": 0.08125819134993446,
      "grad_norm": 0.1796875,
      "learning_rate": 9.187418086500655e-06,
      "loss": 1.2641,
      "step": 62
    },
    {
      "epoch": 0.08256880733944955,
      "grad_norm": 0.17578125,
      "learning_rate": 9.174311926605506e-06,
      "loss": 1.2309,
      "step": 63
    },
    {
      "epoch": 0.08387942332896461,
      "grad_norm": 0.171875,
      "learning_rate": 9.161205766710354e-06,
      "loss": 1.2382,
      "step": 64
    },
    {
      "epoch": 0.08519003931847968,
      "grad_norm": 0.1748046875,
      "learning_rate": 9.148099606815203e-06,
      "loss": 1.223,
      "step": 65
    },
    {
      "epoch": 0.08650065530799476,
      "grad_norm": 0.1767578125,
      "learning_rate": 9.134993446920052e-06,
      "loss": 1.2475,
      "step": 66
    },
    {
      "epoch": 0.08781127129750983,
      "grad_norm": 0.1640625,
      "learning_rate": 9.121887287024903e-06,
      "loss": 1.2017,
      "step": 67
    },
    {
      "epoch": 0.0891218872870249,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.108781127129752e-06,
      "loss": 1.2381,
      "step": 68
    },
    {
      "epoch": 0.09043250327653997,
      "grad_norm": 0.16015625,
      "learning_rate": 9.0956749672346e-06,
      "loss": 1.1767,
      "step": 69
    },
    {
      "epoch": 0.09174311926605505,
      "grad_norm": 0.1640625,
      "learning_rate": 9.08256880733945e-06,
      "loss": 1.1802,
      "step": 70
    },
    {
      "epoch": 0.09305373525557012,
      "grad_norm": 0.162109375,
      "learning_rate": 9.0694626474443e-06,
      "loss": 1.1855,
      "step": 71
    },
    {
      "epoch": 0.09436435124508519,
      "grad_norm": 0.16796875,
      "learning_rate": 9.056356487549149e-06,
      "loss": 1.235,
      "step": 72
    },
    {
      "epoch": 0.09567496723460026,
      "grad_norm": 0.1630859375,
      "learning_rate": 9.043250327653998e-06,
      "loss": 1.2176,
      "step": 73
    },
    {
      "epoch": 0.09698558322411534,
      "grad_norm": 0.146484375,
      "learning_rate": 9.030144167758848e-06,
      "loss": 1.0679,
      "step": 74
    },
    {
      "epoch": 0.0982961992136304,
      "grad_norm": 0.154296875,
      "learning_rate": 9.017038007863697e-06,
      "loss": 1.1711,
      "step": 75
    },
    {
      "epoch": 0.09960681520314547,
      "grad_norm": 0.1826171875,
      "learning_rate": 9.003931847968546e-06,
      "loss": 1.2938,
      "step": 76
    },
    {
      "epoch": 0.10091743119266056,
      "grad_norm": 0.166015625,
      "learning_rate": 8.990825688073395e-06,
      "loss": 1.2285,
      "step": 77
    },
    {
      "epoch": 0.10222804718217562,
      "grad_norm": 0.154296875,
      "learning_rate": 8.977719528178246e-06,
      "loss": 1.1362,
      "step": 78
    },
    {
      "epoch": 0.10353866317169069,
      "grad_norm": 0.158203125,
      "learning_rate": 8.964613368283094e-06,
      "loss": 1.1319,
      "step": 79
    },
    {
      "epoch": 0.10484927916120576,
      "grad_norm": 0.150390625,
      "learning_rate": 8.951507208387943e-06,
      "loss": 1.1595,
      "step": 80
    },
    {
      "epoch": 0.10615989515072084,
      "grad_norm": 0.1484375,
      "learning_rate": 8.938401048492792e-06,
      "loss": 1.1481,
      "step": 81
    },
    {
      "epoch": 0.10747051114023591,
      "grad_norm": 0.185546875,
      "learning_rate": 8.925294888597643e-06,
      "loss": 1.2817,
      "step": 82
    },
    {
      "epoch": 0.10878112712975098,
      "grad_norm": 0.1513671875,
      "learning_rate": 8.912188728702492e-06,
      "loss": 1.1217,
      "step": 83
    },
    {
      "epoch": 0.11009174311926606,
      "grad_norm": 0.1689453125,
      "learning_rate": 8.89908256880734e-06,
      "loss": 1.2165,
      "step": 84
    },
    {
      "epoch": 0.11140235910878113,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.88597640891219e-06,
      "loss": 1.1469,
      "step": 85
    },
    {
      "epoch": 0.1127129750982962,
      "grad_norm": 0.162109375,
      "learning_rate": 8.872870249017038e-06,
      "loss": 1.1946,
      "step": 86
    },
    {
      "epoch": 0.11402359108781127,
      "grad_norm": 0.1494140625,
      "learning_rate": 8.859764089121889e-06,
      "loss": 1.1542,
      "step": 87
    },
    {
      "epoch": 0.11533420707732635,
      "grad_norm": 0.1455078125,
      "learning_rate": 8.846657929226738e-06,
      "loss": 1.1733,
      "step": 88
    },
    {
      "epoch": 0.11664482306684142,
      "grad_norm": 0.140625,
      "learning_rate": 8.833551769331587e-06,
      "loss": 1.1442,
      "step": 89
    },
    {
      "epoch": 0.11795543905635648,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.820445609436435e-06,
      "loss": 1.0989,
      "step": 90
    },
    {
      "epoch": 0.11926605504587157,
      "grad_norm": 0.1396484375,
      "learning_rate": 8.807339449541286e-06,
      "loss": 1.1514,
      "step": 91
    },
    {
      "epoch": 0.12057667103538663,
      "grad_norm": 0.142578125,
      "learning_rate": 8.794233289646135e-06,
      "loss": 1.1793,
      "step": 92
    },
    {
      "epoch": 0.1218872870249017,
      "grad_norm": 0.12158203125,
      "learning_rate": 8.781127129750984e-06,
      "loss": 1.0437,
      "step": 93
    },
    {
      "epoch": 0.12319790301441677,
      "grad_norm": 0.12353515625,
      "learning_rate": 8.768020969855833e-06,
      "loss": 1.1171,
      "step": 94
    },
    {
      "epoch": 0.12450851900393185,
      "grad_norm": 0.123046875,
      "learning_rate": 8.754914809960682e-06,
      "loss": 1.0807,
      "step": 95
    },
    {
      "epoch": 0.1258191349934469,
      "grad_norm": 0.12890625,
      "learning_rate": 8.741808650065532e-06,
      "loss": 1.1298,
      "step": 96
    },
    {
      "epoch": 0.127129750982962,
      "grad_norm": 0.140625,
      "learning_rate": 8.728702490170381e-06,
      "loss": 1.2133,
      "step": 97
    },
    {
      "epoch": 0.12844036697247707,
      "grad_norm": 0.1376953125,
      "learning_rate": 8.71559633027523e-06,
      "loss": 1.1301,
      "step": 98
    },
    {
      "epoch": 0.12975098296199214,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.702490170380079e-06,
      "loss": 1.0789,
      "step": 99
    },
    {
      "epoch": 0.1310615989515072,
      "grad_norm": 0.1376953125,
      "learning_rate": 8.689384010484928e-06,
      "loss": 1.1856,
      "step": 100
    },
    {
      "epoch": 0.13237221494102228,
      "grad_norm": 0.1201171875,
      "learning_rate": 8.676277850589778e-06,
      "loss": 1.0449,
      "step": 101
    },
    {
      "epoch": 0.13368283093053734,
      "grad_norm": 0.134765625,
      "learning_rate": 8.663171690694627e-06,
      "loss": 1.1543,
      "step": 102
    },
    {
      "epoch": 0.1349934469200524,
      "grad_norm": 0.11181640625,
      "learning_rate": 8.650065530799476e-06,
      "loss": 1.0142,
      "step": 103
    },
    {
      "epoch": 0.1363040629095675,
      "grad_norm": 0.1474609375,
      "learning_rate": 8.636959370904325e-06,
      "loss": 1.2604,
      "step": 104
    },
    {
      "epoch": 0.13761467889908258,
      "grad_norm": 0.126953125,
      "learning_rate": 8.623853211009175e-06,
      "loss": 1.1036,
      "step": 105
    },
    {
      "epoch": 0.13892529488859764,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.610747051114024e-06,
      "loss": 1.1137,
      "step": 106
    },
    {
      "epoch": 0.1402359108781127,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.597640891218873e-06,
      "loss": 1.133,
      "step": 107
    },
    {
      "epoch": 0.14154652686762778,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.584534731323722e-06,
      "loss": 1.0935,
      "step": 108
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.1171875,
      "learning_rate": 8.571428571428571e-06,
      "loss": 1.0659,
      "step": 109
    },
    {
      "epoch": 0.14416775884665792,
      "grad_norm": 0.12158203125,
      "learning_rate": 8.558322411533421e-06,
      "loss": 1.1022,
      "step": 110
    },
    {
      "epoch": 0.145478374836173,
      "grad_norm": 0.130859375,
      "learning_rate": 8.54521625163827e-06,
      "loss": 1.1436,
      "step": 111
    },
    {
      "epoch": 0.14678899082568808,
      "grad_norm": 0.11669921875,
      "learning_rate": 8.53211009174312e-06,
      "loss": 1.0806,
      "step": 112
    },
    {
      "epoch": 0.14809960681520315,
      "grad_norm": 0.123046875,
      "learning_rate": 8.519003931847968e-06,
      "loss": 1.1075,
      "step": 113
    },
    {
      "epoch": 0.14941022280471822,
      "grad_norm": 0.1123046875,
      "learning_rate": 8.505897771952819e-06,
      "loss": 1.0394,
      "step": 114
    },
    {
      "epoch": 0.15072083879423329,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.492791612057668e-06,
      "loss": 1.138,
      "step": 115
    },
    {
      "epoch": 0.15203145478374835,
      "grad_norm": 0.12255859375,
      "learning_rate": 8.479685452162516e-06,
      "loss": 1.1298,
      "step": 116
    },
    {
      "epoch": 0.15334207077326342,
      "grad_norm": 0.1240234375,
      "learning_rate": 8.466579292267367e-06,
      "loss": 1.0441,
      "step": 117
    },
    {
      "epoch": 0.15465268676277852,
      "grad_norm": 0.1171875,
      "learning_rate": 8.453473132372216e-06,
      "loss": 1.078,
      "step": 118
    },
    {
      "epoch": 0.1559633027522936,
      "grad_norm": 0.130859375,
      "learning_rate": 8.440366972477065e-06,
      "loss": 1.0481,
      "step": 119
    },
    {
      "epoch": 0.15727391874180865,
      "grad_norm": 0.126953125,
      "learning_rate": 8.427260812581914e-06,
      "loss": 1.108,
      "step": 120
    },
    {
      "epoch": 0.15858453473132372,
      "grad_norm": 0.11181640625,
      "learning_rate": 8.414154652686764e-06,
      "loss": 1.0522,
      "step": 121
    },
    {
      "epoch": 0.1598951507208388,
      "grad_norm": 0.12353515625,
      "learning_rate": 8.401048492791613e-06,
      "loss": 1.1116,
      "step": 122
    },
    {
      "epoch": 0.16120576671035386,
      "grad_norm": 0.12890625,
      "learning_rate": 8.387942332896462e-06,
      "loss": 1.0993,
      "step": 123
    },
    {
      "epoch": 0.16251638269986893,
      "grad_norm": 0.11962890625,
      "learning_rate": 8.374836173001311e-06,
      "loss": 1.0644,
      "step": 124
    },
    {
      "epoch": 0.16382699868938402,
      "grad_norm": 0.11279296875,
      "learning_rate": 8.361730013106161e-06,
      "loss": 1.0056,
      "step": 125
    },
    {
      "epoch": 0.1651376146788991,
      "grad_norm": 0.11328125,
      "learning_rate": 8.34862385321101e-06,
      "loss": 1.0006,
      "step": 126
    },
    {
      "epoch": 0.16644823066841416,
      "grad_norm": 0.13671875,
      "learning_rate": 8.33551769331586e-06,
      "loss": 1.1509,
      "step": 127
    },
    {
      "epoch": 0.16775884665792923,
      "grad_norm": 0.11279296875,
      "learning_rate": 8.32241153342071e-06,
      "loss": 1.0404,
      "step": 128
    },
    {
      "epoch": 0.1690694626474443,
      "grad_norm": 0.11083984375,
      "learning_rate": 8.309305373525559e-06,
      "loss": 1.0368,
      "step": 129
    },
    {
      "epoch": 0.17038007863695936,
      "grad_norm": 0.1328125,
      "learning_rate": 8.296199213630408e-06,
      "loss": 1.154,
      "step": 130
    },
    {
      "epoch": 0.17169069462647443,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.283093053735256e-06,
      "loss": 1.0531,
      "step": 131
    },
    {
      "epoch": 0.17300131061598953,
      "grad_norm": 0.123046875,
      "learning_rate": 8.269986893840105e-06,
      "loss": 1.0737,
      "step": 132
    },
    {
      "epoch": 0.1743119266055046,
      "grad_norm": 0.1259765625,
      "learning_rate": 8.256880733944956e-06,
      "loss": 1.0988,
      "step": 133
    },
    {
      "epoch": 0.17562254259501967,
      "grad_norm": 0.1171875,
      "learning_rate": 8.243774574049805e-06,
      "loss": 1.0644,
      "step": 134
    },
    {
      "epoch": 0.17693315858453473,
      "grad_norm": 0.11767578125,
      "learning_rate": 8.230668414154654e-06,
      "loss": 1.0439,
      "step": 135
    },
    {
      "epoch": 0.1782437745740498,
      "grad_norm": 0.12255859375,
      "learning_rate": 8.217562254259502e-06,
      "loss": 1.067,
      "step": 136
    },
    {
      "epoch": 0.17955439056356487,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.204456094364351e-06,
      "loss": 1.0904,
      "step": 137
    },
    {
      "epoch": 0.18086500655307994,
      "grad_norm": 0.10986328125,
      "learning_rate": 8.191349934469202e-06,
      "loss": 0.9656,
      "step": 138
    },
    {
      "epoch": 0.182175622542595,
      "grad_norm": 0.123046875,
      "learning_rate": 8.17824377457405e-06,
      "loss": 1.0849,
      "step": 139
    },
    {
      "epoch": 0.1834862385321101,
      "grad_norm": 0.11669921875,
      "learning_rate": 8.1651376146789e-06,
      "loss": 1.0446,
      "step": 140
    },
    {
      "epoch": 0.18479685452162517,
      "grad_norm": 0.11865234375,
      "learning_rate": 8.152031454783749e-06,
      "loss": 1.0548,
      "step": 141
    },
    {
      "epoch": 0.18610747051114024,
      "grad_norm": 0.1015625,
      "learning_rate": 8.1389252948886e-06,
      "loss": 0.9563,
      "step": 142
    },
    {
      "epoch": 0.1874180865006553,
      "grad_norm": 0.1171875,
      "learning_rate": 8.125819134993448e-06,
      "loss": 1.0506,
      "step": 143
    },
    {
      "epoch": 0.18872870249017037,
      "grad_norm": 0.115234375,
      "learning_rate": 8.112712975098297e-06,
      "loss": 1.0642,
      "step": 144
    },
    {
      "epoch": 0.19003931847968544,
      "grad_norm": 0.10791015625,
      "learning_rate": 8.099606815203146e-06,
      "loss": 1.023,
      "step": 145
    },
    {
      "epoch": 0.1913499344692005,
      "grad_norm": 0.123046875,
      "learning_rate": 8.086500655307995e-06,
      "loss": 1.0591,
      "step": 146
    },
    {
      "epoch": 0.1926605504587156,
      "grad_norm": 0.11328125,
      "learning_rate": 8.073394495412845e-06,
      "loss": 1.0078,
      "step": 147
    },
    {
      "epoch": 0.19397116644823068,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.060288335517694e-06,
      "loss": 1.0094,
      "step": 148
    },
    {
      "epoch": 0.19528178243774574,
      "grad_norm": 0.1142578125,
      "learning_rate": 8.047182175622543e-06,
      "loss": 1.0657,
      "step": 149
    },
    {
      "epoch": 0.1965923984272608,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.034076015727392e-06,
