{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.3269469692015955,
  "eval_steps": 500,
  "global_step": 2500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000653893938403191,
      "grad_norm": 7.78125,
      "learning_rate": 1e-05,
      "loss": 10.9314,
      "step": 5
    },
    {
      "epoch": 0.001307787876806382,
      "grad_norm": 7.15625,
      "learning_rate": 2e-05,
      "loss": 10.8618,
      "step": 10
    },
    {
      "epoch": 0.001961681815209573,
      "grad_norm": 8.625,
      "learning_rate": 3e-05,
      "loss": 10.6602,
      "step": 15
    },
    {
      "epoch": 0.002615575753612764,
      "grad_norm": 3.78125,
      "learning_rate": 4e-05,
      "loss": 10.4672,
      "step": 20
    },
    {
      "epoch": 0.003269469692015955,
      "grad_norm": 3.21875,
      "learning_rate": 5e-05,
      "loss": 10.3043,
      "step": 25
    },
    {
      "epoch": 0.003923363630419146,
      "grad_norm": 2.921875,
      "learning_rate": 6e-05,
      "loss": 10.2382,
      "step": 30
    },
    {
      "epoch": 0.004577257568822337,
      "grad_norm": 2.84375,
      "learning_rate": 7.000000000000001e-05,
      "loss": 10.072,
      "step": 35
    },
    {
      "epoch": 0.005231151507225528,
      "grad_norm": 2.84375,
      "learning_rate": 8e-05,
      "loss": 9.9208,
      "step": 40
    },
    {
      "epoch": 0.005885045445628719,
      "grad_norm": 2.5,
      "learning_rate": 8.999999999999999e-05,
      "loss": 9.7828,
      "step": 45
    },
    {
      "epoch": 0.00653893938403191,
      "grad_norm": 2.53125,
      "learning_rate": 0.0001,
      "loss": 9.5983,
      "step": 50
    },
    {
      "epoch": 0.007192833322435101,
      "grad_norm": 2.515625,
      "learning_rate": 0.00011,
      "loss": 9.3536,
      "step": 55
    },
    {
      "epoch": 0.007846727260838291,
      "grad_norm": 2.265625,
      "learning_rate": 0.00012,
      "loss": 9.1994,
      "step": 60
    },
    {
      "epoch": 0.008500621199241483,
      "grad_norm": 1.96875,
      "learning_rate": 0.00013000000000000002,
      "loss": 9.0285,
      "step": 65
    },
    {
      "epoch": 0.009154515137644674,
      "grad_norm": 1.734375,
      "learning_rate": 0.00014000000000000001,
      "loss": 8.8309,
      "step": 70
    },
    {
      "epoch": 0.009808409076047865,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00015,
      "loss": 8.7217,
      "step": 75
    },
    {
      "epoch": 0.010462303014451056,
      "grad_norm": 1.296875,
      "learning_rate": 0.00016,
      "loss": 8.5904,
      "step": 80
    },
    {
      "epoch": 0.011116196952854247,
      "grad_norm": 1.140625,
      "learning_rate": 0.00017,
      "loss": 8.5347,
      "step": 85
    },
    {
      "epoch": 0.011770090891257438,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00017999999999999998,
      "loss": 8.5,
      "step": 90
    },
    {
      "epoch": 0.01242398482966063,
      "grad_norm": 1.265625,
      "learning_rate": 0.00019,
      "loss": 8.4191,
      "step": 95
    },
    {
      "epoch": 0.01307787876806382,
      "grad_norm": 1.296875,
      "learning_rate": 0.0002,
      "loss": 8.4152,
      "step": 100
    },
    {
      "epoch": 0.013731772706467011,
      "grad_norm": 1.25,
      "learning_rate": 0.00021,
      "loss": 8.4117,
      "step": 105
    },
    {
      "epoch": 0.014385666644870202,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00022,
      "loss": 8.3862,
      "step": 110
    },
    {
      "epoch": 0.015039560583273394,
      "grad_norm": 1.6484375,
      "learning_rate": 0.00023,
      "loss": 8.3845,
      "step": 115
    },
    {
      "epoch": 0.015693454521676583,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00024,
      "loss": 8.3856,
      "step": 120
    },
    {
      "epoch": 0.016347348460079774,
      "grad_norm": 2.140625,
      "learning_rate": 0.00025,
      "loss": 8.3267,
      "step": 125
    },
    {
      "epoch": 0.017001242398482965,
      "grad_norm": 1.5625,
      "learning_rate": 0.00026000000000000003,
      "loss": 8.3117,
      "step": 130
    },
    {
      "epoch": 0.017655136336886156,
      "grad_norm": 1.46875,
      "learning_rate": 0.00027,
      "loss": 8.2819,
      "step": 135
    },
    {
      "epoch": 0.018309030275289347,
      "grad_norm": 1.890625,
      "learning_rate": 0.00028000000000000003,
      "loss": 8.3206,
      "step": 140
    },
    {
      "epoch": 0.01896292421369254,
      "grad_norm": 1.390625,
      "learning_rate": 0.00029,
      "loss": 8.2799,
      "step": 145
    },
    {
      "epoch": 0.01961681815209573,
      "grad_norm": 1.609375,
      "learning_rate": 0.0003,
      "loss": 8.2797,
      "step": 150
    },
    {
      "epoch": 0.02027071209049892,
      "grad_norm": 1.65625,
      "learning_rate": 0.00031,
      "loss": 8.2185,
      "step": 155
    },
    {
      "epoch": 0.02092460602890211,
      "grad_norm": 1.734375,
      "learning_rate": 0.00032,
      "loss": 8.1803,
      "step": 160
    },
    {
      "epoch": 0.021578499967305303,
      "grad_norm": 2.109375,
      "learning_rate": 0.00033,
      "loss": 8.1965,
      "step": 165
    },
    {
      "epoch": 0.022232393905708494,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00034,
      "loss": 8.1966,
      "step": 170
    },
    {
      "epoch": 0.022886287844111685,
      "grad_norm": 2.3125,
      "learning_rate": 0.00035,
      "loss": 8.1858,
      "step": 175
    },
    {
      "epoch": 0.023540181782514876,
      "grad_norm": 1.8359375,
      "learning_rate": 0.00035999999999999997,
      "loss": 8.1155,
      "step": 180
    },
    {
      "epoch": 0.024194075720918067,
      "grad_norm": 1.703125,
      "learning_rate": 0.00037,
      "loss": 8.0795,
      "step": 185
    },
    {
      "epoch": 0.02484796965932126,
      "grad_norm": 1.6953125,
      "learning_rate": 0.00038,
      "loss": 8.0587,
      "step": 190
    },
    {
      "epoch": 0.02550186359772445,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00039000000000000005,
      "loss": 8.0238,
      "step": 195
    },
    {
      "epoch": 0.02615575753612764,
      "grad_norm": 1.421875,
      "learning_rate": 0.0004,
      "loss": 8.0203,
      "step": 200
    },
    {
      "epoch": 0.02680965147453083,
      "grad_norm": 1.65625,
      "learning_rate": 0.00041,
      "loss": 8.0557,
      "step": 205
    },
    {
      "epoch": 0.027463545412934023,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00042,
      "loss": 8.0153,
      "step": 210
    },
    {
      "epoch": 0.028117439351337214,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00043,
      "loss": 8.0324,
      "step": 215
    },
    {
      "epoch": 0.028771333289740405,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00044,
      "loss": 7.943,
      "step": 220
    },
    {
      "epoch": 0.029425227228143596,
      "grad_norm": 1.765625,
      "learning_rate": 0.00045000000000000004,
      "loss": 7.9174,
      "step": 225
    },
    {
      "epoch": 0.030079121166546787,
      "grad_norm": 1.484375,
      "learning_rate": 0.00046,
      "loss": 7.9345,
      "step": 230
    },
    {
      "epoch": 0.030733015104949978,
      "grad_norm": 1.625,
      "learning_rate": 0.00047,
      "loss": 7.8904,
      "step": 235
    },
    {
      "epoch": 0.031386909043353166,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00048,
      "loss": 7.9578,
      "step": 240
    },
    {
      "epoch": 0.03204080298175636,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00049,
      "loss": 7.8584,
      "step": 245
    },
    {
      "epoch": 0.03269469692015955,
      "grad_norm": 1.953125,
      "learning_rate": 0.0005,
      "loss": 7.8777,
      "step": 250
    },
    {
      "epoch": 0.03334859085856274,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00051,
      "loss": 7.8879,
      "step": 255
    },
    {
      "epoch": 0.03400248479696593,
      "grad_norm": 1.5859375,
      "learning_rate": 0.0005200000000000001,
      "loss": 7.8749,
      "step": 260
    },
    {
      "epoch": 0.03465637873536912,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0005300000000000001,
      "loss": 7.8588,
      "step": 265
    },
    {
      "epoch": 0.03531027267377231,
      "grad_norm": 1.8125,
      "learning_rate": 0.00054,
      "loss": 7.8164,
      "step": 270
    },
    {
      "epoch": 0.035964166612175504,
      "grad_norm": 1.890625,
      "learning_rate": 0.00055,
      "loss": 7.8153,
      "step": 275
    },
    {
      "epoch": 0.036618060550578695,
      "grad_norm": 1.828125,
      "learning_rate": 0.0005600000000000001,
      "loss": 7.8396,
      "step": 280
    },
    {
      "epoch": 0.037271954488981886,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00057,
      "loss": 7.8175,
      "step": 285
    },
    {
      "epoch": 0.03792584842738508,
      "grad_norm": 1.609375,
      "learning_rate": 0.00058,
      "loss": 7.8173,
      "step": 290
    },
    {
      "epoch": 0.03857974236578827,
      "grad_norm": 1.765625,
      "learning_rate": 0.00059,
      "loss": 7.7171,
      "step": 295
    },
    {
      "epoch": 0.03923363630419146,
      "grad_norm": 1.46875,
      "learning_rate": 0.0006,
      "loss": 7.7434,
      "step": 300
    },
    {
      "epoch": 0.03988753024259465,
      "grad_norm": 1.75,
      "learning_rate": 0.00061,
      "loss": 7.7722,
      "step": 305
    },
    {
      "epoch": 0.04054142418099784,
      "grad_norm": 1.5859375,
      "learning_rate": 0.00062,
      "loss": 7.7211,
      "step": 310
    },
    {
      "epoch": 0.04119531811940103,
      "grad_norm": 1.390625,
      "learning_rate": 0.00063,
      "loss": 7.7748,
      "step": 315
    },
    {
      "epoch": 0.04184921205780422,
      "grad_norm": 1.59375,
      "learning_rate": 0.00064,
      "loss": 7.6554,
      "step": 320
    },
    {
      "epoch": 0.042503105996207415,
      "grad_norm": 1.5,
      "learning_rate": 0.0006500000000000001,
      "loss": 7.6331,
      "step": 325
    },
    {
      "epoch": 0.043156999934610606,
      "grad_norm": 1.484375,
      "learning_rate": 0.00066,
      "loss": 7.7316,
      "step": 330
    },
    {
      "epoch": 0.0438108938730138,
      "grad_norm": 1.65625,
      "learning_rate": 0.00067,
      "loss": 7.6819,
      "step": 335
    },
    {
      "epoch": 0.04446478781141699,
      "grad_norm": 1.65625,
      "learning_rate": 0.00068,
      "loss": 7.6488,
      "step": 340
    },
    {
      "epoch": 0.04511868174982018,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00069,
      "loss": 7.5754,
      "step": 345
    },
    {
      "epoch": 0.04577257568822337,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0007,
      "loss": 7.6011,
      "step": 350
    },
    {
      "epoch": 0.04642646962662656,
      "grad_norm": 1.703125,
      "learning_rate": 0.00071,
      "loss": 7.5742,
      "step": 355
    },
    {
      "epoch": 0.04708036356502975,
      "grad_norm": 1.5,
      "learning_rate": 0.0007199999999999999,
      "loss": 7.5733,
      "step": 360
    },
    {
      "epoch": 0.04773425750343294,
      "grad_norm": 1.390625,
      "learning_rate": 0.00073,
      "loss": 7.5406,
      "step": 365
    },
    {
      "epoch": 0.048388151441836134,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00074,
      "loss": 7.5902,
      "step": 370
    },
    {
      "epoch": 0.049042045380239326,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00075,
      "loss": 7.5615,
      "step": 375
    },
    {
      "epoch": 0.04969593931864252,
      "grad_norm": 1.5625,
      "learning_rate": 0.00076,
      "loss": 7.5224,
      "step": 380
    },
    {
      "epoch": 0.05034983325704571,
      "grad_norm": 1.546875,
      "learning_rate": 0.0007700000000000001,
      "loss": 7.5185,
      "step": 385
    },
    {
      "epoch": 0.0510037271954489,
      "grad_norm": 1.65625,
      "learning_rate": 0.0007800000000000001,
      "loss": 7.4919,
      "step": 390
    },
    {
      "epoch": 0.05165762113385209,
      "grad_norm": 1.59375,
      "learning_rate": 0.00079,
      "loss": 7.4876,
      "step": 395
    },
    {
      "epoch": 0.05231151507225528,
      "grad_norm": 1.578125,
      "learning_rate": 0.0008,
      "loss": 7.4722,
      "step": 400
    },
    {
      "epoch": 0.05296540901065847,
      "grad_norm": 1.6640625,
      "learning_rate": 0.0008100000000000001,
      "loss": 7.498,
      "step": 405
    },
    {
      "epoch": 0.05361930294906166,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00082,
      "loss": 7.4047,
      "step": 410
    },
    {
      "epoch": 0.054273196887464854,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00083,
      "loss": 7.4376,
      "step": 415
    },
    {
      "epoch": 0.054927090825868045,
      "grad_norm": 1.734375,
      "learning_rate": 0.00084,
      "loss": 7.4518,
      "step": 420
    },
    {
      "epoch": 0.055580984764271237,
      "grad_norm": 1.453125,
      "learning_rate": 0.00085,
      "loss": 7.4957,
      "step": 425
    },
    {
      "epoch": 0.05623487870267443,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00086,
      "loss": 7.4782,
      "step": 430
    },
    {
      "epoch": 0.05688877264107762,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00087,
      "loss": 7.4289,
      "step": 435
    },
    {
      "epoch": 0.05754266657948081,
      "grad_norm": 3.09375,
      "learning_rate": 0.00088,
      "loss": 7.4412,
      "step": 440
    },
    {
      "epoch": 0.058196560517884,
      "grad_norm": 1.625,
      "learning_rate": 0.0008900000000000001,
      "loss": 7.413,
      "step": 445
    },
    {
      "epoch": 0.05885045445628719,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009000000000000001,
      "loss": 7.4372,
      "step": 450
    },
    {
      "epoch": 0.05950434839469038,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00091,
      "loss": 7.3155,
      "step": 455
    },
    {
      "epoch": 0.060158242333093574,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00092,
      "loss": 7.4253,
      "step": 460
    },
    {
      "epoch": 0.060812136271496765,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00093,
      "loss": 7.3755,
      "step": 465
    },
    {
      "epoch": 0.061466030209899956,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00094,
      "loss": 7.391,
      "step": 470
    },
    {
      "epoch": 0.06211992414830315,
      "grad_norm": 1.5703125,
      "learning_rate": 0.00095,
      "loss": 7.3909,
      "step": 475
    },
    {
      "epoch": 0.06277381808670633,
      "grad_norm": 1.5,
      "learning_rate": 0.00096,
      "loss": 7.3484,
      "step": 480
    },
    {
      "epoch": 0.06342771202510952,
      "grad_norm": 1.609375,
      "learning_rate": 0.0009699999999999999,
      "loss": 7.3424,
      "step": 485
    },
    {
      "epoch": 0.06408160596351271,
      "grad_norm": 1.6953125,
      "learning_rate": 0.00098,
      "loss": 7.3676,
      "step": 490
    },