      "loss": 1.0694,
      "step": 150
    },
    {
      "epoch": 0.19790301441677588,
      "grad_norm": 0.1240234375,
      "learning_rate": 8.02096985583224e-06,
      "loss": 1.0499,
      "step": 151
    },
    {
      "epoch": 0.19921363040629095,
      "grad_norm": 0.1201171875,
      "learning_rate": 8.007863695937091e-06,
      "loss": 1.0521,
      "step": 152
    },
    {
      "epoch": 0.20052424639580602,
      "grad_norm": 0.1279296875,
      "learning_rate": 7.99475753604194e-06,
      "loss": 1.0872,
      "step": 153
    },
    {
      "epoch": 0.2018348623853211,
      "grad_norm": 0.1123046875,
      "learning_rate": 7.981651376146789e-06,
      "loss": 0.9765,
      "step": 154
    },
    {
      "epoch": 0.20314547837483618,
      "grad_norm": 0.10205078125,
      "learning_rate": 7.968545216251638e-06,
      "loss": 0.95,
      "step": 155
    },
    {
      "epoch": 0.20445609436435125,
      "grad_norm": 0.10546875,
      "learning_rate": 7.955439056356489e-06,
      "loss": 0.9681,
      "step": 156
    },
    {
      "epoch": 0.20576671035386632,
      "grad_norm": 0.1337890625,
      "learning_rate": 7.942332896461337e-06,
      "loss": 1.0637,
      "step": 157
    },
    {
      "epoch": 0.20707732634338138,
      "grad_norm": 0.12353515625,
      "learning_rate": 7.929226736566186e-06,
      "loss": 1.0796,
      "step": 158
    },
    {
      "epoch": 0.20838794233289645,
      "grad_norm": 0.11962890625,
      "learning_rate": 7.916120576671035e-06,
      "loss": 1.0473,
      "step": 159
    },
    {
      "epoch": 0.20969855832241152,
      "grad_norm": 0.107421875,
      "learning_rate": 7.903014416775884e-06,
      "loss": 0.9582,
      "step": 160
    },
    {
      "epoch": 0.21100917431192662,
      "grad_norm": 0.11328125,
      "learning_rate": 7.889908256880735e-06,
      "loss": 1.0292,
      "step": 161
    },
    {
      "epoch": 0.21231979030144169,
      "grad_norm": 0.1044921875,
      "learning_rate": 7.876802096985583e-06,
      "loss": 1.027,
      "step": 162
    },
    {
      "epoch": 0.21363040629095675,
      "grad_norm": 0.109375,
      "learning_rate": 7.863695937090432e-06,
      "loss": 0.9905,
      "step": 163
    },
    {
      "epoch": 0.21494102228047182,
      "grad_norm": 0.109375,
      "learning_rate": 7.850589777195283e-06,
      "loss": 0.9685,
      "step": 164
    },
    {
      "epoch": 0.2162516382699869,
      "grad_norm": 0.111328125,
      "learning_rate": 7.837483617300132e-06,
      "loss": 0.9984,
      "step": 165
    },
    {
      "epoch": 0.21756225425950196,
      "grad_norm": 0.09765625,
      "learning_rate": 7.82437745740498e-06,
      "loss": 0.9624,
      "step": 166
    },
    {
      "epoch": 0.21887287024901703,
      "grad_norm": 0.1171875,
      "learning_rate": 7.81127129750983e-06,
      "loss": 1.0142,
      "step": 167
    },
    {
      "epoch": 0.22018348623853212,
      "grad_norm": 0.11767578125,
      "learning_rate": 7.79816513761468e-06,
      "loss": 1.0403,
      "step": 168
    },
    {
      "epoch": 0.2214941022280472,
      "grad_norm": 0.10791015625,
      "learning_rate": 7.785058977719529e-06,
      "loss": 0.9952,
      "step": 169
    },
    {
      "epoch": 0.22280471821756226,
      "grad_norm": 0.103515625,
      "learning_rate": 7.771952817824378e-06,
      "loss": 0.968,
      "step": 170
    },
    {
      "epoch": 0.22411533420707733,
      "grad_norm": 0.10546875,
      "learning_rate": 7.758846657929228e-06,
      "loss": 0.9589,
      "step": 171
    },
    {
      "epoch": 0.2254259501965924,
      "grad_norm": 0.109375,
      "learning_rate": 7.745740498034077e-06,
      "loss": 1.012,
      "step": 172
    },
    {
      "epoch": 0.22673656618610746,
      "grad_norm": 0.107421875,
      "learning_rate": 7.732634338138926e-06,
      "loss": 1.017,
      "step": 173
    },
    {
      "epoch": 0.22804718217562253,
      "grad_norm": 0.099609375,
      "learning_rate": 7.719528178243775e-06,
      "loss": 0.9515,
      "step": 174
    },
    {
      "epoch": 0.22935779816513763,
      "grad_norm": 0.10009765625,
      "learning_rate": 7.706422018348626e-06,
      "loss": 0.9591,
      "step": 175
    },
    {
      "epoch": 0.2306684141546527,
      "grad_norm": 0.10791015625,
      "learning_rate": 7.693315858453475e-06,
      "loss": 0.9612,
      "step": 176
    },
    {
      "epoch": 0.23197903014416776,
      "grad_norm": 0.09326171875,
      "learning_rate": 7.680209698558323e-06,
      "loss": 0.936,
      "step": 177
    },
    {
      "epoch": 0.23328964613368283,
      "grad_norm": 0.1025390625,
      "learning_rate": 7.667103538663172e-06,
      "loss": 0.987,
      "step": 178
    },
    {
      "epoch": 0.2346002621231979,
      "grad_norm": 0.10107421875,
      "learning_rate": 7.653997378768021e-06,
      "loss": 0.9288,
      "step": 179
    },
    {
      "epoch": 0.23591087811271297,
      "grad_norm": 0.103515625,
      "learning_rate": 7.640891218872872e-06,
      "loss": 0.9708,
      "step": 180
    },
    {
      "epoch": 0.23722149410222804,
      "grad_norm": 0.10107421875,
      "learning_rate": 7.627785058977721e-06,
      "loss": 0.9012,
      "step": 181
    },
    {
      "epoch": 0.23853211009174313,
      "grad_norm": 0.10009765625,
      "learning_rate": 7.6146788990825695e-06,
      "loss": 0.97,
      "step": 182
    },
    {
      "epoch": 0.2398427260812582,
      "grad_norm": 0.0947265625,
      "learning_rate": 7.601572739187418e-06,
      "loss": 0.9259,
      "step": 183
    },
    {
      "epoch": 0.24115334207077327,
      "grad_norm": 0.10546875,
      "learning_rate": 7.588466579292268e-06,
      "loss": 0.9549,
      "step": 184
    },
    {
      "epoch": 0.24246395806028834,
      "grad_norm": 0.10595703125,
      "learning_rate": 7.575360419397118e-06,
      "loss": 0.9635,
      "step": 185
    },
    {
      "epoch": 0.2437745740498034,
      "grad_norm": 0.1103515625,
      "learning_rate": 7.562254259501967e-06,
      "loss": 0.9415,
      "step": 186
    },
    {
      "epoch": 0.24508519003931847,
      "grad_norm": 0.1005859375,
      "learning_rate": 7.549148099606816e-06,
      "loss": 0.9457,
      "step": 187
    },
    {
      "epoch": 0.24639580602883354,
      "grad_norm": 0.0986328125,
      "learning_rate": 7.5360419397116645e-06,
      "loss": 0.9088,
      "step": 188
    },
    {
      "epoch": 0.24770642201834864,
      "grad_norm": 0.09130859375,
      "learning_rate": 7.522935779816515e-06,
      "loss": 0.8947,
      "step": 189
    },
    {
      "epoch": 0.2490170380078637,
      "grad_norm": 0.1044921875,
      "learning_rate": 7.509829619921364e-06,
      "loss": 0.9703,
      "step": 190
    },
    {
      "epoch": 0.2503276539973788,
      "grad_norm": 0.10205078125,
      "learning_rate": 7.496723460026213e-06,
      "loss": 0.9181,
      "step": 191
    },
    {
      "epoch": 0.2516382699868938,
      "grad_norm": 0.123046875,
      "learning_rate": 7.483617300131062e-06,
      "loss": 1.0153,
      "step": 192
    },
    {
      "epoch": 0.2529488859764089,
      "grad_norm": 0.1083984375,
      "learning_rate": 7.470511140235911e-06,
      "loss": 0.9802,
      "step": 193
    },
    {
      "epoch": 0.254259501965924,
      "grad_norm": 0.103515625,
      "learning_rate": 7.457404980340761e-06,
      "loss": 0.9592,
      "step": 194
    },
    {
      "epoch": 0.25557011795543905,
      "grad_norm": 0.083984375,
      "learning_rate": 7.44429882044561e-06,
      "loss": 0.8833,
      "step": 195
    },
    {
      "epoch": 0.25688073394495414,
      "grad_norm": 0.0947265625,
      "learning_rate": 7.431192660550459e-06,
      "loss": 0.941,
      "step": 196
    },
    {
      "epoch": 0.2581913499344692,
      "grad_norm": 0.09814453125,
      "learning_rate": 7.418086500655308e-06,
      "loss": 0.9528,
      "step": 197
    },
    {
      "epoch": 0.2595019659239843,
      "grad_norm": 0.09619140625,
      "learning_rate": 7.404980340760158e-06,
      "loss": 0.8998,
      "step": 198
    },
    {
      "epoch": 0.2608125819134993,
      "grad_norm": 0.09228515625,
      "learning_rate": 7.391874180865007e-06,
      "loss": 0.9324,
      "step": 199
    },
    {
      "epoch": 0.2621231979030144,
      "grad_norm": 0.10107421875,
      "learning_rate": 7.378768020969856e-06,
      "loss": 0.9305,
      "step": 200
    },
    {
      "epoch": 0.2634338138925295,
      "grad_norm": 0.111328125,
      "learning_rate": 7.365661861074706e-06,
      "loss": 0.9688,
      "step": 201
    },
    {
      "epoch": 0.26474442988204455,
      "grad_norm": 0.126953125,
      "learning_rate": 7.352555701179555e-06,
      "loss": 1.0586,
      "step": 202
    },
    {
      "epoch": 0.26605504587155965,
      "grad_norm": 0.09814453125,
      "learning_rate": 7.3394495412844045e-06,
      "loss": 0.8897,
      "step": 203
    },
    {
      "epoch": 0.2673656618610747,
      "grad_norm": 0.0869140625,
      "learning_rate": 7.326343381389253e-06,
      "loss": 0.8846,
      "step": 204
    },
    {
      "epoch": 0.2686762778505898,
      "grad_norm": 0.10302734375,
      "learning_rate": 7.313237221494103e-06,
      "loss": 0.9742,
      "step": 205
    },
    {
      "epoch": 0.2699868938401048,
      "grad_norm": 0.0947265625,
      "learning_rate": 7.300131061598952e-06,
      "loss": 0.9316,
      "step": 206
    },
    {
      "epoch": 0.2712975098296199,
      "grad_norm": 0.115234375,
      "learning_rate": 7.287024901703801e-06,
      "loss": 1.0154,
      "step": 207
    },
    {
      "epoch": 0.272608125819135,
      "grad_norm": 0.09765625,
      "learning_rate": 7.273918741808651e-06,
      "loss": 0.9601,
      "step": 208
    },
    {
      "epoch": 0.27391874180865006,
      "grad_norm": 0.08984375,
      "learning_rate": 7.2608125819135e-06,
      "loss": 0.8991,
      "step": 209
    },
    {
      "epoch": 0.27522935779816515,
      "grad_norm": 0.09912109375,
      "learning_rate": 7.247706422018349e-06,
      "loss": 0.936,
      "step": 210
    },
    {
      "epoch": 0.2765399737876802,
      "grad_norm": 0.1044921875,
      "learning_rate": 7.234600262123198e-06,
      "loss": 0.9497,
      "step": 211
    },
    {
      "epoch": 0.2778505897771953,
      "grad_norm": 0.10595703125,
      "learning_rate": 7.221494102228049e-06,
      "loss": 0.9594,
      "step": 212
    },
    {
      "epoch": 0.27916120576671033,
      "grad_norm": 0.095703125,
      "learning_rate": 7.2083879423328975e-06,
      "loss": 0.8151,
      "step": 213
    },
    {
      "epoch": 0.2804718217562254,
      "grad_norm": 0.11474609375,
      "learning_rate": 7.195281782437746e-06,
      "loss": 0.9903,
      "step": 214
    },
    {
      "epoch": 0.2817824377457405,
      "grad_norm": 0.10888671875,
      "learning_rate": 7.182175622542595e-06,
      "loss": 0.9934,
      "step": 215
    },
    {
      "epoch": 0.28309305373525556,
      "grad_norm": 0.10009765625,
      "learning_rate": 7.169069462647444e-06,
      "loss": 0.8955,
      "step": 216
    },
    {
      "epoch": 0.28440366972477066,
      "grad_norm": 0.0869140625,
      "learning_rate": 7.155963302752295e-06,
      "loss": 0.8998,
      "step": 217
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.09228515625,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.9119,
      "step": 218
    },
    {
      "epoch": 0.2870249017038008,
      "grad_norm": 0.0966796875,
      "learning_rate": 7.1297509829619924e-06,
      "loss": 0.9361,
      "step": 219
    },
    {
      "epoch": 0.28833551769331583,
      "grad_norm": 0.1005859375,
      "learning_rate": 7.116644823066841e-06,
      "loss": 0.9309,
      "step": 220
    },
    {
      "epoch": 0.28964613368283093,
      "grad_norm": 0.11083984375,
      "learning_rate": 7.103538663171692e-06,
      "loss": 0.9438,
      "step": 221
    },
    {
      "epoch": 0.290956749672346,
      "grad_norm": 0.08935546875,
      "learning_rate": 7.090432503276541e-06,
      "loss": 0.8638,
      "step": 222
    },
    {
      "epoch": 0.29226736566186107,
      "grad_norm": 0.09326171875,
      "learning_rate": 7.07732634338139e-06,
      "loss": 0.8844,
      "step": 223
    },
    {
      "epoch": 0.29357798165137616,
      "grad_norm": 0.10205078125,
      "learning_rate": 7.0642201834862385e-06,
      "loss": 0.9305,
      "step": 224
    },
    {
      "epoch": 0.2948885976408912,
      "grad_norm": 0.1123046875,
      "learning_rate": 7.051114023591088e-06,
      "loss": 0.9847,
      "step": 225
    },
    {
      "epoch": 0.2961992136304063,
      "grad_norm": 0.12890625,
      "learning_rate": 7.038007863695938e-06,
      "loss": 0.9791,
      "step": 226
    },
    {
      "epoch": 0.29750982961992134,
      "grad_norm": 0.099609375,
      "learning_rate": 7.024901703800787e-06,
      "loss": 0.9613,
      "step": 227
    },
    {
      "epoch": 0.29882044560943644,
      "grad_norm": 0.083984375,
      "learning_rate": 7.011795543905637e-06,
      "loss": 0.8941,
      "step": 228
    },
    {
      "epoch": 0.30013106159895153,
      "grad_norm": 0.09912109375,
      "learning_rate": 6.9986893840104855e-06,
      "loss": 0.9126,
      "step": 229
    },
    {
      "epoch": 0.30144167758846657,
      "grad_norm": 0.10205078125,
      "learning_rate": 6.985583224115334e-06,
      "loss": 0.942,
      "step": 230
    },
    {
      "epoch": 0.30275229357798167,
      "grad_norm": 0.10595703125,
      "learning_rate": 6.972477064220184e-06,
      "loss": 0.917,
      "step": 231
    },
    {
      "epoch": 0.3040629095674967,
      "grad_norm": 0.11376953125,
      "learning_rate": 6.959370904325034e-06,
      "loss": 0.9761,
      "step": 232
    },
    {
      "epoch": 0.3053735255570118,
      "grad_norm": 0.109375,
      "learning_rate": 6.946264744429883e-06,
      "loss": 0.9472,
      "step": 233
    },
    {
      "epoch": 0.30668414154652685,
      "grad_norm": 0.10546875,
      "learning_rate": 6.9331585845347315e-06,
      "loss": 0.9342,
      "step": 234
    },
    {
      "epoch": 0.30799475753604194,
      "grad_norm": 0.1005859375,
      "learning_rate": 6.920052424639582e-06,
      "loss": 0.9272,
      "step": 235
    },
    {
      "epoch": 0.30930537352555704,
      "grad_norm": 0.09716796875,
      "learning_rate": 6.906946264744431e-06,
      "loss": 0.9006,
      "step": 236
    },
    {
      "epoch": 0.3106159895150721,
      "grad_norm": 0.08935546875,
      "learning_rate": 6.89384010484928e-06,
      "loss": 0.8775,
      "step": 237
    },
    {
      "epoch": 0.3119266055045872,
      "grad_norm": 0.1015625,
      "learning_rate": 6.880733944954129e-06,
      "loss": 0.93,
      "step": 238
    },
    {
      "epoch": 0.3132372214941022,
      "grad_norm": 0.1044921875,
      "learning_rate": 6.867627785058978e-06,
      "loss": 0.9344,
      "step": 239
    },
    {
      "epoch": 0.3145478374836173,
      "grad_norm": 0.08349609375,
      "learning_rate": 6.854521625163828e-06,
      "loss": 0.8809,
      "step": 240
    },
    {
      "epoch": 0.31585845347313235,
      "grad_norm": 0.1015625,
      "learning_rate": 6.841415465268677e-06,
      "loss": 0.9134,
      "step": 241
    },
    {
      "epoch": 0.31716906946264745,
      "grad_norm": 0.109375,
      "learning_rate": 6.828309305373526e-06,
      "loss": 0.9409,
      "step": 242
    },
    {
      "epoch": 0.31847968545216254,
      "grad_norm": 0.119140625,
      "learning_rate": 6.815203145478375e-06,
      "loss": 0.9688,
      "step": 243
    },
    {
      "epoch": 0.3197903014416776,
      "grad_norm": 0.10009765625,
      "learning_rate": 6.8020969855832246e-06,
      "loss": 0.9215,
      "step": 244
    },
    {
      "epoch": 0.3211009174311927,
      "grad_norm": 0.111328125,
      "learning_rate": 6.788990825688074e-06,
      "loss": 0.9651,
      "step": 245
    },
    {
      "epoch": 0.3224115334207077,
      "grad_norm": 0.10888671875,
      "learning_rate": 6.775884665792923e-06,
      "loss": 0.9618,
      "step": 246
    },
    {
      "epoch": 0.3237221494102228,
      "grad_norm": 0.10400390625,
      "learning_rate": 6.762778505897772e-06,
      "loss": 0.9181,
      "step": 247
    },
    {
      "epoch": 0.32503276539973786,
      "grad_norm": 0.09814453125,
      "learning_rate": 6.749672346002622e-06,
      "loss": 0.8983,
      "step": 248
    },
    {
      "epoch": 0.32634338138925295,
      "grad_norm": 0.10986328125,
      "learning_rate": 6.7365661861074715e-06,
      "loss": 0.8959,
      "step": 249
    },
    {
      "epoch": 0.32765399737876805,
      "grad_norm": 0.09814453125,
      "learning_rate": 6.72346002621232e-06,
      "loss": 0.8851,
      "step": 250
    },
    {
      "epoch": 0.3289646133682831,
      "grad_norm": 0.10498046875,
      "learning_rate": 6.710353866317169e-06,
      "loss": 0.8834,
      "step": 251
    },
    {
      "epoch": 0.3302752293577982,
      "grad_norm": 0.111328125,
      "learning_rate": 6.697247706422019e-06,
      "loss": 0.9838,
      "step": 252
    },
    {
      "epoch": 0.3315858453473132,
      "grad_norm": 0.09423828125,
      "learning_rate": 6.684141546526868e-06,
      "loss": 0.9082,
      "step": 253
    },
    {
      "epoch": 0.3328964613368283,
      "grad_norm": 0.09375,
      "learning_rate": 6.671035386631718e-06,
      "loss": 0.9052,
      "step": 254
    },
    {
      "epoch": 0.33420707732634336,
      "grad_norm": 0.1005859375,
      "learning_rate": 6.657929226736567e-06,
      "loss": 0.921,
      "step": 255
    },
    {
      "epoch": 0.33551769331585846,
      "grad_norm": 0.09375,
      "learning_rate": 6.644823066841416e-06,
      "loss": 0.912,
      "step": 256
    },
    {
      "epoch": 0.33682830930537355,
      "grad_norm": 0.09765625,
      "learning_rate": 6.631716906946265e-06,
      "loss": 0.9113,
      "step": 257
    },
    {
      "epoch": 0.3381389252948886,
      "grad_norm": 0.1123046875,
      "learning_rate": 6.618610747051114e-06,
      "loss": 0.92,
      "step": 258
    },
    {
      "epoch": 0.3394495412844037,
      "grad_norm": 0.09912109375,
      "learning_rate": 6.6055045871559645e-06,
      "loss": 0.9471,
      "step": 259
    },
    {
      "epoch": 0.34076015727391873,
      "grad_norm": 0.09716796875,
      "learning_rate": 6.592398427260813e-06,
      "loss": 0.9197,
      "step": 260
    },
    {
      "epoch": 0.3420707732634338,
      "grad_norm": 0.0927734375,
      "learning_rate": 6.579292267365662e-06,
      "loss": 0.8569,
      "step": 261
    },
    {
      "epoch": 0.34338138925294887,
      "grad_norm": 0.119140625,
      "learning_rate": 6.566186107470511e-06,
      "loss": 0.8676,
      "step": 262
    },
    {
      "epoch": 0.34469200524246396,
      "grad_norm": 0.11083984375,
      "learning_rate": 6.553079947575362e-06,
      "loss": 0.9409,
      "step": 263
    },
    {
      "epoch": 0.34600262123197906,
      "grad_norm": 0.115234375,
      "learning_rate": 6.539973787680211e-06,
      "loss": 0.9069,
      "step": 264
    },
    {
      "epoch": 0.3473132372214941,
      "grad_norm": 0.0908203125,
      "learning_rate": 6.5268676277850595e-06,
      "loss": 0.9058,
      "step": 265
    },
    {
      "epoch": 0.3486238532110092,
      "grad_norm": 0.10693359375,
      "learning_rate": 6.513761467889908e-06,
      "loss": 0.9314,
      "step": 266
    },
    {
      "epoch": 0.34993446920052423,
      "grad_norm": 0.09912109375,
      "learning_rate": 6.500655307994757e-06,
      "loss": 0.919,
      "step": 267
    },
    {
      "epoch": 0.35124508519003933,
      "grad_norm": 0.103515625,
      "learning_rate": 6.487549148099608e-06,
      "loss": 0.9022,
      "step": 268
    },
    {
      "epoch": 0.35255570117955437,
      "grad_norm": 0.1015625,
      "learning_rate": 6.474442988204457e-06,
      "loss": 0.873,
      "step": 269
    },
    {
      "epoch": 0.35386631716906947,
      "grad_norm": 0.0986328125,
      "learning_rate": 6.4613368283093056e-06,
      "loss": 0.8797,
      "step": 270
    },
    {
      "epoch": 0.35517693315858456,
      "grad_norm": 0.10546875,
      "learning_rate": 6.448230668414155e-06,
      "loss": 0.9295,
      "step": 271
    },
    {
      "epoch": 0.3564875491480996,
      "grad_norm": 0.09130859375,
      "learning_rate": 6.435124508519004e-06,
      "loss": 0.9065,
      "step": 272
    },
    {
      "epoch": 0.3577981651376147,
      "grad_norm": 0.08544921875,
      "learning_rate": 6.422018348623854e-06,
      "loss": 0.8883,
      "step": 273
    },
    {
      "epoch": 0.35910878112712974,
      "grad_norm": 0.10791015625,
      "learning_rate": 6.408912188728703e-06,
      "loss": 0.949,
      "step": 274
    },
    {
      "epoch": 0.36041939711664484,
      "grad_norm": 0.1103515625,
      "learning_rate": 6.3958060288335525e-06,
      "loss": 1.0008,
      "step": 275
    },
    {
      "epoch": 0.3617300131061599,
      "grad_norm": 0.10009765625,
      "learning_rate": 6.382699868938401e-06,
      "loss": 0.9013,
      "step": 276
    },
    {
      "epoch": 0.36304062909567497,
      "grad_norm": 0.1103515625,
      "learning_rate": 6.369593709043251e-06,
      "loss": 0.8853,
      "step": 277
    },
    {
      "epoch": 0.36435124508519,
      "grad_norm": 0.09814453125,
      "learning_rate": 6.3564875491481e-06,
      "loss": 0.8884,
      "step": 278
    },
    {
      "epoch": 0.3656618610747051,
      "grad_norm": 0.0927734375,
      "learning_rate": 6.34338138925295e-06,
      "loss": 0.8861,
      "step": 279
    },
    {
      "epoch": 0.3669724770642202,
      "grad_norm": 0.1201171875,
      "learning_rate": 6.330275229357799e-06,
      "loss": 0.9303,
      "step": 280
    },
    {
      "epoch": 0.36828309305373524,
      "grad_norm": 0.10302734375,
      "learning_rate": 6.3171690694626475e-06,
      "loss": 0.9221,
      "step": 281
    },
    {
      "epoch": 0.36959370904325034,
      "grad_norm": 0.0830078125,
      "learning_rate": 6.304062909567498e-06,
      "loss": 0.8284,
      "step": 282
    },
    {
      "epoch": 0.3709043250327654,
      "grad_norm": 0.10009765625,
      "learning_rate": 6.290956749672347e-06,
      "loss": 0.8977,
      "step": 283
    },
    {
      "epoch": 0.3722149410222805,
      "grad_norm": 0.44921875,
      "learning_rate": 6.277850589777196e-06,
      "loss": 0.9326,
      "step": 284
    },
    {
      "epoch": 0.3735255570117955,
      "grad_norm": 0.12255859375,
      "learning_rate": 6.264744429882045e-06,
      "loss": 0.9963,
      "step": 285
    },
    {
      "epoch": 0.3748361730013106,
      "grad_norm": 0.091796875,
      "learning_rate": 6.2516382699868936e-06,
      "loss": 0.8851,
      "step": 286
    },
    {
      "epoch": 0.3761467889908257,
      "grad_norm": 0.10400390625,
      "learning_rate": 6.238532110091744e-06,
      "loss": 0.9229,
      "step": 287
    },
    {
      "epoch": 0.37745740498034075,
      "grad_norm": 0.08984375,
      "learning_rate": 6.225425950196593e-06,
      "loss": 0.8724,
      "step": 288
    },
    {
      "epoch": 0.37876802096985585,
      "grad_norm": 0.10595703125,
      "learning_rate": 6.212319790301442e-06,
      "loss": 0.8938,
      "step": 289
    },
    {
      "epoch": 0.3800786369593709,
      "grad_norm": 0.1025390625,
      "learning_rate": 6.199213630406291e-06,
      "loss": 0.8916,
      "step": 290
    },
    {
      "epoch": 0.381389252948886,
      "grad_norm": 0.12060546875,
      "learning_rate": 6.186107470511141e-06,
      "loss": 0.9642,
      "step": 291
    },
    {
      "epoch": 0.382699868938401,
      "grad_norm": 0.09716796875,
      "learning_rate": 6.17300131061599e-06,
      "loss": 0.9056,
      "step": 292
    },
    {
      "epoch": 0.3840104849279161,
      "grad_norm": 0.09521484375,
      "learning_rate": 6.159895150720839e-06,
      "loss": 0.834,
      "step": 293
    },
    {
      "epoch": 0.3853211009174312,
      "grad_norm": 0.0966796875,
      "learning_rate": 6.146788990825688e-06,
      "loss": 0.866,
      "step": 294
    },
    {
      "epoch": 0.38663171690694625,
      "grad_norm": 0.099609375,
      "learning_rate": 6.133682830930538e-06,
      "loss": 0.8584,
      "step": 295
    },
    {
      "epoch": 0.38794233289646135,
      "grad_norm": 0.1025390625,
      "learning_rate": 6.120576671035387e-06,
      "loss": 0.8973,
      "step": 296
    },
    {
      "epoch": 0.3892529488859764,
      "grad_norm": 0.10888671875,
      "learning_rate": 6.107470511140236e-06,
      "loss": 0.9327,
      "step": 297
    },
    {
      "epoch": 0.3905635648754915,
      "grad_norm": 0.09765625,
      "learning_rate": 6.094364351245086e-06,
      "loss": 0.8952,
      "step": 298
    },
    {
      "epoch": 0.3918741808650065,
      "grad_norm": 0.08251953125,
      "learning_rate": 6.081258191349935e-06,
      "loss": 0.8542,
      "step": 299
    },
    {
      "epoch": 0.3931847968545216,
      "grad_norm": 0.09814453125,
      "learning_rate": 6.068152031454784e-06,
      "loss": 0.8459,
      "step": 300
    },
    {
      "epoch": 0.3944954128440367,
      "grad_norm": 0.12060546875,
      "learning_rate": 6.0550458715596335e-06,
      "loss": 0.9723,
      "step": 301
    },
    {
      "epoch": 0.39580602883355176,
      "grad_norm": 0.0859375,
      "learning_rate": 6.041939711664483e-06,
      "loss": 0.8052,
      "step": 302
    },
    {
      "epoch": 0.39711664482306686,
      "grad_norm": 0.1015625,
      "learning_rate": 6.028833551769332e-06,
      "loss": 0.8934,
      "step": 303
    },
    {
      "epoch": 0.3984272608125819,
      "grad_norm": 0.09423828125,
      "learning_rate": 6.015727391874181e-06,
      "loss": 0.8904,
      "step": 304
    },
    {
      "epoch": 0.399737876802097,
      "grad_norm": 0.08447265625,
      "learning_rate": 6.002621231979031e-06,
      "loss": 0.8462,
      "step": 305
    },
    {
      "epoch": 0.40104849279161203,
      "grad_norm": 0.09375,
      "learning_rate": 5.9895150720838804e-06,
      "loss": 0.8518,
      "step": 306
    },
    {
      "epoch": 0.40235910878112713,
      "grad_norm": 0.08642578125,
      "learning_rate": 5.976408912188729e-06,
      "loss": 0.8333,
      "step": 307
    },
    {
      "epoch": 0.4036697247706422,
      "grad_norm": 0.10791015625,
      "learning_rate": 5.963302752293578e-06,
      "loss": 0.9149,
      "step": 308
    },
    {
      "epoch": 0.40498034076015726,
      "grad_norm": 0.10302734375,
      "learning_rate": 5.950196592398427e-06,
      "loss": 0.9044,
      "step": 309
    },
    {
      "epoch": 0.40629095674967236,
      "grad_norm": 0.08544921875,
      "learning_rate": 5.937090432503278e-06,
      "loss": 0.8535,
      "step": 310
    },
    {
      "epoch": 0.4076015727391874,
      "grad_norm": 0.087890625,
      "learning_rate": 5.9239842726081265e-06,
      "loss": 0.8441,
      "step": 311
    },
    {
      "epoch": 0.4089121887287025,
      "grad_norm": 0.09228515625,
      "learning_rate": 5.910878112712975e-06,
      "loss": 0.9094,
      "step": 312
    },
    {
      "epoch": 0.41022280471821754,
      "grad_norm": 0.1044921875,
      "learning_rate": 5.897771952817824e-06,
      "loss": 0.9137,
      "step": 313
    },
    {
      "epoch": 0.41153342070773263,
      "grad_norm": 0.0927734375,
      "learning_rate": 5.884665792922674e-06,
      "loss": 0.8653,
      "step": 314
    },
    {
      "epoch": 0.41284403669724773,
      "grad_norm": 0.09423828125,
      "learning_rate": 5.871559633027524e-06,
      "loss": 0.8478,
      "step": 315
    },
    {
      "epoch": 0.41415465268676277,
      "grad_norm": 0.1064453125,
      "learning_rate": 5.858453473132373e-06,
      "loss": 0.9632,
      "step": 316
    },
    {
      "epoch": 0.41546526867627787,
      "grad_norm": 0.10498046875,
      "learning_rate": 5.8453473132372215e-06,
      "loss": 0.9283,
      "step": 317
    },
    {
      "epoch": 0.4167758846657929,
      "grad_norm": 0.095703125,
      "learning_rate": 5.832241153342071e-06,
      "loss": 0.8875,
      "step": 318
    },
    {
      "epoch": 0.418086500655308,
      "grad_norm": 0.1728515625,
      "learning_rate": 5.819134993446921e-06,
      "loss": 0.9899,
      "step": 319
    },
    {
      "epoch": 0.41939711664482304,
      "grad_norm": 0.08837890625,
      "learning_rate": 5.80602883355177e-06,
      "loss": 0.8889,
      "step": 320
    },
    {
      "epoch": 0.42070773263433814,
      "grad_norm": 0.1015625,
      "learning_rate": 5.792922673656619e-06,
      "loss": 0.89,
      "step": 321
    },
    {
      "epoch": 0.42201834862385323,
      "grad_norm": 0.09765625,
      "learning_rate": 5.7798165137614684e-06,
      "loss": 0.845,
      "step": 322
    },
    {
      "epoch": 0.4233289646133683,
      "grad_norm": 0.0966796875,
      "learning_rate": 5.766710353866317e-06,
      "loss": 0.8742,
      "step": 323
    },
    {
      "epoch": 0.42463958060288337,
      "grad_norm": 0.09228515625,
      "learning_rate": 5.753604193971167e-06,
      "loss": 0.8777,
      "step": 324
    },
    {
      "epoch": 0.4259501965923984,
      "grad_norm": 0.1025390625,
      "learning_rate": 5.740498034076017e-06,
      "loss": 0.8747,
      "step": 325
    },