    {
      "epoch": 0.0647354999019159,
      "grad_norm": 1.53125,
      "learning_rate": 0.00099,
      "loss": 7.2811,
      "step": 495
    },
    {
      "epoch": 0.0653893938403191,
      "grad_norm": 1.53125,
      "learning_rate": 0.001,
      "loss": 7.2938,
      "step": 500
    },
    {
      "epoch": 0.06604328777872229,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009999954680464025,
      "loss": 7.2968,
      "step": 505
    },
    {
      "epoch": 0.06669718171712548,
      "grad_norm": 1.4921875,
      "learning_rate": 0.000999981872276892,
      "loss": 7.3436,
      "step": 510
    },
    {
      "epoch": 0.06735107565552867,
      "grad_norm": 1.53125,
      "learning_rate": 0.0009999592129653152,
      "loss": 7.33,
      "step": 515
    },
    {
      "epoch": 0.06800496959393186,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009999274905680762,
      "loss": 7.3327,
      "step": 520
    },
    {
      "epoch": 0.06865886353233505,
      "grad_norm": 1.4453125,
      "learning_rate": 0.000999886705724128,
      "loss": 7.2898,
      "step": 525
    },
    {
      "epoch": 0.06931275747073824,
      "grad_norm": 1.5078125,
      "learning_rate": 0.0009998368592549595,
      "loss": 7.253,
      "step": 530
    },
    {
      "epoch": 0.06996665140914143,
      "grad_norm": 1.6171875,
      "learning_rate": 0.0009997779521645793,
      "loss": 7.2486,
      "step": 535
    },
    {
      "epoch": 0.07062054534754462,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009997099856394946,
      "loss": 7.2762,
      "step": 540
    },
    {
      "epoch": 0.07127443928594782,
      "grad_norm": 1.5,
      "learning_rate": 0.0009996329610486886,
      "loss": 7.3113,
      "step": 545
    },
    {
      "epoch": 0.07192833322435101,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009995468799435915,
      "loss": 7.2034,
      "step": 550
    },
    {
      "epoch": 0.0725822271627542,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0009994517440580508,
      "loss": 7.3104,
      "step": 555
    },
    {
      "epoch": 0.07323612110115739,
      "grad_norm": 1.375,
      "learning_rate": 0.0009993475553082944,
      "loss": 7.2426,
      "step": 560
    },
    {
      "epoch": 0.07389001503956058,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0009992343157928943,
      "loss": 7.3101,
      "step": 565
    },
    {
      "epoch": 0.07454390897796377,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009991120277927223,
      "loss": 7.2279,
      "step": 570
    },
    {
      "epoch": 0.07519780291636696,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009989806937709053,
      "loss": 7.2125,
      "step": 575
    },
    {
      "epoch": 0.07585169685477015,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009988403163727754,
      "loss": 7.201,
      "step": 580
    },
    {
      "epoch": 0.07650559079317334,
      "grad_norm": 1.453125,
      "learning_rate": 0.0009986908984258162,
      "loss": 7.2014,
      "step": 585
    },
    {
      "epoch": 0.07715948473157654,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009985324429396065,
      "loss": 7.1926,
      "step": 590
    },
    {
      "epoch": 0.07781337866997973,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009983649531057593,
      "loss": 7.2502,
      "step": 595
    },
    {
      "epoch": 0.07846727260838292,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009981884322978576,
      "loss": 7.1844,
      "step": 600
    },
    {
      "epoch": 0.07912116654678611,
      "grad_norm": 1.7421875,
      "learning_rate": 0.000998002884071386,
      "loss": 7.2272,
      "step": 605
    },
    {
      "epoch": 0.0797750604851893,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009978083121636601,
      "loss": 7.1626,
      "step": 610
    },
    {
      "epoch": 0.08042895442359249,
      "grad_norm": 1.4609375,
      "learning_rate": 0.0009976047204937502,
      "loss": 7.1802,
      "step": 615
    },
    {
      "epoch": 0.08108284836199568,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0009973921131624032,
      "loss": 7.1191,
      "step": 620
    },
    {
      "epoch": 0.08173674230039887,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009971704944519593,
      "loss": 7.1534,
      "step": 625
    },
    {
      "epoch": 0.08239063623880206,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009969398688262658,
      "loss": 7.1173,
      "step": 630
    },
    {
      "epoch": 0.08304453017720526,
      "grad_norm": 1.34375,
      "learning_rate": 0.000996700240930588,
      "loss": 7.1504,
      "step": 635
    },
    {
      "epoch": 0.08369842411560845,
      "grad_norm": 1.5,
      "learning_rate": 0.000996451615591515,
      "loss": 7.1148,
      "step": 640
    },
    {
      "epoch": 0.08435231805401164,
      "grad_norm": 1.359375,
      "learning_rate": 0.000996193997816862,
      "loss": 7.161,
      "step": 645
    },
    {
      "epoch": 0.08500621199241483,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009959273927955706,
      "loss": 7.0662,
      "step": 650
    },
    {
      "epoch": 0.08566010593081802,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009956518058976034,
      "loss": 7.113,
      "step": 655
    },
    {
      "epoch": 0.08631399986922121,
      "grad_norm": 1.5,
      "learning_rate": 0.0009953672426738357,
      "loss": 7.1168,
      "step": 660
    },
    {
      "epoch": 0.0869678938076244,
      "grad_norm": 1.546875,
      "learning_rate": 0.000995073708855945,
      "loss": 7.0961,
      "step": 665
    },
    {
      "epoch": 0.0876217877460276,
      "grad_norm": 1.3203125,
      "learning_rate": 0.000994771210356294,
      "loss": 7.1477,
      "step": 670
    },
    {
      "epoch": 0.08827568168443078,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009944597532678121,
      "loss": 7.0991,
      "step": 675
    },
    {
      "epoch": 0.08892957562283398,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009941393438638731,
      "loss": 7.1046,
      "step": 680
    },
    {
      "epoch": 0.08958346956123717,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009938099885981684,
      "loss": 7.061,
      "step": 685
    },
    {
      "epoch": 0.09023736349964036,
      "grad_norm": 2.234375,
      "learning_rate": 0.000993471694104577,
      "loss": 7.0491,
      "step": 690
    },
    {
      "epoch": 0.09089125743804355,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0009931244671970319,
      "loss": 7.0846,
      "step": 695
    },
    {
      "epoch": 0.09154515137644674,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009927683148693834,
      "loss": 7.0289,
      "step": 700
    },
    {
      "epoch": 0.09219904531484993,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009924032442952568,
      "loss": 7.0298,
      "step": 705
    },
    {
      "epoch": 0.09285293925325312,
      "grad_norm": 1.40625,
      "learning_rate": 0.00099202926282791,
      "loss": 6.9826,
      "step": 710
    },
    {
      "epoch": 0.09350683319165631,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009916463780000832,
      "loss": 7.0655,
      "step": 715
    },
    {
      "epoch": 0.0941607271300595,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000991254597523849,
      "loss": 7.0537,
      "step": 720
    },
    {
      "epoch": 0.0948146210684627,
      "grad_norm": 1.4609375,
      "learning_rate": 0.000990853929290456,
      "loss": 6.9711,
      "step": 725
    },
    {
      "epoch": 0.09546851500686589,
      "grad_norm": 1.515625,
      "learning_rate": 0.0009904443813701703,
      "loss": 7.0258,
      "step": 730
    },
    {
      "epoch": 0.09612240894526908,
      "grad_norm": 1.46875,
      "learning_rate": 0.0009900259620121122,
      "loss": 6.9957,
      "step": 735
    },
    {
      "epoch": 0.09677630288367227,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009895986796440914,
      "loss": 7.0212,
      "step": 740
    },
    {
      "epoch": 0.09743019682207546,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009891625428724363,
      "loss": 6.9204,
      "step": 745
    },
    {
      "epoch": 0.09808409076047865,
      "grad_norm": 1.1875,
      "learning_rate": 0.0009887175604818205,
      "loss": 6.9733,
      "step": 750
    },
    {
      "epoch": 0.09873798469888184,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009882637414350868,
      "loss": 7.0053,
      "step": 755
    },
    {
      "epoch": 0.09939187863728503,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009878010948730653,
      "loss": 6.9522,
      "step": 760
    },
    {
      "epoch": 0.10004577257568822,
      "grad_norm": 1.328125,
      "learning_rate": 0.0009873296301143905,
      "loss": 6.9587,
      "step": 765
    },
    {
      "epoch": 0.10069966651409142,
      "grad_norm": 1.328125,
      "learning_rate": 0.000986849356655313,
      "loss": 6.9453,
      "step": 770
    },
    {
      "epoch": 0.1013535604524946,
      "grad_norm": 1.328125,
      "learning_rate": 0.000986360284169509,
      "loss": 6.9191,
      "step": 775
    },
    {
      "epoch": 0.1020074543908978,
      "grad_norm": 1.1875,
      "learning_rate": 0.000985862422507884,
      "loss": 7.017,
      "step": 780
    },
    {
      "epoch": 0.10266134832930099,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009853557816983756,
      "loss": 6.8862,
      "step": 785
    },
    {
      "epoch": 0.10331524226770418,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009848403719457507,
      "loss": 6.991,
      "step": 790
    },
    {
      "epoch": 0.10396913620610737,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009843162036314015,
      "loss": 6.9641,
      "step": 795
    },
    {
      "epoch": 0.10462303014451056,
      "grad_norm": 1.3671875,
      "learning_rate": 0.000983783287313134,
      "loss": 6.8792,
      "step": 800
    },
    {
      "epoch": 0.10527692408291375,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009832416337249574,
      "loss": 6.8891,
      "step": 805
    },
    {
      "epoch": 0.10593081802131694,
      "grad_norm": 1.34375,
      "learning_rate": 0.000982691253776867,
      "loss": 6.8837,
      "step": 810
    },
    {
      "epoch": 0.10658471195972014,
      "grad_norm": 1.390625,
      "learning_rate": 0.0009821321585546244,
      "loss": 6.8944,
      "step": 815
    },
    {
      "epoch": 0.10723860589812333,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0009815643593195345,
      "loss": 6.839,
      "step": 820
    },
    {
      "epoch": 0.10789249983652652,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009809878675082187,
      "loss": 6.8525,
      "step": 825
    },
    {
      "epoch": 0.10854639377492971,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009804026947323843,
      "loss": 6.8457,
      "step": 830
    },
    {
      "epoch": 0.1092002877133329,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009798088527785903,
      "loss": 6.9034,
      "step": 835
    },
    {
      "epoch": 0.10985418165173609,
      "grad_norm": 1.375,
      "learning_rate": 0.0009792063536080114,
      "loss": 6.8747,
      "step": 840
    },
    {
      "epoch": 0.11050807559013928,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009785952093561952,
      "loss": 6.8496,
      "step": 845
    },
    {
      "epoch": 0.11116196952854247,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009779754323328191,
      "loss": 6.8699,
      "step": 850
    },
    {
      "epoch": 0.11181586346694566,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009773470350214418,
      "loss": 6.8479,
      "step": 855
    },
    {
      "epoch": 0.11246975740534886,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009767100300792518,
      "loss": 6.8258,
      "step": 860
    },
    {
      "epoch": 0.11312365134375205,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009760644303368132,
      "loss": 6.7887,
      "step": 865
    },
    {
      "epoch": 0.11377754528215524,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0009754102487978059,
      "loss": 6.7641,
      "step": 870
    },
    {
      "epoch": 0.11443143922055843,
      "grad_norm": 2.140625,
      "learning_rate": 0.0009747474986387655,
      "loss": 6.9452,
      "step": 875
    },
    {
      "epoch": 0.11508533315896162,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009740761932088156,
      "loss": 6.8388,
      "step": 880
    },
    {
      "epoch": 0.11573922709736481,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009733963460294015,
      "loss": 6.8429,
      "step": 885
    },
    {
      "epoch": 0.116393121035768,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009727079707940157,
      "loss": 6.7286,
      "step": 890
    },
    {
      "epoch": 0.11704701497417119,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009720110813679232,
      "loss": 6.7943,
      "step": 895
    },
    {
      "epoch": 0.11770090891257438,
      "grad_norm": 1.671875,
      "learning_rate": 0.0009713056917878818,
      "loss": 6.7801,
      "step": 900
    },
    {
      "epoch": 0.11835480285097758,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009705918162618599,
      "loss": 6.7698,
      "step": 905
    },
    {
      "epoch": 0.11900869678938077,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00096986946916875,
      "loss": 6.8375,
      "step": 910
    },
    {
      "epoch": 0.11966259072778396,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009691386650580782,
      "loss": 6.7538,
      "step": 915
    },
    {
      "epoch": 0.12031648466618715,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009683994186497132,
      "loss": 6.6733,
      "step": 920
    },
    {
      "epoch": 0.12097037860459034,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0009676517448335676,
      "loss": 6.7383,
      "step": 925
    },
    {
      "epoch": 0.12162427254299353,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009668956586692994,
      "loss": 6.6867,
      "step": 930
    },
    {
      "epoch": 0.12227816648139672,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009661311753860083,
      "loss": 6.7132,
      "step": 935
    },
    {
      "epoch": 0.12293206041979991,
      "grad_norm": 1.21875,
      "learning_rate": 0.0009653583103819289,
      "loss": 6.7865,
      "step": 940
    },
    {
      "epoch": 0.1235859543582031,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009645770792241201,
      "loss": 6.8216,
      "step": 945
    },
    {
      "epoch": 0.1242398482966063,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009637874976481528,
      "loss": 6.6835,
      "step": 950
    },
    {
      "epoch": 0.12489374223500949,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009629895815577914,
      "loss": 6.7692,
      "step": 955
    },
    {