    {
      "epoch": 0.4272608125819135,
      "grad_norm": 0.11962890625,
      "learning_rate": 5.727391874180866e-06,
      "loss": 0.8858,
      "step": 326
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.0908203125,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.8502,
      "step": 327
    },
    {
      "epoch": 0.42988204456094364,
      "grad_norm": 0.09912109375,
      "learning_rate": 5.701179554390564e-06,
      "loss": 0.8953,
      "step": 328
    },
    {
      "epoch": 0.43119266055045874,
      "grad_norm": 0.10888671875,
      "learning_rate": 5.688073394495414e-06,
      "loss": 0.8451,
      "step": 329
    },
    {
      "epoch": 0.4325032765399738,
      "grad_norm": 0.0986328125,
      "learning_rate": 5.674967234600263e-06,
      "loss": 0.9342,
      "step": 330
    },
    {
      "epoch": 0.4338138925294889,
      "grad_norm": 0.09619140625,
      "learning_rate": 5.661861074705112e-06,
      "loss": 0.8322,
      "step": 331
    },
    {
      "epoch": 0.4351245085190039,
      "grad_norm": 0.1103515625,
      "learning_rate": 5.648754914809961e-06,
      "loss": 0.9098,
      "step": 332
    },
    {
      "epoch": 0.436435124508519,
      "grad_norm": 0.09814453125,
      "learning_rate": 5.635648754914811e-06,
      "loss": 0.8983,
      "step": 333
    },
    {
      "epoch": 0.43774574049803405,
      "grad_norm": 0.107421875,
      "learning_rate": 5.62254259501966e-06,
      "loss": 0.8822,
      "step": 334
    },
    {
      "epoch": 0.43905635648754915,
      "grad_norm": 0.11865234375,
      "learning_rate": 5.609436435124509e-06,
      "loss": 0.9757,
      "step": 335
    },
    {
      "epoch": 0.44036697247706424,
      "grad_norm": 0.09326171875,
      "learning_rate": 5.596330275229358e-06,
      "loss": 0.8545,
      "step": 336
    },
    {
      "epoch": 0.4416775884665793,
      "grad_norm": 0.09375,
      "learning_rate": 5.583224115334207e-06,
      "loss": 0.8864,
      "step": 337
    },
    {
      "epoch": 0.4429882044560944,
      "grad_norm": 0.09716796875,
      "learning_rate": 5.570117955439057e-06,
      "loss": 0.8633,
      "step": 338
    },
    {
      "epoch": 0.4442988204456094,
      "grad_norm": 0.10009765625,
      "learning_rate": 5.557011795543906e-06,
      "loss": 0.8907,
      "step": 339
    },
    {
      "epoch": 0.4456094364351245,
      "grad_norm": 0.1015625,
      "learning_rate": 5.543905635648755e-06,
      "loss": 0.9361,
      "step": 340
    },
    {
      "epoch": 0.44692005242463956,
      "grad_norm": 0.10302734375,
      "learning_rate": 5.530799475753605e-06,
      "loss": 0.8908,
      "step": 341
    },
    {
      "epoch": 0.44823066841415465,
      "grad_norm": 0.08984375,
      "learning_rate": 5.5176933158584545e-06,
      "loss": 0.8725,
      "step": 342
    },
    {
      "epoch": 0.44954128440366975,
      "grad_norm": 0.0869140625,
      "learning_rate": 5.504587155963303e-06,
      "loss": 0.8167,
      "step": 343
    },
    {
      "epoch": 0.4508519003931848,
      "grad_norm": 0.1005859375,
      "learning_rate": 5.491480996068152e-06,
      "loss": 0.9276,
      "step": 344
    },
    {
      "epoch": 0.4521625163826999,
      "grad_norm": 0.08984375,
      "learning_rate": 5.478374836173002e-06,
      "loss": 0.8772,
      "step": 345
    },
    {
      "epoch": 0.4534731323722149,
      "grad_norm": 0.0869140625,
      "learning_rate": 5.465268676277851e-06,
      "loss": 0.8358,
      "step": 346
    },
    {
      "epoch": 0.45478374836173,
      "grad_norm": 0.08837890625,
      "learning_rate": 5.4521625163827006e-06,
      "loss": 0.8637,
      "step": 347
    },
    {
      "epoch": 0.45609436435124506,
      "grad_norm": 0.09326171875,
      "learning_rate": 5.4390563564875494e-06,
      "loss": 0.9064,
      "step": 348
    },
    {
      "epoch": 0.45740498034076016,
      "grad_norm": 0.10009765625,
      "learning_rate": 5.425950196592399e-06,
      "loss": 0.8661,
      "step": 349
    },
    {
      "epoch": 0.45871559633027525,
      "grad_norm": 0.0927734375,
      "learning_rate": 5.412844036697248e-06,
      "loss": 0.847,
      "step": 350
    },
    {
      "epoch": 0.4600262123197903,
      "grad_norm": 0.0927734375,
      "learning_rate": 5.399737876802097e-06,
      "loss": 0.86,
      "step": 351
    },
    {
      "epoch": 0.4613368283093054,
      "grad_norm": 0.10302734375,
      "learning_rate": 5.3866317169069475e-06,
      "loss": 0.8319,
      "step": 352
    },
    {
      "epoch": 0.46264744429882043,
      "grad_norm": 0.0888671875,
      "learning_rate": 5.373525557011796e-06,
      "loss": 0.8655,
      "step": 353
    },
    {
      "epoch": 0.4639580602883355,
      "grad_norm": 0.08984375,
      "learning_rate": 5.360419397116645e-06,
      "loss": 0.8668,
      "step": 354
    },
    {
      "epoch": 0.46526867627785057,
      "grad_norm": 0.087890625,
      "learning_rate": 5.347313237221494e-06,
      "loss": 0.8541,
      "step": 355
    },
    {
      "epoch": 0.46657929226736566,
      "grad_norm": 0.08642578125,
      "learning_rate": 5.334207077326345e-06,
      "loss": 0.8526,
      "step": 356
    },
    {
      "epoch": 0.46788990825688076,
      "grad_norm": 0.0859375,
      "learning_rate": 5.3211009174311936e-06,
      "loss": 0.836,
      "step": 357
    },
    {
      "epoch": 0.4692005242463958,
      "grad_norm": 0.115234375,
      "learning_rate": 5.3079947575360424e-06,
      "loss": 0.8742,
      "step": 358
    },
    {
      "epoch": 0.4705111402359109,
      "grad_norm": 0.09423828125,
      "learning_rate": 5.294888597640891e-06,
      "loss": 0.8559,
      "step": 359
    },
    {
      "epoch": 0.47182175622542594,
      "grad_norm": 0.11328125,
      "learning_rate": 5.28178243774574e-06,
      "loss": 0.8729,
      "step": 360
    },
    {
      "epoch": 0.47313237221494103,
      "grad_norm": 0.0859375,
      "learning_rate": 5.268676277850591e-06,
      "loss": 0.8849,
      "step": 361
    },
    {
      "epoch": 0.4744429882044561,
      "grad_norm": 0.11328125,
      "learning_rate": 5.25557011795544e-06,
      "loss": 0.9169,
      "step": 362
    },
    {
      "epoch": 0.47575360419397117,
      "grad_norm": 0.09521484375,
      "learning_rate": 5.2424639580602885e-06,
      "loss": 0.911,
      "step": 363
    },
    {
      "epoch": 0.47706422018348627,
      "grad_norm": 0.095703125,
      "learning_rate": 5.229357798165137e-06,
      "loss": 0.9222,
      "step": 364
    },
    {
      "epoch": 0.4783748361730013,
      "grad_norm": 0.11181640625,
      "learning_rate": 5.216251638269987e-06,
      "loss": 0.9473,
      "step": 365
    },
    {
      "epoch": 0.4796854521625164,
      "grad_norm": 0.10546875,
      "learning_rate": 5.203145478374837e-06,
      "loss": 0.8922,
      "step": 366
    },
    {
      "epoch": 0.48099606815203144,
      "grad_norm": 0.09521484375,
      "learning_rate": 5.190039318479686e-06,
      "loss": 0.8529,
      "step": 367
    },
    {
      "epoch": 0.48230668414154654,
      "grad_norm": 0.08984375,
      "learning_rate": 5.1769331585845355e-06,
      "loss": 0.8587,
      "step": 368
    },
    {
      "epoch": 0.4836173001310616,
      "grad_norm": 0.0830078125,
      "learning_rate": 5.163826998689384e-06,
      "loss": 0.8101,
      "step": 369
    },
    {
      "epoch": 0.4849279161205767,
      "grad_norm": 0.10595703125,
      "learning_rate": 5.150720838794234e-06,
      "loss": 0.8956,
      "step": 370
    },
    {
      "epoch": 0.48623853211009177,
      "grad_norm": 0.08349609375,
      "learning_rate": 5.137614678899083e-06,
      "loss": 0.793,
      "step": 371
    },
    {
      "epoch": 0.4875491480996068,
      "grad_norm": 0.10107421875,
      "learning_rate": 5.124508519003933e-06,
      "loss": 0.901,
      "step": 372
    },
    {
      "epoch": 0.4888597640891219,
      "grad_norm": 0.09326171875,
      "learning_rate": 5.1114023591087816e-06,
      "loss": 0.8753,
      "step": 373
    },
    {
      "epoch": 0.49017038007863695,
      "grad_norm": 0.0888671875,
      "learning_rate": 5.0982961992136304e-06,
      "loss": 0.8777,
      "step": 374
    },
    {
      "epoch": 0.49148099606815204,
      "grad_norm": 0.09521484375,
      "learning_rate": 5.08519003931848e-06,
      "loss": 0.8757,
      "step": 375
    },
    {
      "epoch": 0.4927916120576671,
      "grad_norm": 0.08837890625,
      "learning_rate": 5.07208387942333e-06,
      "loss": 0.8753,
      "step": 376
    },
    {
      "epoch": 0.4941022280471822,
      "grad_norm": 0.08642578125,
      "learning_rate": 5.058977719528179e-06,
      "loss": 0.8613,
      "step": 377
    },
    {
      "epoch": 0.4954128440366973,
      "grad_norm": 0.08544921875,
      "learning_rate": 5.045871559633028e-06,
      "loss": 0.8793,
      "step": 378
    },
    {
      "epoch": 0.4967234600262123,
      "grad_norm": 0.09375,
      "learning_rate": 5.0327653997378765e-06,
      "loss": 0.8717,
      "step": 379
    },
    {
      "epoch": 0.4980340760157274,
      "grad_norm": 0.08984375,
      "learning_rate": 5.019659239842727e-06,
      "loss": 0.83,
      "step": 380
    },
    {
      "epoch": 0.49934469200524245,
      "grad_norm": 0.0966796875,
      "learning_rate": 5.006553079947576e-06,
      "loss": 0.8717,
      "step": 381
    },
    {
      "epoch": 0.5006553079947575,
      "grad_norm": 0.1005859375,
      "learning_rate": 4.993446920052425e-06,
      "loss": 0.9068,
      "step": 382
    },
    {
      "epoch": 0.5019659239842726,
      "grad_norm": 0.09375,
      "learning_rate": 4.9803407601572746e-06,
      "loss": 0.8594,
      "step": 383
    },
    {
      "epoch": 0.5032765399737876,
      "grad_norm": 0.08349609375,
      "learning_rate": 4.9672346002621235e-06,
      "loss": 0.8479,
      "step": 384
    },
    {
      "epoch": 0.5045871559633027,
      "grad_norm": 0.130859375,
      "learning_rate": 4.954128440366973e-06,
      "loss": 0.8989,
      "step": 385
    },
    {
      "epoch": 0.5058977719528178,
      "grad_norm": 0.1044921875,
      "learning_rate": 4.941022280471822e-06,
      "loss": 0.8959,
      "step": 386
    },
    {
      "epoch": 0.5072083879423329,
      "grad_norm": 0.08056640625,
      "learning_rate": 4.927916120576671e-06,
      "loss": 0.8326,
      "step": 387
    },
    {
      "epoch": 0.508519003931848,
      "grad_norm": 0.09375,
      "learning_rate": 4.914809960681521e-06,
      "loss": 0.8961,
      "step": 388
    },
    {
      "epoch": 0.509829619921363,
      "grad_norm": 0.10009765625,
      "learning_rate": 4.9017038007863695e-06,
      "loss": 0.8708,
      "step": 389
    },
    {
      "epoch": 0.5111402359108781,
      "grad_norm": 0.0927734375,
      "learning_rate": 4.888597640891219e-06,
      "loss": 0.8549,
      "step": 390
    },
    {
      "epoch": 0.5124508519003932,
      "grad_norm": 0.109375,
      "learning_rate": 4.875491480996068e-06,
      "loss": 0.8987,
      "step": 391
    },
    {
      "epoch": 0.5137614678899083,
      "grad_norm": 0.0966796875,
      "learning_rate": 4.862385321100918e-06,
      "loss": 0.8615,
      "step": 392
    },
    {
      "epoch": 0.5150720838794234,
      "grad_norm": 0.1044921875,
      "learning_rate": 4.849279161205767e-06,
      "loss": 0.8803,
      "step": 393
    },
    {
      "epoch": 0.5163826998689384,
      "grad_norm": 0.08056640625,
      "learning_rate": 4.8361730013106165e-06,
      "loss": 0.87,
      "step": 394
    },
    {
      "epoch": 0.5176933158584535,
      "grad_norm": 0.130859375,
      "learning_rate": 4.823066841415465e-06,
      "loss": 1.0097,
      "step": 395
    },
    {
      "epoch": 0.5190039318479686,
      "grad_norm": 0.0947265625,
      "learning_rate": 4.809960681520315e-06,
      "loss": 0.8873,
      "step": 396
    },
    {
      "epoch": 0.5203145478374837,
      "grad_norm": 0.09130859375,
      "learning_rate": 4.796854521625165e-06,
      "loss": 0.88,
      "step": 397
    },
    {
      "epoch": 0.5216251638269986,
      "grad_norm": 0.0966796875,
      "learning_rate": 4.783748361730014e-06,
      "loss": 0.8896,
      "step": 398
    },
    {
      "epoch": 0.5229357798165137,
      "grad_norm": 0.0859375,
      "learning_rate": 4.770642201834863e-06,
      "loss": 0.8432,
      "step": 399
    },
    {
      "epoch": 0.5242463958060288,
      "grad_norm": 0.1025390625,
      "learning_rate": 4.757536041939712e-06,
      "loss": 0.9044,
      "step": 400
    },
    {
      "epoch": 0.5255570117955439,
      "grad_norm": 0.09716796875,
      "learning_rate": 4.744429882044561e-06,
      "loss": 0.8561,
      "step": 401
    },
    {
      "epoch": 0.526867627785059,
      "grad_norm": 0.1005859375,
      "learning_rate": 4.731323722149411e-06,
      "loss": 0.8843,
      "step": 402
    },
    {
      "epoch": 0.528178243774574,
      "grad_norm": 0.09765625,
      "learning_rate": 4.71821756225426e-06,
      "loss": 0.8224,
      "step": 403
    },
    {
      "epoch": 0.5294888597640891,
      "grad_norm": 0.08740234375,
      "learning_rate": 4.7051114023591095e-06,
      "loss": 0.8476,
      "step": 404
    },
    {
      "epoch": 0.5307994757536042,
      "grad_norm": 0.0869140625,
      "learning_rate": 4.692005242463958e-06,
      "loss": 0.8652,
      "step": 405
    },
    {
      "epoch": 0.5321100917431193,
      "grad_norm": 0.1044921875,
      "learning_rate": 4.678899082568808e-06,
      "loss": 0.8609,
      "step": 406
    },
    {
      "epoch": 0.5334207077326344,
      "grad_norm": 0.08935546875,
      "learning_rate": 4.665792922673657e-06,
      "loss": 0.8656,
      "step": 407
    },
    {
      "epoch": 0.5347313237221494,
      "grad_norm": 0.09130859375,
      "learning_rate": 4.652686762778506e-06,
      "loss": 0.8402,
      "step": 408
    },
    {
      "epoch": 0.5360419397116645,
      "grad_norm": 0.0810546875,
      "learning_rate": 4.639580602883356e-06,
      "loss": 0.8096,
      "step": 409
    },
    {
      "epoch": 0.5373525557011796,
      "grad_norm": 0.0966796875,
      "learning_rate": 4.6264744429882045e-06,
      "loss": 0.8851,
      "step": 410
    },
    {
      "epoch": 0.5386631716906947,
      "grad_norm": 0.08447265625,
      "learning_rate": 4.613368283093054e-06,
      "loss": 0.8061,
      "step": 411
    },
    {
      "epoch": 0.5399737876802096,
      "grad_norm": 0.08935546875,
      "learning_rate": 4.600262123197903e-06,
      "loss": 0.8385,
      "step": 412
    },
    {
      "epoch": 0.5412844036697247,
      "grad_norm": 0.095703125,
      "learning_rate": 4.587155963302753e-06,
      "loss": 0.8718,
      "step": 413
    },
    {
      "epoch": 0.5425950196592398,
      "grad_norm": 0.0849609375,
      "learning_rate": 4.574049803407602e-06,
      "loss": 0.8323,
      "step": 414
    },
    {
      "epoch": 0.5439056356487549,
      "grad_norm": 0.09912109375,
      "learning_rate": 4.560943643512451e-06,
      "loss": 0.9231,
      "step": 415
    },
    {
      "epoch": 0.54521625163827,
      "grad_norm": 0.099609375,
      "learning_rate": 4.5478374836173e-06,
      "loss": 0.8611,
      "step": 416
    },
    {
      "epoch": 0.546526867627785,
      "grad_norm": 0.10595703125,
      "learning_rate": 4.53473132372215e-06,
      "loss": 0.862,
      "step": 417
    },
    {
      "epoch": 0.5478374836173001,
      "grad_norm": 0.08447265625,
      "learning_rate": 4.521625163826999e-06,
      "loss": 0.8414,
      "step": 418
    },
    {
      "epoch": 0.5491480996068152,
      "grad_norm": 0.0849609375,
      "learning_rate": 4.508519003931849e-06,
      "loss": 0.8363,
      "step": 419
    },
    {
      "epoch": 0.5504587155963303,
      "grad_norm": 0.09375,
      "learning_rate": 4.4954128440366975e-06,
      "loss": 0.8394,
      "step": 420
    },
    {
      "epoch": 0.5517693315858454,
      "grad_norm": 0.10009765625,
      "learning_rate": 4.482306684141547e-06,
      "loss": 0.8798,
      "step": 421
    },
    {
      "epoch": 0.5530799475753604,
      "grad_norm": 0.08544921875,
      "learning_rate": 4.469200524246396e-06,
      "loss": 0.8672,
      "step": 422
    },
    {
      "epoch": 0.5543905635648755,
      "grad_norm": 0.1044921875,
      "learning_rate": 4.456094364351246e-06,
      "loss": 0.8717,
      "step": 423
    },
    {
      "epoch": 0.5557011795543906,
      "grad_norm": 0.0830078125,
      "learning_rate": 4.442988204456095e-06,
      "loss": 0.8614,
      "step": 424
    },
    {
      "epoch": 0.5570117955439057,
      "grad_norm": 0.09912109375,
      "learning_rate": 4.429882044560944e-06,
      "loss": 0.8832,
      "step": 425
    },
    {
      "epoch": 0.5583224115334207,
      "grad_norm": 0.08935546875,
      "learning_rate": 4.416775884665793e-06,
      "loss": 0.8659,
      "step": 426
    },
    {
      "epoch": 0.5596330275229358,
      "grad_norm": 0.08251953125,
      "learning_rate": 4.403669724770643e-06,
      "loss": 0.827,
      "step": 427
    },
    {
      "epoch": 0.5609436435124509,
      "grad_norm": 0.08740234375,
      "learning_rate": 4.390563564875492e-06,
      "loss": 0.8626,
      "step": 428
    },
    {
      "epoch": 0.562254259501966,
      "grad_norm": 0.09130859375,
      "learning_rate": 4.377457404980341e-06,
      "loss": 0.8755,
      "step": 429
    },
    {
      "epoch": 0.563564875491481,
      "grad_norm": 0.08984375,
      "learning_rate": 4.3643512450851905e-06,
      "loss": 0.8616,
      "step": 430
    },
    {
      "epoch": 0.564875491480996,
      "grad_norm": 0.1298828125,
      "learning_rate": 4.351245085190039e-06,
      "loss": 0.9583,
      "step": 431
    },
    {
      "epoch": 0.5661861074705111,
      "grad_norm": 0.09765625,
      "learning_rate": 4.338138925294889e-06,
      "loss": 0.8579,
      "step": 432
    },
    {
      "epoch": 0.5674967234600262,
      "grad_norm": 0.0908203125,
      "learning_rate": 4.325032765399738e-06,
      "loss": 0.8487,
      "step": 433
    },
    {
      "epoch": 0.5688073394495413,
      "grad_norm": 0.095703125,
      "learning_rate": 4.311926605504588e-06,
      "loss": 0.9197,
      "step": 434
    },
    {
      "epoch": 0.5701179554390564,
      "grad_norm": 0.0810546875,
      "learning_rate": 4.298820445609437e-06,
      "loss": 0.8562,
      "step": 435
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.09521484375,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.8741,
      "step": 436
    },
    {
      "epoch": 0.5727391874180865,
      "grad_norm": 0.0849609375,
      "learning_rate": 4.272608125819135e-06,
      "loss": 0.8329,
      "step": 437
    },
    {
      "epoch": 0.5740498034076016,
      "grad_norm": 0.0947265625,
      "learning_rate": 4.259501965923984e-06,
      "loss": 0.8682,
      "step": 438
    },
    {
      "epoch": 0.5753604193971167,
      "grad_norm": 0.0927734375,
      "learning_rate": 4.246395806028834e-06,
      "loss": 0.832,
      "step": 439
    },
    {
      "epoch": 0.5766710353866317,
      "grad_norm": 0.080078125,
      "learning_rate": 4.2332896461336835e-06,
      "loss": 0.7371,
      "step": 440
    },
    {
      "epoch": 0.5779816513761468,
      "grad_norm": 0.08642578125,
      "learning_rate": 4.220183486238532e-06,
      "loss": 0.787,
      "step": 441
    },
    {
      "epoch": 0.5792922673656619,
      "grad_norm": 0.091796875,
      "learning_rate": 4.207077326343382e-06,
      "loss": 0.8569,
      "step": 442
    },
    {
      "epoch": 0.580602883355177,
      "grad_norm": 0.09375,
      "learning_rate": 4.193971166448231e-06,
      "loss": 0.8624,
      "step": 443
    },
    {
      "epoch": 0.581913499344692,
      "grad_norm": 0.09716796875,
      "learning_rate": 4.180865006553081e-06,
      "loss": 0.8681,
      "step": 444
    },
    {
      "epoch": 0.583224115334207,
      "grad_norm": 0.09716796875,
      "learning_rate": 4.16775884665793e-06,
      "loss": 0.8535,
      "step": 445
    },
    {
      "epoch": 0.5845347313237221,
      "grad_norm": 0.0947265625,
      "learning_rate": 4.154652686762779e-06,
      "loss": 0.8643,
      "step": 446
    },
    {
      "epoch": 0.5858453473132372,
      "grad_norm": 0.099609375,
      "learning_rate": 4.141546526867628e-06,
      "loss": 0.8925,
      "step": 447
    },
    {
      "epoch": 0.5871559633027523,
      "grad_norm": 0.1064453125,
      "learning_rate": 4.128440366972478e-06,
      "loss": 0.8615,
      "step": 448
    },
    {
      "epoch": 0.5884665792922673,
      "grad_norm": 0.08544921875,
      "learning_rate": 4.115334207077327e-06,
      "loss": 0.8465,
      "step": 449
    },
    {
      "epoch": 0.5897771952817824,
      "grad_norm": 0.0830078125,
      "learning_rate": 4.102228047182176e-06,
      "loss": 0.7813,
      "step": 450
    },
    {
      "epoch": 0.5910878112712975,
      "grad_norm": 0.10302734375,
      "learning_rate": 4.089121887287025e-06,
      "loss": 0.8841,
      "step": 451
    },
    {
      "epoch": 0.5923984272608126,
      "grad_norm": 0.1025390625,
      "learning_rate": 4.076015727391874e-06,
      "loss": 0.9014,
      "step": 452
    },
    {
      "epoch": 0.5937090432503277,
      "grad_norm": 0.080078125,
      "learning_rate": 4.062909567496724e-06,
      "loss": 0.8079,
      "step": 453
    },
    {
      "epoch": 0.5950196592398427,
      "grad_norm": 0.09814453125,
      "learning_rate": 4.049803407601573e-06,
      "loss": 0.8983,
      "step": 454
    },
    {
      "epoch": 0.5963302752293578,
      "grad_norm": 0.08935546875,
      "learning_rate": 4.036697247706423e-06,
      "loss": 0.838,
      "step": 455
    },
    {
      "epoch": 0.5976408912188729,
      "grad_norm": 0.09033203125,
      "learning_rate": 4.0235910878112715e-06,
      "loss": 0.8754,
      "step": 456
    },
    {
      "epoch": 0.598951507208388,
      "grad_norm": 0.09765625,
      "learning_rate": 4.01048492791612e-06,
      "loss": 0.8545,
      "step": 457
    },
    {
      "epoch": 0.6002621231979031,
      "grad_norm": 0.09228515625,
      "learning_rate": 3.99737876802097e-06,
      "loss": 0.8233,
      "step": 458
    },
    {
      "epoch": 0.601572739187418,
      "grad_norm": 0.322265625,
      "learning_rate": 3.984272608125819e-06,
      "loss": 1.1009,
      "step": 459
    },
    {
      "epoch": 0.6028833551769331,
      "grad_norm": 0.10205078125,
      "learning_rate": 3.971166448230669e-06,
      "loss": 0.8428,
      "step": 460
    },
    {
      "epoch": 0.6041939711664482,
      "grad_norm": 0.09130859375,
      "learning_rate": 3.958060288335518e-06,
      "loss": 0.8053,
      "step": 461
    },
    {
      "epoch": 0.6055045871559633,
      "grad_norm": 0.1044921875,
      "learning_rate": 3.944954128440367e-06,
      "loss": 0.8954,
      "step": 462
    },
    {
      "epoch": 0.6068152031454783,
      "grad_norm": 0.0908203125,
      "learning_rate": 3.931847968545216e-06,
      "loss": 0.847,
      "step": 463
    },
    {
      "epoch": 0.6081258191349934,
      "grad_norm": 0.08203125,
      "learning_rate": 3.918741808650066e-06,
      "loss": 0.8027,
      "step": 464
    },
    {
      "epoch": 0.6094364351245085,
      "grad_norm": 0.1015625,
      "learning_rate": 3.905635648754915e-06,
      "loss": 0.8583,
      "step": 465
    },
    {
      "epoch": 0.6107470511140236,
      "grad_norm": 0.10400390625,
      "learning_rate": 3.8925294888597645e-06,
      "loss": 0.8867,
      "step": 466
    },
    {
      "epoch": 0.6120576671035387,
      "grad_norm": 0.0810546875,
      "learning_rate": 3.879423328964614e-06,
      "loss": 0.8511,
      "step": 467
    },
    {
      "epoch": 0.6133682830930537,
      "grad_norm": 0.08447265625,
      "learning_rate": 3.866317169069463e-06,
      "loss": 0.7994,
      "step": 468
    },
    {
      "epoch": 0.6146788990825688,
      "grad_norm": 0.08642578125,
      "learning_rate": 3.853211009174313e-06,
      "loss": 0.7958,
      "step": 469
    },
    {
      "epoch": 0.6159895150720839,
      "grad_norm": 0.09375,
      "learning_rate": 3.840104849279162e-06,
      "loss": 0.8538,
      "step": 470
    },
    {
      "epoch": 0.617300131061599,
      "grad_norm": 0.1142578125,
      "learning_rate": 3.826998689384011e-06,
      "loss": 0.8669,
      "step": 471
    },
    {
      "epoch": 0.6186107470511141,
      "grad_norm": 0.091796875,
      "learning_rate": 3.8138925294888603e-06,
      "loss": 0.844,
      "step": 472
    },
    {
      "epoch": 0.6199213630406291,
      "grad_norm": 0.17578125,
      "learning_rate": 3.800786369593709e-06,
      "loss": 0.8102,
      "step": 473
    },
    {
      "epoch": 0.6212319790301442,
      "grad_norm": 0.0947265625,
      "learning_rate": 3.787680209698559e-06,
      "loss": 0.8565,
      "step": 474
    },
    {
      "epoch": 0.6225425950196593,
      "grad_norm": 0.11572265625,
      "learning_rate": 3.774574049803408e-06,
      "loss": 0.8768,
      "step": 475
    },
    {
      "epoch": 0.6238532110091743,
      "grad_norm": 0.09326171875,
      "learning_rate": 3.7614678899082575e-06,
      "loss": 0.8301,
      "step": 476
    },
    {
      "epoch": 0.6251638269986893,
      "grad_norm": 0.10302734375,
      "learning_rate": 3.7483617300131064e-06,
      "loss": 0.8675,
      "step": 477
    },
    {
      "epoch": 0.6264744429882044,
      "grad_norm": 0.08642578125,
      "learning_rate": 3.7352555701179553e-06,
      "loss": 0.8506,
      "step": 478
    },
    {
      "epoch": 0.6277850589777195,
      "grad_norm": 0.0986328125,
      "learning_rate": 3.722149410222805e-06,
      "loss": 0.8497,
      "step": 479
    },
    {
      "epoch": 0.6290956749672346,
      "grad_norm": 0.08447265625,
      "learning_rate": 3.709043250327654e-06,
      "loss": 0.8406,
      "step": 480
    },
    {
      "epoch": 0.6304062909567497,
      "grad_norm": 0.0986328125,
      "learning_rate": 3.6959370904325036e-06,
      "loss": 0.8673,
      "step": 481
    },
    {
      "epoch": 0.6317169069462647,
      "grad_norm": 0.1064453125,
      "learning_rate": 3.682830930537353e-06,
      "loss": 0.8574,
      "step": 482
    },
    {
      "epoch": 0.6330275229357798,
      "grad_norm": 0.0986328125,
      "learning_rate": 3.6697247706422022e-06,
      "loss": 0.8517,
      "step": 483
    },
    {
      "epoch": 0.6343381389252949,
      "grad_norm": 0.10791015625,
      "learning_rate": 3.6566186107470515e-06,
      "loss": 0.8969,
      "step": 484
    },
    {
      "epoch": 0.63564875491481,
      "grad_norm": 0.08935546875,
      "learning_rate": 3.6435124508519004e-06,
      "loss": 0.828,
      "step": 485
    },
    {
      "epoch": 0.6369593709043251,
      "grad_norm": 0.1005859375,
      "learning_rate": 3.63040629095675e-06,
      "loss": 0.888,
      "step": 486
    },
    {
      "epoch": 0.6382699868938401,
      "grad_norm": 0.09765625,
      "learning_rate": 3.617300131061599e-06,
      "loss": 0.8397,
      "step": 487
    },
    {
      "epoch": 0.6395806028833552,
      "grad_norm": 0.08447265625,
      "learning_rate": 3.6041939711664487e-06,
      "loss": 0.816,
      "step": 488
    },
    {
      "epoch": 0.6408912188728703,
      "grad_norm": 0.0927734375,
      "learning_rate": 3.5910878112712976e-06,
      "loss": 0.8358,
      "step": 489
    },
    {
      "epoch": 0.6422018348623854,
      "grad_norm": 0.09130859375,
      "learning_rate": 3.5779816513761473e-06,
      "loss": 0.8721,
      "step": 490
    },
    {
      "epoch": 0.6435124508519003,
      "grad_norm": 0.08984375,
      "learning_rate": 3.5648754914809962e-06,
      "loss": 0.8537,
      "step": 491
    },
    {
      "epoch": 0.6448230668414154,
      "grad_norm": 0.0927734375,
      "learning_rate": 3.551769331585846e-06,
      "loss": 0.8555,
      "step": 492
    },
    {
      "epoch": 0.6461336828309305,
      "grad_norm": 0.08056640625,
      "learning_rate": 3.538663171690695e-06,
      "loss": 0.824,
      "step": 493
    },
    {
      "epoch": 0.6474442988204456,
      "grad_norm": 0.09326171875,
      "learning_rate": 3.525557011795544e-06,
      "loss": 0.861,
      "step": 494
    },
    {
      "epoch": 0.6487549148099607,
      "grad_norm": 0.1015625,
      "learning_rate": 3.5124508519003934e-06,
      "loss": 0.8718,
      "step": 495
    },
    {
      "epoch": 0.6500655307994757,
      "grad_norm": 0.08642578125,
      "learning_rate": 3.4993446920052427e-06,
      "loss": 0.8362,
      "step": 496
    },
    {
      "epoch": 0.6513761467889908,
      "grad_norm": 0.09521484375,
      "learning_rate": 3.486238532110092e-06,
      "loss": 0.8618,
      "step": 497
    },
    {
      "epoch": 0.6526867627785059,
      "grad_norm": 0.08837890625,
      "learning_rate": 3.4731323722149413e-06,
      "loss": 0.7983,
      "step": 498
    },
    {
      "epoch": 0.653997378768021,
      "grad_norm": 0.095703125,
      "learning_rate": 3.460026212319791e-06,
      "loss": 0.8572,
      "step": 499
    },
    {
      "epoch": 0.6553079947575361,
      "grad_norm": 0.08935546875,
      "learning_rate": 3.44692005242464e-06,
      "loss": 0.8217,
      "step": 500
    },
    {
      "epoch": 0.6566186107470511,
      "grad_norm": 0.09033203125,
      "learning_rate": 3.433813892529489e-06,
      "loss": 0.8224,