      "epoch": 0.12554763617341266,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009621833470246749,
      "loss": 6.8167,
      "step": 960
    },
    {
      "epoch": 0.12620153011181587,
      "grad_norm": 1.390625,
      "learning_rate": 0.000961368810287992,
      "loss": 6.6619,
      "step": 965
    },
    {
      "epoch": 0.12685542405021905,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009605459877541544,
      "loss": 6.6922,
      "step": 970
    },
    {
      "epoch": 0.12750931798862225,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009597148959964672,
      "loss": 6.7218,
      "step": 975
    },
    {
      "epoch": 0.12816321192702543,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009588755517547935,
      "loss": 6.7344,
      "step": 980
    },
    {
      "epoch": 0.12881710586542863,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009580279719352186,
      "loss": 6.7226,
      "step": 985
    },
    {
      "epoch": 0.1294709998038318,
      "grad_norm": 1.3125,
      "learning_rate": 0.0009571721736097089,
      "loss": 6.7793,
      "step": 990
    },
    {
      "epoch": 0.13012489374223501,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0009563081740157675,
      "loss": 6.69,
      "step": 995
    },
    {
      "epoch": 0.1307787876806382,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0009554359905560886,
      "loss": 6.7232,
      "step": 1000
    },
    {
      "epoch": 0.1314326816190414,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009545556407982051,
      "loss": 6.6788,
      "step": 1005
    },
    {
      "epoch": 0.13208657555744457,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0009536671424741358,
      "loss": 6.5623,
      "step": 1010
    },
    {
      "epoch": 0.13274046949584778,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009527705134800284,
      "loss": 6.6733,
      "step": 1015
    },
    {
      "epoch": 0.13339436343425096,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009518657718757985,
      "loss": 6.6977,
      "step": 1020
    },
    {
      "epoch": 0.13404825737265416,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009509529358847655,
      "loss": 6.6997,
      "step": 1025
    },
    {
      "epoch": 0.13470215131105734,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009500320238932869,
      "loss": 6.7829,
      "step": 1030
    },
    {
      "epoch": 0.13535604524946054,
      "grad_norm": 1.25,
      "learning_rate": 0.0009491030544503867,
      "loss": 6.7429,
      "step": 1035
    },
    {
      "epoch": 0.13600993918786372,
      "grad_norm": 1.421875,
      "learning_rate": 0.0009481660462673824,
      "loss": 6.6804,
      "step": 1040
    },
    {
      "epoch": 0.13666383312626693,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009472210182175074,
      "loss": 6.7292,
      "step": 1045
    },
    {
      "epoch": 0.1373177270646701,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009462679893355321,
      "loss": 6.6747,
      "step": 1050
    },
    {
      "epoch": 0.1379716210030733,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009453069788173795,
      "loss": 6.6158,
      "step": 1055
    },
    {
      "epoch": 0.13862551494147649,
      "grad_norm": 1.375,
      "learning_rate": 0.0009443380060197386,
      "loss": 6.6826,
      "step": 1060
    },
    {
      "epoch": 0.1392794088798797,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0009433610904596752,
      "loss": 6.6824,
      "step": 1065
    },
    {
      "epoch": 0.13993330281828287,
      "grad_norm": 1.171875,
      "learning_rate": 0.0009423762518142381,
      "loss": 6.717,
      "step": 1070
    },
    {
      "epoch": 0.14058719675668607,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009413835099200629,
      "loss": 6.6834,
      "step": 1075
    },
    {
      "epoch": 0.14124109069508925,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000940382884772973,
      "loss": 6.5807,
      "step": 1080
    },
    {
      "epoch": 0.14189498463349245,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009393743965275757,
      "loss": 6.6025,
      "step": 1085
    },
    {
      "epoch": 0.14254887857189563,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009383580654968576,
      "loss": 6.7214,
      "step": 1090
    },
    {
      "epoch": 0.14320277251029884,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009373339121517746,
      "loss": 6.6022,
      "step": 1095
    },
    {
      "epoch": 0.14385666644870201,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0009363019571208398,
      "loss": 6.5895,
      "step": 1100
    },
    {
      "epoch": 0.14451056038710522,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009352622211897077,
      "loss": 6.6554,
      "step": 1105
    },
    {
      "epoch": 0.1451644543255084,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0009342147253007562,
      "loss": 6.6725,
      "step": 1110
    },
    {
      "epoch": 0.1458183482639116,
      "grad_norm": 1.296875,
      "learning_rate": 0.0009331594905526644,
      "loss": 6.6015,
      "step": 1115
    },
    {
      "epoch": 0.14647224220231478,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0009320965381999872,
      "loss": 6.5637,
      "step": 1120
    },
    {
      "epoch": 0.14712613614071798,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009310258896527278,
      "loss": 6.7042,
      "step": 1125
    },
    {
      "epoch": 0.14778003007912116,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0009299475664759069,
      "loss": 6.6233,
      "step": 1130
    },
    {
      "epoch": 0.14843392401752437,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0009288615903891266,
      "loss": 6.6531,
      "step": 1135
    },
    {
      "epoch": 0.14908781795592754,
      "grad_norm": 1.234375,
      "learning_rate": 0.0009277679832661348,
      "loss": 6.6379,
      "step": 1140
    },
    {
      "epoch": 0.14974171189433075,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009266667671343836,
      "loss": 6.6197,
      "step": 1145
    },
    {
      "epoch": 0.15039560583273393,
      "grad_norm": 1.203125,
      "learning_rate": 0.0009255579641745861,
      "loss": 6.5892,
      "step": 1150
    },
    {
      "epoch": 0.15104949977113713,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0009244415967202692,
      "loss": 6.6192,
      "step": 1155
    },
    {
      "epoch": 0.1517033937095403,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009233176872573241,
      "loss": 6.5783,
      "step": 1160
    },
    {
      "epoch": 0.1523572876479435,
      "grad_norm": 1.265625,
      "learning_rate": 0.0009221862584235527,
      "loss": 6.611,
      "step": 1165
    },
    {
      "epoch": 0.1530111815863467,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009210473330082134,
      "loss": 6.4943,
      "step": 1170
    },
    {
      "epoch": 0.1536650755247499,
      "grad_norm": 2.421875,
      "learning_rate": 0.0009199009339515595,
      "loss": 6.6599,
      "step": 1175
    },
    {
      "epoch": 0.15431896946315307,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009187470843443794,
      "loss": 6.5553,
      "step": 1180
    },
    {
      "epoch": 0.15497286340155628,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0009175858074275303,
      "loss": 6.6543,
      "step": 1185
    },
    {
      "epoch": 0.15562675733995945,
      "grad_norm": 1.25,
      "learning_rate": 0.0009164171265914704,
      "loss": 6.6098,
      "step": 1190
    },
    {
      "epoch": 0.15628065127836266,
      "grad_norm": 1.359375,
      "learning_rate": 0.0009152410653757879,
      "loss": 6.5691,
      "step": 1195
    },
    {
      "epoch": 0.15693454521676584,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009140576474687263,
      "loss": 6.5794,
      "step": 1200
    },
    {
      "epoch": 0.15758843915516904,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009128668967067084,
      "loss": 6.5345,
      "step": 1205
    },
    {
      "epoch": 0.15824233309357222,
      "grad_norm": 1.40625,
      "learning_rate": 0.0009116688370738549,
      "loss": 6.5239,
      "step": 1210
    },
    {
      "epoch": 0.15889622703197542,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009104634927015022,
      "loss": 6.6302,
      "step": 1215
    },
    {
      "epoch": 0.1595501209703786,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0009092508878677157,
      "loss": 6.5343,
      "step": 1220
    },
    {
      "epoch": 0.1602040149087818,
      "grad_norm": 1.34375,
      "learning_rate": 0.0009080310469968015,
      "loss": 6.6725,
      "step": 1225
    },
    {
      "epoch": 0.16085790884718498,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0009068039946588139,
      "loss": 6.5371,
      "step": 1230
    },
    {
      "epoch": 0.1615118027855882,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0009055697555690607,
      "loss": 6.4683,
      "step": 1235
    },
    {
      "epoch": 0.16216569672399136,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009043283545876054,
      "loss": 6.5462,
      "step": 1240
    },
    {
      "epoch": 0.16281959066239457,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0009030798167187664,
      "loss": 6.5269,
      "step": 1245
    },
    {
      "epoch": 0.16347348460079775,
      "grad_norm": 1.28125,
      "learning_rate": 0.0009018241671106134,
      "loss": 6.5138,
      "step": 1250
    },
    {
      "epoch": 0.16412737853920095,
      "grad_norm": 1.203125,
      "learning_rate": 0.000900561431054461,
      "loss": 6.5613,
      "step": 1255
    },
    {
      "epoch": 0.16478127247760413,
      "grad_norm": 1.265625,
      "learning_rate": 0.000899291633984359,
      "loss": 6.532,
      "step": 1260
    },
    {
      "epoch": 0.16543516641600733,
      "grad_norm": 1.65625,
      "learning_rate": 0.0008980148014765801,
      "loss": 6.6125,
      "step": 1265
    },
    {
      "epoch": 0.1660890603544105,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008967309592491052,
      "loss": 6.5438,
      "step": 1270
    },
    {
      "epoch": 0.16674295429281372,
      "grad_norm": 2.046875,
      "learning_rate": 0.000895440133161105,
      "loss": 6.4701,
      "step": 1275
    },
    {
      "epoch": 0.1673968482312169,
      "grad_norm": 1.296875,
      "learning_rate": 0.0008941423492124193,
      "loss": 6.5871,
      "step": 1280
    },
    {
      "epoch": 0.1680507421696201,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008928376335430331,
      "loss": 6.476,
      "step": 1285
    },
    {
      "epoch": 0.16870463610802328,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00089152601243255,
      "loss": 6.4908,
      "step": 1290
    },
    {
      "epoch": 0.16935853004642648,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008902075122996638,
      "loss": 6.5745,
      "step": 1295
    },
    {
      "epoch": 0.17001242398482966,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000888882159701625,
      "loss": 6.449,
      "step": 1300
    },
    {
      "epoch": 0.17066631792323286,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008875499813337067,
      "loss": 6.5743,
      "step": 1305
    },
    {
      "epoch": 0.17132021186163604,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0008862110040286674,
      "loss": 6.4647,
      "step": 1310
    },
    {
      "epoch": 0.17197410580003925,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008848652547562088,
      "loss": 6.463,
      "step": 1315
    },
    {
      "epoch": 0.17262799973844242,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0008835127606224346,
      "loss": 6.4772,
      "step": 1320
    },
    {
      "epoch": 0.17328189367684563,
      "grad_norm": 1.359375,
      "learning_rate": 0.0008821535488693033,
      "loss": 6.4907,
      "step": 1325
    },
    {
      "epoch": 0.1739357876152488,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008807876468740795,
      "loss": 6.5567,
      "step": 1330
    },
    {
      "epoch": 0.174589681553652,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008794150821487834,
      "loss": 6.3954,
      "step": 1335
    },
    {
      "epoch": 0.1752435754920552,
      "grad_norm": 1.34375,
      "learning_rate": 0.0008780358823396352,
      "loss": 6.5513,
      "step": 1340
    },
    {
      "epoch": 0.1758974694304584,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008766500752264999,
      "loss": 6.4675,
      "step": 1345
    },
    {
      "epoch": 0.17655136336886157,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008752576887223263,
      "loss": 6.5179,
      "step": 1350
    },
    {
      "epoch": 0.17720525730726477,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008738587508725857,
      "loss": 6.4961,
      "step": 1355
    },
    {
      "epoch": 0.17785915124566795,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008724532898547066,
      "loss": 6.4981,
      "step": 1360
    },
    {
      "epoch": 0.17851304518407116,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008710413339775073,
      "loss": 6.479,
      "step": 1365
    },
    {
      "epoch": 0.17916693912247433,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008696229116806256,
      "loss": 6.4585,
      "step": 1370
    },
    {
      "epoch": 0.17982083306087754,
      "grad_norm": 1.1875,
      "learning_rate": 0.0008681980515339464,
      "loss": 6.4498,
      "step": 1375
    },
    {
      "epoch": 0.18047472699928072,
      "grad_norm": 1.09375,
      "learning_rate": 0.0008667667822370252,
      "loss": 6.4633,
      "step": 1380
    },
    {
      "epoch": 0.18112862093768392,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0008653291326185112,
      "loss": 6.5417,
      "step": 1385
    },
    {
      "epoch": 0.1817825148760871,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0008638851316355662,
      "loss": 6.4837,
      "step": 1390
    },
    {
      "epoch": 0.1824364088144903,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0008624348083732806,
      "loss": 6.5195,
      "step": 1395
    },
    {
      "epoch": 0.18309030275289348,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008609781920440892,
      "loss": 6.4522,
      "step": 1400
    },
    {
      "epoch": 0.18374419669129669,
      "grad_norm": 1.3125,
      "learning_rate": 0.000859515311987181,
      "loss": 6.3718,
      "step": 1405
    },
    {
      "epoch": 0.18439809062969986,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008580461976679099,
      "loss": 6.4162,
      "step": 1410
    },
    {
      "epoch": 0.18505198456810307,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008565708786771999,
      "loss": 6.4548,
      "step": 1415
    },
    {
      "epoch": 0.18570587850650624,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0008550893847309494,
      "loss": 6.4219,
      "step": 1420