      "step": 501
    },
    {
      "epoch": 0.6579292267365662,
      "grad_norm": 0.09619140625,
      "learning_rate": 3.4207077326343385e-06,
      "loss": 0.8775,
      "step": 502
    },
    {
      "epoch": 0.6592398427260813,
      "grad_norm": 0.095703125,
      "learning_rate": 3.4076015727391874e-06,
      "loss": 0.8743,
      "step": 503
    },
    {
      "epoch": 0.6605504587155964,
      "grad_norm": 0.10888671875,
      "learning_rate": 3.394495412844037e-06,
      "loss": 0.8613,
      "step": 504
    },
    {
      "epoch": 0.6618610747051114,
      "grad_norm": 0.09326171875,
      "learning_rate": 3.381389252948886e-06,
      "loss": 0.8204,
      "step": 505
    },
    {
      "epoch": 0.6631716906946264,
      "grad_norm": 0.099609375,
      "learning_rate": 3.3682830930537357e-06,
      "loss": 0.9019,
      "step": 506
    },
    {
      "epoch": 0.6644823066841415,
      "grad_norm": 0.0849609375,
      "learning_rate": 3.3551769331585846e-06,
      "loss": 0.8578,
      "step": 507
    },
    {
      "epoch": 0.6657929226736566,
      "grad_norm": 0.095703125,
      "learning_rate": 3.342070773263434e-06,
      "loss": 0.8256,
      "step": 508
    },
    {
      "epoch": 0.6671035386631717,
      "grad_norm": 0.1201171875,
      "learning_rate": 3.3289646133682837e-06,
      "loss": 0.9122,
      "step": 509
    },
    {
      "epoch": 0.6684141546526867,
      "grad_norm": 0.087890625,
      "learning_rate": 3.3158584534731325e-06,
      "loss": 0.8423,
      "step": 510
    },
    {
      "epoch": 0.6697247706422018,
      "grad_norm": 0.125,
      "learning_rate": 3.3027522935779823e-06,
      "loss": 0.9521,
      "step": 511
    },
    {
      "epoch": 0.6710353866317169,
      "grad_norm": 0.08447265625,
      "learning_rate": 3.289646133682831e-06,
      "loss": 0.8626,
      "step": 512
    },
    {
      "epoch": 0.672346002621232,
      "grad_norm": 0.08837890625,
      "learning_rate": 3.276539973787681e-06,
      "loss": 0.85,
      "step": 513
    },
    {
      "epoch": 0.6736566186107471,
      "grad_norm": 0.1015625,
      "learning_rate": 3.2634338138925297e-06,
      "loss": 0.9004,
      "step": 514
    },
    {
      "epoch": 0.6749672346002621,
      "grad_norm": 0.0927734375,
      "learning_rate": 3.2503276539973786e-06,
      "loss": 0.8526,
      "step": 515
    },
    {
      "epoch": 0.6762778505897772,
      "grad_norm": 0.0986328125,
      "learning_rate": 3.2372214941022283e-06,
      "loss": 0.8744,
      "step": 516
    },
    {
      "epoch": 0.6775884665792923,
      "grad_norm": 0.0859375,
      "learning_rate": 3.2241153342070776e-06,
      "loss": 0.8476,
      "step": 517
    },
    {
      "epoch": 0.6788990825688074,
      "grad_norm": 0.103515625,
      "learning_rate": 3.211009174311927e-06,
      "loss": 0.8958,
      "step": 518
    },
    {
      "epoch": 0.6802096985583224,
      "grad_norm": 0.09326171875,
      "learning_rate": 3.1979030144167763e-06,
      "loss": 0.8467,
      "step": 519
    },
    {
      "epoch": 0.6815203145478375,
      "grad_norm": 0.09375,
      "learning_rate": 3.1847968545216256e-06,
      "loss": 0.8136,
      "step": 520
    },
    {
      "epoch": 0.6828309305373526,
      "grad_norm": 0.09814453125,
      "learning_rate": 3.171690694626475e-06,
      "loss": 0.8534,
      "step": 521
    },
    {
      "epoch": 0.6841415465268676,
      "grad_norm": 0.0947265625,
      "learning_rate": 3.1585845347313237e-06,
      "loss": 0.8944,
      "step": 522
    },
    {
      "epoch": 0.6854521625163827,
      "grad_norm": 0.0947265625,
      "learning_rate": 3.1454783748361735e-06,
      "loss": 0.8361,
      "step": 523
    },
    {
      "epoch": 0.6867627785058977,
      "grad_norm": 0.09765625,
      "learning_rate": 3.1323722149410223e-06,
      "loss": 0.879,
      "step": 524
    },
    {
      "epoch": 0.6880733944954128,
      "grad_norm": 0.09619140625,
      "learning_rate": 3.119266055045872e-06,
      "loss": 0.8758,
      "step": 525
    },
    {
      "epoch": 0.6893840104849279,
      "grad_norm": 0.11083984375,
      "learning_rate": 3.106159895150721e-06,
      "loss": 0.9117,
      "step": 526
    },
    {
      "epoch": 0.690694626474443,
      "grad_norm": 0.09130859375,
      "learning_rate": 3.0930537352555707e-06,
      "loss": 0.848,
      "step": 527
    },
    {
      "epoch": 0.6920052424639581,
      "grad_norm": 0.1025390625,
      "learning_rate": 3.0799475753604195e-06,
      "loss": 0.8588,
      "step": 528
    },
    {
      "epoch": 0.6933158584534731,
      "grad_norm": 0.09765625,
      "learning_rate": 3.066841415465269e-06,
      "loss": 0.8596,
      "step": 529
    },
    {
      "epoch": 0.6946264744429882,
      "grad_norm": 0.0849609375,
      "learning_rate": 3.053735255570118e-06,
      "loss": 0.8478,
      "step": 530
    },
    {
      "epoch": 0.6959370904325033,
      "grad_norm": 0.08544921875,
      "learning_rate": 3.0406290956749675e-06,
      "loss": 0.8543,
      "step": 531
    },
    {
      "epoch": 0.6972477064220184,
      "grad_norm": 0.0830078125,
      "learning_rate": 3.0275229357798168e-06,
      "loss": 0.8399,
      "step": 532
    },
    {
      "epoch": 0.6985583224115334,
      "grad_norm": 0.09375,
      "learning_rate": 3.014416775884666e-06,
      "loss": 0.8798,
      "step": 533
    },
    {
      "epoch": 0.6998689384010485,
      "grad_norm": 0.0966796875,
      "learning_rate": 3.0013106159895154e-06,
      "loss": 0.8754,
      "step": 534
    },
    {
      "epoch": 0.7011795543905636,
      "grad_norm": 0.09033203125,
      "learning_rate": 2.9882044560943647e-06,
      "loss": 0.825,
      "step": 535
    },
    {
      "epoch": 0.7024901703800787,
      "grad_norm": 0.08935546875,
      "learning_rate": 2.9750982961992135e-06,
      "loss": 0.822,
      "step": 536
    },
    {
      "epoch": 0.7038007863695938,
      "grad_norm": 0.091796875,
      "learning_rate": 2.9619921363040633e-06,
      "loss": 0.8401,
      "step": 537
    },
    {
      "epoch": 0.7051114023591087,
      "grad_norm": 0.08349609375,
      "learning_rate": 2.948885976408912e-06,
      "loss": 0.8338,
      "step": 538
    },
    {
      "epoch": 0.7064220183486238,
      "grad_norm": 0.0986328125,
      "learning_rate": 2.935779816513762e-06,
      "loss": 0.8804,
      "step": 539
    },
    {
      "epoch": 0.7077326343381389,
      "grad_norm": 0.08935546875,
      "learning_rate": 2.9226736566186107e-06,
      "loss": 0.8282,
      "step": 540
    },
    {
      "epoch": 0.709043250327654,
      "grad_norm": 0.09716796875,
      "learning_rate": 2.9095674967234605e-06,
      "loss": 0.8461,
      "step": 541
    },
    {
      "epoch": 0.7103538663171691,
      "grad_norm": 0.0927734375,
      "learning_rate": 2.8964613368283093e-06,
      "loss": 0.8664,
      "step": 542
    },
    {
      "epoch": 0.7116644823066841,
      "grad_norm": 0.1005859375,
      "learning_rate": 2.8833551769331587e-06,
      "loss": 0.8794,
      "step": 543
    },
    {
      "epoch": 0.7129750982961992,
      "grad_norm": 0.08984375,
      "learning_rate": 2.8702490170380084e-06,
      "loss": 0.8405,
      "step": 544
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.1044921875,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.8692,
      "step": 545
    },
    {
      "epoch": 0.7155963302752294,
      "grad_norm": 0.09619140625,
      "learning_rate": 2.844036697247707e-06,
      "loss": 0.8795,
      "step": 546
    },
    {
      "epoch": 0.7169069462647444,
      "grad_norm": 0.0947265625,
      "learning_rate": 2.830930537352556e-06,
      "loss": 0.8931,
      "step": 547
    },
    {
      "epoch": 0.7182175622542595,
      "grad_norm": 0.10986328125,
      "learning_rate": 2.8178243774574056e-06,
      "loss": 0.937,
      "step": 548
    },
    {
      "epoch": 0.7195281782437746,
      "grad_norm": 0.09130859375,
      "learning_rate": 2.8047182175622545e-06,
      "loss": 0.8622,
      "step": 549
    },
    {
      "epoch": 0.7208387942332897,
      "grad_norm": 0.09521484375,
      "learning_rate": 2.7916120576671033e-06,
      "loss": 0.8853,
      "step": 550
    },
    {
      "epoch": 0.7221494102228048,
      "grad_norm": 0.08984375,
      "learning_rate": 2.778505897771953e-06,
      "loss": 0.8372,
      "step": 551
    },
    {
      "epoch": 0.7234600262123198,
      "grad_norm": 0.09033203125,
      "learning_rate": 2.7653997378768024e-06,
      "loss": 0.8319,
      "step": 552
    },
    {
      "epoch": 0.7247706422018348,
      "grad_norm": 0.09423828125,
      "learning_rate": 2.7522935779816517e-06,
      "loss": 0.8221,
      "step": 553
    },
    {
      "epoch": 0.7260812581913499,
      "grad_norm": 0.08154296875,
      "learning_rate": 2.739187418086501e-06,
      "loss": 0.7893,
      "step": 554
    },
    {
      "epoch": 0.727391874180865,
      "grad_norm": 0.09130859375,
      "learning_rate": 2.7260812581913503e-06,
      "loss": 0.8701,
      "step": 555
    },
    {
      "epoch": 0.72870249017038,
      "grad_norm": 0.08935546875,
      "learning_rate": 2.7129750982961996e-06,
      "loss": 0.8164,
      "step": 556
    },
    {
      "epoch": 0.7300131061598951,
      "grad_norm": 0.1005859375,
      "learning_rate": 2.6998689384010485e-06,
      "loss": 0.8293,
      "step": 557
    },
    {
      "epoch": 0.7313237221494102,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.686762778505898e-06,
      "loss": 0.9142,
      "step": 558
    },
    {
      "epoch": 0.7326343381389253,
      "grad_norm": 0.09033203125,
      "learning_rate": 2.673656618610747e-06,
      "loss": 0.8517,
      "step": 559
    },
    {
      "epoch": 0.7339449541284404,
      "grad_norm": 0.09033203125,
      "learning_rate": 2.6605504587155968e-06,
      "loss": 0.8473,
      "step": 560
    },
    {
      "epoch": 0.7352555701179554,
      "grad_norm": 0.09326171875,
      "learning_rate": 2.6474442988204457e-06,
      "loss": 0.8502,
      "step": 561
    },
    {
      "epoch": 0.7365661861074705,
      "grad_norm": 0.1005859375,
      "learning_rate": 2.6343381389252954e-06,
      "loss": 0.8867,
      "step": 562
    },
    {
      "epoch": 0.7378768020969856,
      "grad_norm": 0.10693359375,
      "learning_rate": 2.6212319790301443e-06,
      "loss": 0.8873,
      "step": 563
    },
    {
      "epoch": 0.7391874180865007,
      "grad_norm": 0.11328125,
      "learning_rate": 2.6081258191349936e-06,
      "loss": 0.9348,
      "step": 564
    },
    {
      "epoch": 0.7404980340760158,
      "grad_norm": 0.08349609375,
      "learning_rate": 2.595019659239843e-06,
      "loss": 0.8116,
      "step": 565
    },
    {
      "epoch": 0.7418086500655308,
      "grad_norm": 0.1044921875,
      "learning_rate": 2.581913499344692e-06,
      "loss": 0.8726,
      "step": 566
    },
    {
      "epoch": 0.7431192660550459,
      "grad_norm": 0.10595703125,
      "learning_rate": 2.5688073394495415e-06,
      "loss": 0.9041,
      "step": 567
    },
    {
      "epoch": 0.744429882044561,
      "grad_norm": 0.0986328125,
      "learning_rate": 2.5557011795543908e-06,
      "loss": 0.8594,
      "step": 568
    },
    {
      "epoch": 0.745740498034076,
      "grad_norm": 0.09765625,
      "learning_rate": 2.54259501965924e-06,
      "loss": 0.8508,
      "step": 569
    },
    {
      "epoch": 0.747051114023591,
      "grad_norm": 0.095703125,
      "learning_rate": 2.5294888597640894e-06,
      "loss": 0.859,
      "step": 570
    },
    {
      "epoch": 0.7483617300131061,
      "grad_norm": 0.09716796875,
      "learning_rate": 2.5163826998689383e-06,
      "loss": 0.8472,
      "step": 571
    },
    {
      "epoch": 0.7496723460026212,
      "grad_norm": 0.10888671875,
      "learning_rate": 2.503276539973788e-06,
      "loss": 0.9117,
      "step": 572
    },
    {
      "epoch": 0.7509829619921363,
      "grad_norm": 0.0888671875,
      "learning_rate": 2.4901703800786373e-06,
      "loss": 0.7996,
      "step": 573
    },
    {
      "epoch": 0.7522935779816514,
      "grad_norm": 0.10986328125,
      "learning_rate": 2.4770642201834866e-06,
      "loss": 0.8995,
      "step": 574
    },
    {
      "epoch": 0.7536041939711664,
      "grad_norm": 0.0908203125,
      "learning_rate": 2.4639580602883355e-06,
      "loss": 0.8158,
      "step": 575
    },
    {
      "epoch": 0.7549148099606815,
      "grad_norm": 0.08837890625,
      "learning_rate": 2.4508519003931848e-06,
      "loss": 0.8351,
      "step": 576
    },
    {
      "epoch": 0.7562254259501966,
      "grad_norm": 0.09619140625,
      "learning_rate": 2.437745740498034e-06,
      "loss": 0.8231,
      "step": 577
    },
    {
      "epoch": 0.7575360419397117,
      "grad_norm": 0.0849609375,
      "learning_rate": 2.4246395806028834e-06,
      "loss": 0.8436,
      "step": 578
    },
    {
      "epoch": 0.7588466579292268,
      "grad_norm": 0.119140625,
      "learning_rate": 2.4115334207077327e-06,
      "loss": 0.951,
      "step": 579
    },
    {
      "epoch": 0.7601572739187418,
      "grad_norm": 0.099609375,
      "learning_rate": 2.3984272608125824e-06,
      "loss": 0.8615,
      "step": 580
    },
    {
      "epoch": 0.7614678899082569,
      "grad_norm": 0.095703125,
      "learning_rate": 2.3853211009174317e-06,
      "loss": 0.8217,
      "step": 581
    },
    {
      "epoch": 0.762778505897772,
      "grad_norm": 0.09521484375,
      "learning_rate": 2.3722149410222806e-06,
      "loss": 0.8587,
      "step": 582
    },
    {
      "epoch": 0.7640891218872871,
      "grad_norm": 0.09130859375,
      "learning_rate": 2.35910878112713e-06,
      "loss": 0.8474,
      "step": 583
    },
    {
      "epoch": 0.765399737876802,
      "grad_norm": 0.0986328125,
      "learning_rate": 2.346002621231979e-06,
      "loss": 0.922,
      "step": 584
    },
    {
      "epoch": 0.7667103538663171,
      "grad_norm": 0.10107421875,
      "learning_rate": 2.3328964613368285e-06,
      "loss": 0.8675,
      "step": 585
    },
    {
      "epoch": 0.7680209698558322,
      "grad_norm": 0.09619140625,
      "learning_rate": 2.319790301441678e-06,
      "loss": 0.8212,
      "step": 586
    },
    {
      "epoch": 0.7693315858453473,
      "grad_norm": 0.10791015625,
      "learning_rate": 2.306684141546527e-06,
      "loss": 0.8521,
      "step": 587
    },
    {
      "epoch": 0.7706422018348624,
      "grad_norm": 0.09814453125,
      "learning_rate": 2.2935779816513764e-06,
      "loss": 0.8703,
      "step": 588
    },
    {
      "epoch": 0.7719528178243774,
      "grad_norm": 0.08740234375,
      "learning_rate": 2.2804718217562257e-06,
      "loss": 0.8151,
      "step": 589
    },
    {
      "epoch": 0.7732634338138925,
      "grad_norm": 0.07666015625,
      "learning_rate": 2.267365661861075e-06,
      "loss": 0.7884,
      "step": 590
    },
    {
      "epoch": 0.7745740498034076,
      "grad_norm": 0.09033203125,
      "learning_rate": 2.2542595019659243e-06,
      "loss": 0.8229,
      "step": 591
    },
    {
      "epoch": 0.7758846657929227,
      "grad_norm": 0.09765625,
      "learning_rate": 2.2411533420707736e-06,
      "loss": 0.8589,
      "step": 592
    },
    {
      "epoch": 0.7771952817824378,
      "grad_norm": 0.09033203125,
      "learning_rate": 2.228047182175623e-06,
      "loss": 0.8547,
      "step": 593
    },
    {
      "epoch": 0.7785058977719528,
      "grad_norm": 0.109375,
      "learning_rate": 2.214941022280472e-06,
      "loss": 0.9085,
      "step": 594
    },
    {
      "epoch": 0.7798165137614679,
      "grad_norm": 0.1025390625,
      "learning_rate": 2.2018348623853215e-06,
      "loss": 0.8684,
      "step": 595
    },
    {
      "epoch": 0.781127129750983,
      "grad_norm": 0.08544921875,
      "learning_rate": 2.1887287024901704e-06,
      "loss": 0.8371,
      "step": 596
    },
    {
      "epoch": 0.7824377457404981,
      "grad_norm": 0.09912109375,
      "learning_rate": 2.1756225425950197e-06,
      "loss": 0.8456,
      "step": 597
    },
    {
      "epoch": 0.783748361730013,
      "grad_norm": 0.08740234375,
      "learning_rate": 2.162516382699869e-06,
      "loss": 0.8163,
      "step": 598
    },
    {
      "epoch": 0.7850589777195282,
      "grad_norm": 0.09521484375,
      "learning_rate": 2.1494102228047183e-06,
      "loss": 0.8369,
      "step": 599
    },
    {
      "epoch": 0.7863695937090432,
      "grad_norm": 0.0908203125,
      "learning_rate": 2.1363040629095676e-06,
      "loss": 0.8605,
      "step": 600
    },
    {
      "epoch": 0.7876802096985583,
      "grad_norm": 0.083984375,
      "learning_rate": 2.123197903014417e-06,
      "loss": 0.8092,
      "step": 601
    },
    {
      "epoch": 0.7889908256880734,
      "grad_norm": 0.09521484375,
      "learning_rate": 2.110091743119266e-06,
      "loss": 0.8318,
      "step": 602
    },
    {
      "epoch": 0.7903014416775884,
      "grad_norm": 0.1005859375,
      "learning_rate": 2.0969855832241155e-06,
      "loss": 0.8978,
      "step": 603
    },
    {
      "epoch": 0.7916120576671035,
      "grad_norm": 0.1064453125,
      "learning_rate": 2.083879423328965e-06,
      "loss": 0.8885,
      "step": 604
    },
    {
      "epoch": 0.7929226736566186,
      "grad_norm": 0.10107421875,
      "learning_rate": 2.070773263433814e-06,
      "loss": 0.868,
      "step": 605
    },
    {
      "epoch": 0.7942332896461337,
      "grad_norm": 0.0908203125,
      "learning_rate": 2.0576671035386634e-06,
      "loss": 0.872,
      "step": 606
    },
    {
      "epoch": 0.7955439056356488,
      "grad_norm": 0.0888671875,
      "learning_rate": 2.0445609436435127e-06,
      "loss": 0.8372,
      "step": 607
    },
    {
      "epoch": 0.7968545216251638,
      "grad_norm": 0.08837890625,
      "learning_rate": 2.031454783748362e-06,
      "loss": 0.8122,
      "step": 608
    },
    {
      "epoch": 0.7981651376146789,
      "grad_norm": 0.09130859375,
      "learning_rate": 2.0183486238532113e-06,
      "loss": 0.8637,
      "step": 609
    },
    {
      "epoch": 0.799475753604194,
      "grad_norm": 0.09521484375,
      "learning_rate": 2.00524246395806e-06,
      "loss": 0.8584,
      "step": 610
    },
    {
      "epoch": 0.8007863695937091,
      "grad_norm": 0.10205078125,
      "learning_rate": 1.9921363040629095e-06,
      "loss": 0.8644,
      "step": 611
    },
    {
      "epoch": 0.8020969855832241,
      "grad_norm": 0.0986328125,
      "learning_rate": 1.979030144167759e-06,
      "loss": 0.8899,
      "step": 612
    },
    {
      "epoch": 0.8034076015727392,
      "grad_norm": 0.0966796875,
      "learning_rate": 1.965923984272608e-06,
      "loss": 0.8957,
      "step": 613
    },
    {
      "epoch": 0.8047182175622543,
      "grad_norm": 0.09716796875,
      "learning_rate": 1.9528178243774574e-06,
      "loss": 0.8588,
      "step": 614
    },
    {
      "epoch": 0.8060288335517694,
      "grad_norm": 0.1083984375,
      "learning_rate": 1.939711664482307e-06,
      "loss": 0.8818,
      "step": 615
    },
    {
      "epoch": 0.8073394495412844,
      "grad_norm": 0.08642578125,
      "learning_rate": 1.9266055045871564e-06,
      "loss": 0.8423,
      "step": 616
    },
    {
      "epoch": 0.8086500655307994,
      "grad_norm": 0.10498046875,
      "learning_rate": 1.9134993446920053e-06,
      "loss": 0.9107,
      "step": 617
    },
    {
      "epoch": 0.8099606815203145,
      "grad_norm": 0.10400390625,
      "learning_rate": 1.9003931847968546e-06,
      "loss": 0.856,
      "step": 618
    },
    {
      "epoch": 0.8112712975098296,
      "grad_norm": 0.0908203125,
      "learning_rate": 1.887287024901704e-06,
      "loss": 0.8194,
      "step": 619
    },
    {
      "epoch": 0.8125819134993447,
      "grad_norm": 0.09765625,
      "learning_rate": 1.8741808650065532e-06,
      "loss": 0.8282,
      "step": 620
    },
    {
      "epoch": 0.8138925294888598,
      "grad_norm": 0.0966796875,
      "learning_rate": 1.8610747051114025e-06,
      "loss": 0.8685,
      "step": 621
    },
    {
      "epoch": 0.8152031454783748,
      "grad_norm": 0.0947265625,
      "learning_rate": 1.8479685452162518e-06,
      "loss": 0.8629,
      "step": 622
    },
    {
      "epoch": 0.8165137614678899,
      "grad_norm": 0.10107421875,
      "learning_rate": 1.8348623853211011e-06,
      "loss": 0.8451,
      "step": 623
    },
    {
      "epoch": 0.817824377457405,
      "grad_norm": 0.09375,
      "learning_rate": 1.8217562254259502e-06,
      "loss": 0.8584,
      "step": 624
    },
    {
      "epoch": 0.8191349934469201,
      "grad_norm": 0.091796875,
      "learning_rate": 1.8086500655307995e-06,
      "loss": 0.8165,
      "step": 625
    },
    {
      "epoch": 0.8204456094364351,
      "grad_norm": 0.1162109375,
      "learning_rate": 1.7955439056356488e-06,
      "loss": 0.8471,
      "step": 626
    },
    {
      "epoch": 0.8217562254259502,
      "grad_norm": 0.1279296875,
      "learning_rate": 1.7824377457404981e-06,
      "loss": 0.8677,
      "step": 627
    },
    {
      "epoch": 0.8230668414154653,
      "grad_norm": 0.0888671875,
      "learning_rate": 1.7693315858453474e-06,
      "loss": 0.8728,
      "step": 628
    },
    {
      "epoch": 0.8243774574049804,
      "grad_norm": 0.0927734375,
      "learning_rate": 1.7562254259501967e-06,
      "loss": 0.8703,
      "step": 629
    },
    {
      "epoch": 0.8256880733944955,
      "grad_norm": 0.09716796875,
      "learning_rate": 1.743119266055046e-06,
      "loss": 0.8602,
      "step": 630
    },
    {
      "epoch": 0.8269986893840104,
      "grad_norm": 0.091796875,
      "learning_rate": 1.7300131061598955e-06,
      "loss": 0.8554,
      "step": 631
    },
    {
      "epoch": 0.8283093053735255,
      "grad_norm": 0.0830078125,
      "learning_rate": 1.7169069462647444e-06,
      "loss": 0.7842,
      "step": 632
    },
    {
      "epoch": 0.8296199213630406,
      "grad_norm": 0.09619140625,
      "learning_rate": 1.7038007863695937e-06,
      "loss": 0.8738,
      "step": 633
    },
    {
      "epoch": 0.8309305373525557,
      "grad_norm": 0.099609375,
      "learning_rate": 1.690694626474443e-06,
      "loss": 0.8639,
      "step": 634
    },
    {
      "epoch": 0.8322411533420708,
      "grad_norm": 0.0966796875,
      "learning_rate": 1.6775884665792923e-06,
      "loss": 0.8823,
      "step": 635
    },
    {
      "epoch": 0.8335517693315858,
      "grad_norm": 0.10107421875,
      "learning_rate": 1.6644823066841418e-06,
      "loss": 0.8777,
      "step": 636
    },
    {
      "epoch": 0.8348623853211009,
      "grad_norm": 0.0791015625,
      "learning_rate": 1.6513761467889911e-06,
      "loss": 0.837,
      "step": 637
    },
    {
      "epoch": 0.836173001310616,
      "grad_norm": 0.0947265625,
      "learning_rate": 1.6382699868938404e-06,
      "loss": 0.8582,
      "step": 638
    },
    {
      "epoch": 0.8374836173001311,
      "grad_norm": 0.0947265625,
      "learning_rate": 1.6251638269986893e-06,
      "loss": 0.8156,
      "step": 639
    },
    {
      "epoch": 0.8387942332896461,
      "grad_norm": 0.09033203125,
      "learning_rate": 1.6120576671035388e-06,
      "loss": 0.8631,
      "step": 640
    },
    {
      "epoch": 0.8401048492791612,
      "grad_norm": 0.1171875,
      "learning_rate": 1.5989515072083881e-06,
      "loss": 0.8445,
      "step": 641
    },
    {
      "epoch": 0.8414154652686763,
      "grad_norm": 0.103515625,
      "learning_rate": 1.5858453473132374e-06,
      "loss": 0.838,
      "step": 642
    },
    {
      "epoch": 0.8427260812581914,
      "grad_norm": 0.1044921875,
      "learning_rate": 1.5727391874180867e-06,
      "loss": 0.8624,
      "step": 643
    },
    {
      "epoch": 0.8440366972477065,
      "grad_norm": 0.0986328125,
      "learning_rate": 1.559633027522936e-06,
      "loss": 0.8519,
      "step": 644
    },
    {
      "epoch": 0.8453473132372215,
      "grad_norm": 0.1279296875,
      "learning_rate": 1.5465268676277853e-06,
      "loss": 0.8694,
      "step": 645
    },
    {
      "epoch": 0.8466579292267365,
      "grad_norm": 0.08837890625,
      "learning_rate": 1.5334207077326344e-06,
      "loss": 0.8351,
      "step": 646
    },
    {
      "epoch": 0.8479685452162516,
      "grad_norm": 0.1162109375,
      "learning_rate": 1.5203145478374837e-06,
      "loss": 0.9536,
      "step": 647
    },
    {
      "epoch": 0.8492791612057667,
      "grad_norm": 0.08935546875,
      "learning_rate": 1.507208387942333e-06,
      "loss": 0.8671,
      "step": 648
    },
    {
      "epoch": 0.8505897771952818,
      "grad_norm": 0.08642578125,
      "learning_rate": 1.4941022280471823e-06,
      "loss": 0.8461,
      "step": 649
    },
    {
      "epoch": 0.8519003931847968,
      "grad_norm": 0.1044921875,
      "learning_rate": 1.4809960681520316e-06,
      "loss": 0.8974,
      "step": 650
    },
    {
      "epoch": 0.8532110091743119,
      "grad_norm": 0.09716796875,
      "learning_rate": 1.467889908256881e-06,
      "loss": 0.8587,
      "step": 651
    },
    {
      "epoch": 0.854521625163827,
      "grad_norm": 0.10009765625,
      "learning_rate": 1.4547837483617302e-06,
      "loss": 0.8808,
      "step": 652
    },
    {
      "epoch": 0.8558322411533421,
      "grad_norm": 0.0966796875,
      "learning_rate": 1.4416775884665793e-06,
      "loss": 0.857,
      "step": 653
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.09619140625,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.8254,
      "step": 654
    },
    {
      "epoch": 0.8584534731323722,
      "grad_norm": 0.09375,
      "learning_rate": 1.415465268676278e-06,
      "loss": 0.8865,
      "step": 655
    },
    {
      "epoch": 0.8597640891218873,
      "grad_norm": 0.080078125,
      "learning_rate": 1.4023591087811272e-06,
      "loss": 0.8341,
      "step": 656
    },
    {
      "epoch": 0.8610747051114024,
      "grad_norm": 0.0927734375,
      "learning_rate": 1.3892529488859765e-06,
      "loss": 0.8145,
      "step": 657
    },
    {
      "epoch": 0.8623853211009175,
      "grad_norm": 0.09765625,
      "learning_rate": 1.3761467889908258e-06,
      "loss": 0.8296,
      "step": 658
    },
    {
      "epoch": 0.8636959370904325,
      "grad_norm": 0.14453125,
      "learning_rate": 1.3630406290956751e-06,
      "loss": 0.7887,
      "step": 659
    },
    {
      "epoch": 0.8650065530799476,
      "grad_norm": 0.10693359375,
      "learning_rate": 1.3499344692005242e-06,
      "loss": 0.8713,
      "step": 660
    },
    {
      "epoch": 0.8663171690694627,
      "grad_norm": 0.09814453125,
      "learning_rate": 1.3368283093053735e-06,
      "loss": 0.8357,
      "step": 661
    },
    {
      "epoch": 0.8676277850589778,
      "grad_norm": 0.10791015625,
      "learning_rate": 1.3237221494102228e-06,
      "loss": 0.8622,
      "step": 662
    },
    {
      "epoch": 0.8689384010484927,
      "grad_norm": 0.10205078125,
      "learning_rate": 1.3106159895150721e-06,
      "loss": 0.8534,
      "step": 663
    },
    {
      "epoch": 0.8702490170380078,
      "grad_norm": 0.0830078125,
      "learning_rate": 1.2975098296199214e-06,
      "loss": 0.8119,
      "step": 664
    },
    {
      "epoch": 0.8715596330275229,
      "grad_norm": 0.0810546875,
      "learning_rate": 1.2844036697247707e-06,
      "loss": 0.7674,
      "step": 665
    },
    {
      "epoch": 0.872870249017038,
      "grad_norm": 0.107421875,
      "learning_rate": 1.27129750982962e-06,
      "loss": 0.8659,
      "step": 666
    },
    {
      "epoch": 0.8741808650065531,
      "grad_norm": 0.0908203125,
      "learning_rate": 1.2581913499344691e-06,
      "loss": 0.8658,
      "step": 667
    },
    {
      "epoch": 0.8754914809960681,
      "grad_norm": 0.09033203125,
      "learning_rate": 1.2450851900393186e-06,
      "loss": 0.839,
      "step": 668
    },
    {
      "epoch": 0.8768020969855832,
      "grad_norm": 0.08642578125,
      "learning_rate": 1.2319790301441677e-06,
      "loss": 0.8186,
      "step": 669
    },
    {
      "epoch": 0.8781127129750983,
      "grad_norm": 0.09228515625,
      "learning_rate": 1.218872870249017e-06,
      "loss": 0.8901,
      "step": 670
    },
    {
      "epoch": 0.8794233289646134,
      "grad_norm": 0.09033203125,
      "learning_rate": 1.2057667103538663e-06,
      "loss": 0.8684,
      "step": 671
    },
    {
      "epoch": 0.8807339449541285,
      "grad_norm": 0.0986328125,
      "learning_rate": 1.1926605504587159e-06,
      "loss": 0.8821,
      "step": 672
    },
    {
      "epoch": 0.8820445609436435,
      "grad_norm": 0.08203125,
      "learning_rate": 1.179554390563565e-06,
      "loss": 0.8073,
      "step": 673
    },
    {
      "epoch": 0.8833551769331586,
      "grad_norm": 0.09765625,
      "learning_rate": 1.1664482306684142e-06,
      "loss": 0.8496,
      "step": 674
    },
    {
      "epoch": 0.8846657929226737,
      "grad_norm": 0.07958984375,
      "learning_rate": 1.1533420707732635e-06,
      "loss": 0.7955,
      "step": 675
    },
    {
      "epoch": 0.8859764089121888,
      "grad_norm": 0.0966796875,
      "learning_rate": 1.1402359108781128e-06,
      "loss": 0.8757,
      "step": 676
    },
    {