    },
    {
      "epoch": 0.18635977244490945,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0008536017456694336,
      "loss": 6.4599,
      "step": 1425
    },
    {
      "epoch": 0.18701366638331263,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008521079914567018,
      "loss": 6.4352,
      "step": 1430
    },
    {
      "epoch": 0.1876675603217158,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0008506081521799755,
      "loss": 6.3716,
      "step": 1435
    },
    {
      "epoch": 0.188321454260119,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008491022580490416,
      "loss": 6.405,
      "step": 1440
    },
    {
      "epoch": 0.1889753481985222,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0008475903393956434,
      "loss": 6.4729,
      "step": 1445
    },
    {
      "epoch": 0.1896292421369254,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008460724266728708,
      "loss": 6.3943,
      "step": 1450
    },
    {
      "epoch": 0.19028313607532857,
      "grad_norm": 1.2578125,
      "learning_rate": 0.000844548550454546,
      "loss": 6.555,
      "step": 1455
    },
    {
      "epoch": 0.19093703001373177,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008430187414346081,
      "loss": 6.4777,
      "step": 1460
    },
    {
      "epoch": 0.19159092395213495,
      "grad_norm": 1.328125,
      "learning_rate": 0.000841483030426495,
      "loss": 6.442,
      "step": 1465
    },
    {
      "epoch": 0.19224481789053816,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008399414483625225,
      "loss": 6.41,
      "step": 1470
    },
    {
      "epoch": 0.19289871182894133,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0008383940262932608,
      "loss": 6.3268,
      "step": 1475
    },
    {
      "epoch": 0.19355260576734454,
      "grad_norm": 1.25,
      "learning_rate": 0.0008368407953869104,
      "loss": 6.4615,
      "step": 1480
    },
    {
      "epoch": 0.19420649970574771,
      "grad_norm": 1.5703125,
      "learning_rate": 0.0008352817869286727,
      "loss": 6.3631,
      "step": 1485
    },
    {
      "epoch": 0.19486039364415092,
      "grad_norm": 1.265625,
      "learning_rate": 0.0008337170323201212,
      "loss": 6.3452,
      "step": 1490
    },
    {
      "epoch": 0.1955142875825541,
      "grad_norm": 1.25,
      "learning_rate": 0.0008321465630785681,
      "loss": 6.3599,
      "step": 1495
    },
    {
      "epoch": 0.1961681815209573,
      "grad_norm": 1.234375,
      "learning_rate": 0.0008305704108364302,
      "loss": 6.4382,
      "step": 1500
    },
    {
      "epoch": 0.19682207545936048,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008289886073405908,
      "loss": 6.4176,
      "step": 1505
    },
    {
      "epoch": 0.19747596939776368,
      "grad_norm": 1.203125,
      "learning_rate": 0.0008274011844517614,
      "loss": 6.3905,
      "step": 1510
    },
    {
      "epoch": 0.19812986333616686,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0008258081741438395,
      "loss": 6.4099,
      "step": 1515
    },
    {
      "epoch": 0.19878375727457007,
      "grad_norm": 1.28125,
      "learning_rate": 0.0008242096085032637,
      "loss": 6.3997,
      "step": 1520
    },
    {
      "epoch": 0.19943765121297324,
      "grad_norm": 1.359375,
      "learning_rate": 0.000822605519728369,
      "loss": 6.3918,
      "step": 1525
    },
    {
      "epoch": 0.20009154515137645,
      "grad_norm": 1.25,
      "learning_rate": 0.000820995940128737,
      "loss": 6.3591,
      "step": 1530
    },
    {
      "epoch": 0.20074543908977963,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0008193809021245458,
      "loss": 6.4627,
      "step": 1535
    },
    {
      "epoch": 0.20139933302818283,
      "grad_norm": 1.3125,
      "learning_rate": 0.0008177604382459167,
      "loss": 6.354,
      "step": 1540
    },
    {
      "epoch": 0.202053226966586,
      "grad_norm": 1.34375,
      "learning_rate": 0.000816134581132259,
      "loss": 6.3421,
      "step": 1545
    },
    {
      "epoch": 0.2027071209049892,
      "grad_norm": 1.296875,
      "learning_rate": 0.000814503363531613,
      "loss": 6.4077,
      "step": 1550
    },
    {
      "epoch": 0.2033610148433924,
      "grad_norm": 1.21875,
      "learning_rate": 0.0008128668182999891,
      "loss": 6.316,
      "step": 1555
    },
    {
      "epoch": 0.2040149087817956,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000811224978400708,
      "loss": 6.3829,
      "step": 1560
    },
    {
      "epoch": 0.20466880272019877,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0008095778769037352,
      "loss": 6.4049,
      "step": 1565
    },
    {
      "epoch": 0.20532269665860198,
      "grad_norm": 1.1640625,
      "learning_rate": 0.000807925546985015,
      "loss": 6.3646,
      "step": 1570
    },
    {
      "epoch": 0.20597659059700515,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0008062680219258034,
      "loss": 6.3425,
      "step": 1575
    },
    {
      "epoch": 0.20663048453540836,
      "grad_norm": 1.6875,
      "learning_rate": 0.0008046053351119965,
      "loss": 6.4198,
      "step": 1580
    },
    {
      "epoch": 0.20728437847381154,
      "grad_norm": 1.171875,
      "learning_rate": 0.0008029375200334588,
      "loss": 6.3546,
      "step": 1585
    },
    {
      "epoch": 0.20793827241221474,
      "grad_norm": 1.265625,
      "learning_rate": 0.000801264610283348,
      "loss": 6.3964,
      "step": 1590
    },
    {
      "epoch": 0.20859216635061792,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007995866395574396,
      "loss": 6.3983,
      "step": 1595
    },
    {
      "epoch": 0.20924606028902112,
      "grad_norm": 1.34375,
      "learning_rate": 0.0007979036416534462,
      "loss": 6.3851,
      "step": 1600
    },
    {
      "epoch": 0.2098999542274243,
      "grad_norm": 1.109375,
      "learning_rate": 0.0007962156504703387,
      "loss": 6.3671,
      "step": 1605
    },
    {
      "epoch": 0.2105538481658275,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007945227000076627,
      "loss": 6.3402,
      "step": 1610
    },
    {
      "epoch": 0.21120774210423068,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000792824824364853,
      "loss": 6.3794,
      "step": 1615
    },
    {
      "epoch": 0.2118616360426339,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007911220577405485,
      "loss": 6.4569,
      "step": 1620
    },
    {
      "epoch": 0.21251552998103707,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007894144344319014,
      "loss": 6.3138,
      "step": 1625
    },
    {
      "epoch": 0.21316942391944027,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007877019888338881,
      "loss": 6.2889,
      "step": 1630
    },
    {
      "epoch": 0.21382331785784345,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0007859847554386145,
      "loss": 6.335,
      "step": 1635
    },
    {
      "epoch": 0.21447721179624665,
      "grad_norm": 1.25,
      "learning_rate": 0.0007842627688346239,
      "loss": 6.3554,
      "step": 1640
    },
    {
      "epoch": 0.21513110573464983,
      "grad_norm": 1.1875,
      "learning_rate": 0.0007825360637061976,
      "loss": 6.3695,
      "step": 1645
    },
    {
      "epoch": 0.21578499967305304,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0007808046748326578,
      "loss": 6.3404,
      "step": 1650
    },
    {