      "epoch": 0.8872870249017037,
      "grad_norm": 0.09619140625,
      "learning_rate": 1.1271297509829621e-06,
      "loss": 0.8424,
      "step": 677
    },
    {
      "epoch": 0.8885976408912188,
      "grad_norm": 0.08837890625,
      "learning_rate": 1.1140235910878115e-06,
      "loss": 0.8599,
      "step": 678
    },
    {
      "epoch": 0.8899082568807339,
      "grad_norm": 0.10302734375,
      "learning_rate": 1.1009174311926608e-06,
      "loss": 0.7905,
      "step": 679
    },
    {
      "epoch": 0.891218872870249,
      "grad_norm": 0.09912109375,
      "learning_rate": 1.0878112712975098e-06,
      "loss": 0.811,
      "step": 680
    },
    {
      "epoch": 0.8925294888597641,
      "grad_norm": 0.09375,
      "learning_rate": 1.0747051114023591e-06,
      "loss": 0.8825,
      "step": 681
    },
    {
      "epoch": 0.8938401048492791,
      "grad_norm": 0.0966796875,
      "learning_rate": 1.0615989515072084e-06,
      "loss": 0.8369,
      "step": 682
    },
    {
      "epoch": 0.8951507208387942,
      "grad_norm": 0.09326171875,
      "learning_rate": 1.0484927916120577e-06,
      "loss": 0.8533,
      "step": 683
    },
    {
      "epoch": 0.8964613368283093,
      "grad_norm": 0.095703125,
      "learning_rate": 1.035386631716907e-06,
      "loss": 0.8499,
      "step": 684
    },
    {
      "epoch": 0.8977719528178244,
      "grad_norm": 0.08984375,
      "learning_rate": 1.0222804718217564e-06,
      "loss": 0.831,
      "step": 685
    },
    {
      "epoch": 0.8990825688073395,
      "grad_norm": 0.10009765625,
      "learning_rate": 1.0091743119266057e-06,
      "loss": 0.8551,
      "step": 686
    },
    {
      "epoch": 0.9003931847968545,
      "grad_norm": 0.1044921875,
      "learning_rate": 9.960681520314547e-07,
      "loss": 0.8516,
      "step": 687
    },
    {
      "epoch": 0.9017038007863696,
      "grad_norm": 0.0830078125,
      "learning_rate": 9.82961992136304e-07,
      "loss": 0.7739,
      "step": 688
    },
    {
      "epoch": 0.9030144167758847,
      "grad_norm": 0.119140625,
      "learning_rate": 9.698558322411536e-07,
      "loss": 0.9754,
      "step": 689
    },
    {
      "epoch": 0.9043250327653998,
      "grad_norm": 0.10107421875,
      "learning_rate": 9.567496723460027e-07,
      "loss": 0.8462,
      "step": 690
    },
    {
      "epoch": 0.9056356487549148,
      "grad_norm": 0.09228515625,
      "learning_rate": 9.43643512450852e-07,
      "loss": 0.8372,
      "step": 691
    },
    {
      "epoch": 0.9069462647444299,
      "grad_norm": 0.09521484375,
      "learning_rate": 9.305373525557013e-07,
      "loss": 0.8669,
      "step": 692
    },
    {
      "epoch": 0.908256880733945,
      "grad_norm": 0.08984375,
      "learning_rate": 9.174311926605506e-07,
      "loss": 0.8267,
      "step": 693
    },
    {
      "epoch": 0.90956749672346,
      "grad_norm": 0.08984375,
      "learning_rate": 9.043250327653998e-07,
      "loss": 0.7513,
      "step": 694
    },
    {
      "epoch": 0.9108781127129751,
      "grad_norm": 0.0849609375,
      "learning_rate": 8.912188728702491e-07,
      "loss": 0.8241,
      "step": 695
    },
    {
      "epoch": 0.9121887287024901,
      "grad_norm": 0.08544921875,
      "learning_rate": 8.781127129750984e-07,
      "loss": 0.871,
      "step": 696
    },
    {
      "epoch": 0.9134993446920052,
      "grad_norm": 0.0849609375,
      "learning_rate": 8.650065530799478e-07,
      "loss": 0.8117,
      "step": 697
    },
    {
      "epoch": 0.9148099606815203,
      "grad_norm": 0.10400390625,
      "learning_rate": 8.519003931847969e-07,
      "loss": 0.9001,
      "step": 698
    },
    {
      "epoch": 0.9161205766710354,
      "grad_norm": 0.0751953125,
      "learning_rate": 8.387942332896462e-07,
      "loss": 0.7326,
      "step": 699
    },
    {
      "epoch": 0.9174311926605505,
      "grad_norm": 0.1005859375,
      "learning_rate": 8.256880733944956e-07,
      "loss": 0.8926,
      "step": 700
    },
    {
      "epoch": 0.9187418086500655,
      "grad_norm": 0.08837890625,
      "learning_rate": 8.125819134993447e-07,
      "loss": 0.8292,
      "step": 701
    },
    {
      "epoch": 0.9200524246395806,
      "grad_norm": 0.099609375,
      "learning_rate": 7.994757536041941e-07,
      "loss": 0.8898,
      "step": 702
    },
    {
      "epoch": 0.9213630406290957,
      "grad_norm": 0.10693359375,
      "learning_rate": 7.863695937090434e-07,
      "loss": 0.908,
      "step": 703
    },
    {
      "epoch": 0.9226736566186108,
      "grad_norm": 0.08837890625,
      "learning_rate": 7.732634338138927e-07,
      "loss": 0.878,
      "step": 704
    },
    {
      "epoch": 0.9239842726081258,
      "grad_norm": 0.09326171875,
      "learning_rate": 7.601572739187419e-07,
      "loss": 0.8028,
      "step": 705
    },
    {
      "epoch": 0.9252948885976409,
      "grad_norm": 0.0859375,
      "learning_rate": 7.470511140235912e-07,
      "loss": 0.8004,
      "step": 706
    },
    {
      "epoch": 0.926605504587156,
      "grad_norm": 0.1044921875,
      "learning_rate": 7.339449541284405e-07,
      "loss": 0.8892,
      "step": 707
    },
    {
      "epoch": 0.927916120576671,
      "grad_norm": 0.0966796875,
      "learning_rate": 7.208387942332897e-07,
      "loss": 0.8666,
      "step": 708
    },
    {
      "epoch": 0.9292267365661862,
      "grad_norm": 0.10205078125,
      "learning_rate": 7.07732634338139e-07,
      "loss": 0.8958,
      "step": 709
    },
    {
      "epoch": 0.9305373525557011,
      "grad_norm": 0.11279296875,
      "learning_rate": 6.946264744429883e-07,
      "loss": 0.9089,
      "step": 710
    },
    {
      "epoch": 0.9318479685452162,
      "grad_norm": 0.0966796875,
      "learning_rate": 6.815203145478376e-07,
      "loss": 0.8549,
      "step": 711
    },
    {
      "epoch": 0.9331585845347313,
      "grad_norm": 0.1015625,
      "learning_rate": 6.684141546526868e-07,
      "loss": 0.8955,
      "step": 712
    },
    {
      "epoch": 0.9344692005242464,
      "grad_norm": 0.0859375,
      "learning_rate": 6.553079947575361e-07,
      "loss": 0.8221,
      "step": 713
    },
    {
      "epoch": 0.9357798165137615,
      "grad_norm": 0.09130859375,
      "learning_rate": 6.422018348623854e-07,
      "loss": 0.8364,
      "step": 714
    },
    {
      "epoch": 0.9370904325032765,
      "grad_norm": 0.0830078125,
      "learning_rate": 6.290956749672346e-07,
      "loss": 0.8322,
      "step": 715
    },
    {
      "epoch": 0.9384010484927916,
      "grad_norm": 0.11572265625,
      "learning_rate": 6.159895150720839e-07,
      "loss": 0.8519,
      "step": 716
    },
    {
      "epoch": 0.9397116644823067,
      "grad_norm": 0.09375,
      "learning_rate": 6.028833551769332e-07,
      "loss": 0.8799,
      "step": 717
    },
    {
      "epoch": 0.9410222804718218,
      "grad_norm": 0.0849609375,
      "learning_rate": 5.897771952817825e-07,
      "loss": 0.8028,
      "step": 718
    },
    {
      "epoch": 0.9423328964613368,
      "grad_norm": 0.14453125,
      "learning_rate": 5.766710353866318e-07,
      "loss": 1.0117,
      "step": 719
    },
    {
      "epoch": 0.9436435124508519,
      "grad_norm": 0.0966796875,
      "learning_rate": 5.635648754914811e-07,
      "loss": 0.8829,
      "step": 720
    },
    {
      "epoch": 0.944954128440367,
      "grad_norm": 0.1005859375,
      "learning_rate": 5.504587155963304e-07,
      "loss": 0.8786,
      "step": 721
    },
    {
      "epoch": 0.9462647444298821,
      "grad_norm": 0.0849609375,
      "learning_rate": 5.373525557011796e-07,
      "loss": 0.7952,
      "step": 722
    },
    {
      "epoch": 0.9475753604193972,
      "grad_norm": 0.181640625,
      "learning_rate": 5.242463958060289e-07,
      "loss": 0.8757,
      "step": 723
    },
    {
      "epoch": 0.9488859764089121,
      "grad_norm": 0.0927734375,
      "learning_rate": 5.111402359108782e-07,
      "loss": 0.834,
      "step": 724
    },
    {
      "epoch": 0.9501965923984272,
      "grad_norm": 0.07861328125,
      "learning_rate": 4.980340760157274e-07,
      "loss": 0.7742,
      "step": 725
    },
    {
      "epoch": 0.9515072083879423,
      "grad_norm": 0.08984375,
      "learning_rate": 4.849279161205768e-07,
      "loss": 0.7577,
      "step": 726
    },
    {
      "epoch": 0.9528178243774574,
      "grad_norm": 0.091796875,
      "learning_rate": 4.71821756225426e-07,
      "loss": 0.8595,
      "step": 727
    },
    {
      "epoch": 0.9541284403669725,
      "grad_norm": 0.09521484375,
      "learning_rate": 4.587155963302753e-07,
      "loss": 0.8525,
      "step": 728
    },
    {
      "epoch": 0.9554390563564875,
      "grad_norm": 0.0869140625,
      "learning_rate": 4.4560943643512453e-07,
      "loss": 0.8467,
      "step": 729
    },
    {
      "epoch": 0.9567496723460026,
      "grad_norm": 0.09814453125,
      "learning_rate": 4.325032765399739e-07,
      "loss": 0.8545,
      "step": 730
    },
    {
      "epoch": 0.9580602883355177,
      "grad_norm": 0.09130859375,
      "learning_rate": 4.193971166448231e-07,
      "loss": 0.8175,
      "step": 731
    },
    {
      "epoch": 0.9593709043250328,
      "grad_norm": 0.0869140625,
      "learning_rate": 4.0629095674967233e-07,
      "loss": 0.8406,
      "step": 732
    },
    {
      "epoch": 0.9606815203145478,
      "grad_norm": 0.1044921875,
      "learning_rate": 3.931847968545217e-07,
      "loss": 0.8783,
      "step": 733
    },
    {
      "epoch": 0.9619921363040629,
      "grad_norm": 0.0947265625,
      "learning_rate": 3.8007863695937093e-07,
      "loss": 0.8492,
      "step": 734
    },
    {
      "epoch": 0.963302752293578,
      "grad_norm": 0.087890625,
      "learning_rate": 3.6697247706422023e-07,
      "loss": 0.7609,
      "step": 735
    },
    {
      "epoch": 0.9646133682830931,
      "grad_norm": 0.09375,
      "learning_rate": 3.538663171690695e-07,
      "loss": 0.8502,
      "step": 736
    },
    {
      "epoch": 0.9659239842726082,
      "grad_norm": 0.09130859375,
      "learning_rate": 3.407601572739188e-07,
      "loss": 0.8456,
      "step": 737
    },
    {
      "epoch": 0.9672346002621232,
      "grad_norm": 0.099609375,
      "learning_rate": 3.2765399737876803e-07,
      "loss": 0.873,
      "step": 738
    },
    {
      "epoch": 0.9685452162516383,
      "grad_norm": 0.0966796875,
      "learning_rate": 3.145478374836173e-07,
      "loss": 0.8559,
      "step": 739
    },
    {
      "epoch": 0.9698558322411533,
      "grad_norm": 0.09033203125,
      "learning_rate": 3.014416775884666e-07,
      "loss": 0.8279,
      "step": 740
    },
    {
      "epoch": 0.9711664482306684,
      "grad_norm": 0.09912109375,
      "learning_rate": 2.883355176933159e-07,
      "loss": 0.8649,
      "step": 741
    },
    {
      "epoch": 0.9724770642201835,
      "grad_norm": 0.0927734375,
      "learning_rate": 2.752293577981652e-07,
      "loss": 0.8299,
      "step": 742
    },
    {
      "epoch": 0.9737876802096985,
      "grad_norm": 0.0986328125,
      "learning_rate": 2.6212319790301444e-07,
      "loss": 0.8482,
      "step": 743
    },
    {
      "epoch": 0.9750982961992136,
      "grad_norm": 0.09326171875,
      "learning_rate": 2.490170380078637e-07,
      "loss": 0.8758,
      "step": 744
    },
    {
      "epoch": 0.9764089121887287,
      "grad_norm": 0.0986328125,
      "learning_rate": 2.35910878112713e-07,
      "loss": 0.8242,
      "step": 745
    },
    {
      "epoch": 0.9777195281782438,
      "grad_norm": 0.083984375,
      "learning_rate": 2.2280471821756226e-07,
      "loss": 0.8508,
      "step": 746
    },
    {
      "epoch": 0.9790301441677588,
      "grad_norm": 0.09521484375,
      "learning_rate": 2.0969855832241154e-07,
      "loss": 0.8623,
      "step": 747
    },
    {
      "epoch": 0.9803407601572739,
      "grad_norm": 0.1142578125,
      "learning_rate": 1.9659239842726084e-07,
      "loss": 0.8708,
      "step": 748
    },
    {
      "epoch": 0.981651376146789,
      "grad_norm": 0.1005859375,
      "learning_rate": 1.8348623853211012e-07,
      "loss": 0.847,
      "step": 749
    },
    {
      "epoch": 0.9829619921363041,
      "grad_norm": 0.08935546875,
      "learning_rate": 1.703800786369594e-07,
      "loss": 0.8746,
      "step": 750
    },
    {
      "epoch": 0.9842726081258192,
      "grad_norm": 0.09033203125,
      "learning_rate": 1.5727391874180864e-07,
      "loss": 0.8517,
      "step": 751
    },
    {
      "epoch": 0.9855832241153342,
      "grad_norm": 0.08935546875,
      "learning_rate": 1.4416775884665794e-07,
      "loss": 0.8401,
      "step": 752
    },
    {
      "epoch": 0.9868938401048493,
      "grad_norm": 0.083984375,
      "learning_rate": 1.3106159895150722e-07,
      "loss": 0.8376,
      "step": 753
    },
    {
      "epoch": 0.9882044560943644,
      "grad_norm": 0.08837890625,
      "learning_rate": 1.179554390563565e-07,
      "loss": 0.8049,
      "step": 754
    },
    {
      "epoch": 0.9895150720838795,
      "grad_norm": 0.11865234375,
      "learning_rate": 1.0484927916120577e-07,
      "loss": 0.9423,
      "step": 755
    },
    {
      "epoch": 0.9908256880733946,
      "grad_norm": 0.0859375,
      "learning_rate": 9.174311926605506e-08,
      "loss": 0.8444,
      "step": 756
    },
    {
      "epoch": 0.9921363040629095,
      "grad_norm": 0.0810546875,
      "learning_rate": 7.863695937090432e-08,
      "loss": 0.8502,
      "step": 757
    },
    {
      "epoch": 0.9934469200524246,
      "grad_norm": 0.10693359375,
      "learning_rate": 6.553079947575361e-08,
      "loss": 0.854,
      "step": 758
    },
    {
      "epoch": 0.9947575360419397,
      "grad_norm": 0.0947265625,
      "learning_rate": 5.2424639580602885e-08,
      "loss": 0.8693,
      "step": 759
    },
    {
      "epoch": 0.9960681520314548,
      "grad_norm": 0.1044921875,
      "learning_rate": 3.931847968545216e-08,
      "loss": 0.8721,
      "step": 760
    },
    {
      "epoch": 0.9973787680209698,
      "grad_norm": 0.087890625,
      "learning_rate": 2.6212319790301442e-08,
      "loss": 0.8678,
      "step": 761
    },
    {
      "epoch": 0.9986893840104849,
      "grad_norm": 0.0927734375,
      "learning_rate": 1.3106159895150721e-08,
      "loss": 0.8628,
      "step": 762
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.0791015625,
      "learning_rate": 0.0,
      "loss": 0.7969,
      "step": 763
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 763,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.468456659819692e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}