      "epoch": 0.2164388936114562,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007790686370876671,
      "loss": 6.3522,
      "step": 1655
    },
    {
      "epoch": 0.21709278754985942,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007773279854385256,
      "loss": 6.3317,
      "step": 1660
    },
    {
      "epoch": 0.2177466814882626,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007755827549454663,
      "loss": 6.3401,
      "step": 1665
    },
    {
      "epoch": 0.2184005754266658,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007738329807609499,
      "loss": 6.3169,
      "step": 1670
    },
    {
      "epoch": 0.21905446936506898,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007720786981289562,
      "loss": 6.2968,
      "step": 1675
    },
    {
      "epoch": 0.21970836330347218,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007703199423842736,
      "loss": 6.4019,
      "step": 1680
    },
    {
      "epoch": 0.22036225724187536,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0007685567489517887,
      "loss": 6.3429,
      "step": 1685
    },
    {
      "epoch": 0.22101615118027856,
      "grad_norm": 1.234375,
      "learning_rate": 0.0007667891533457718,
      "loss": 6.3908,
      "step": 1690
    },
    {
      "epoch": 0.22167004511868174,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000765017191169162,
      "loss": 6.3052,
      "step": 1695
    },
    {
      "epoch": 0.22232393905708495,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007632408981128494,
      "loss": 6.3352,
      "step": 1700
    },
    {
      "epoch": 0.22297783299548812,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007614603099549574,
      "loss": 6.3274,
      "step": 1705
    },
    {
      "epoch": 0.22363172693389133,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0007596754625601211,
      "loss": 6.3494,
      "step": 1710
    },
    {
      "epoch": 0.2242856208722945,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007578863918787654,
      "loss": 6.2444,
      "step": 1715
    },
    {
      "epoch": 0.2249395148106977,
      "grad_norm": 1.296875,
      "learning_rate": 0.0007560931339463805,
      "loss": 6.3518,
      "step": 1720
    },
    {
      "epoch": 0.2255934087491009,
      "grad_norm": 1.34375,
      "learning_rate": 0.000754295724882796,
      "loss": 6.2658,
      "step": 1725
    },
    {
      "epoch": 0.2262473026875041,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007524942008914544,
      "loss": 6.3173,
      "step": 1730
    },
    {
      "epoch": 0.22690119662590727,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007506885982586809,
      "loss": 6.3474,
      "step": 1735
    },
    {
      "epoch": 0.22755509056431047,
      "grad_norm": 1.25,
      "learning_rate": 0.000748878953352952,
      "loss": 6.3453,
      "step": 1740
    },
    {
      "epoch": 0.22820898450271365,
      "grad_norm": 1.25,
      "learning_rate": 0.0007470653026241649,
      "loss": 6.3094,
      "step": 1745
    },
    {
      "epoch": 0.22886287844111686,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007452476826029011,
      "loss": 6.3593,
      "step": 1750
    },
    {
      "epoch": 0.22951677237952003,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0007434261298996925,
      "loss": 6.4065,
      "step": 1755
    },
    {
      "epoch": 0.23017066631792324,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0007416006812042827,
      "loss": 6.3603,
      "step": 1760
    },
    {
      "epoch": 0.23082456025632642,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007397713732848884,
      "loss": 6.3235,
      "step": 1765
    },
    {
      "epoch": 0.23147845419472962,
      "grad_norm": 1.21875,
      "learning_rate": 0.000737938242987459,
      "loss": 6.2637,
      "step": 1770
    },
    {
      "epoch": 0.2321323481331328,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0007361013272349343,
      "loss": 6.3023,
      "step": 1775
    },
    {
      "epoch": 0.232786242071536,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0007342606630265008,
      "loss": 6.2601,
      "step": 1780
    },
    {
      "epoch": 0.23344013600993918,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0007324162874368462,
      "loss": 6.2756,
      "step": 1785
    },
    {
      "epoch": 0.23409402994834239,
      "grad_norm": 1.203125,
      "learning_rate": 0.0007305682376154135,
      "loss": 6.291,
      "step": 1790
    },
    {
      "epoch": 0.23474792388674556,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0007287165507856513,
      "loss": 6.305,
      "step": 1795
    },
    {
      "epoch": 0.23540181782514877,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007268612642442657,
      "loss": 6.2698,
      "step": 1800
    },
    {
      "epoch": 0.23605571176355195,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007250024153604678,
      "loss": 6.3144,
      "step": 1805
    },
    {
      "epoch": 0.23670960570195515,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0007231400415752213,
      "loss": 6.2995,
      "step": 1810
    },
    {
      "epoch": 0.23736349964035833,
      "grad_norm": 1.328125,
      "learning_rate": 0.0007212741804004889,
      "loss": 6.3645,
      "step": 1815
    },
    {
      "epoch": 0.23801739357876153,
      "grad_norm": 1.3984375,
      "learning_rate": 0.0007194048694184764,
      "loss": 6.2423,
      "step": 1820
    },
    {
      "epoch": 0.2386712875171647,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007175321462808753,
      "loss": 6.2971,
      "step": 1825
    },
    {
      "epoch": 0.23932518145556791,
      "grad_norm": 1.265625,
      "learning_rate": 0.0007156560487081052,
      "loss": 6.2813,
      "step": 1830
    },
    {
      "epoch": 0.2399790753939711,
      "grad_norm": 1.21875,
      "learning_rate": 0.000713776614488553,
      "loss": 6.2491,
      "step": 1835
    },
    {
      "epoch": 0.2406329693323743,
      "grad_norm": 1.375,
      "learning_rate": 0.0007118938814778132,
      "loss": 6.3472,
      "step": 1840
    },
    {
      "epoch": 0.24128686327077747,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0007100078875979239,
      "loss": 6.2727,
      "step": 1845
    },
    {
      "epoch": 0.24194075720918068,
      "grad_norm": 1.21875,
      "learning_rate": 0.0007081186708366044,
      "loss": 6.2461,
      "step": 1850
    },
    {
      "epoch": 0.24259465114758386,
      "grad_norm": 1.28125,
      "learning_rate": 0.0007062262692464884,
      "loss": 6.365,
      "step": 1855
    },
    {
      "epoch": 0.24324854508598706,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0007043307209443591,
      "loss": 6.2525,
      "step": 1860
    },
    {
      "epoch": 0.24390243902439024,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0007024320641103812,
      "loss": 6.2895,
      "step": 1865
    },
    {
      "epoch": 0.24455633296279344,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0007005303369873305,
      "loss": 6.2936,
      "step": 1870
    },
    {
      "epoch": 0.24521022690119662,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006986255778798253,
      "loss": 6.2707,
      "step": 1875
    },
    {
      "epoch": 0.24586412083959983,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0006967178251535539,
      "loss": 6.256,
      "step": 1880
    },
    {
      "epoch": 0.246518014778003,
      "grad_norm": 1.1875,
      "learning_rate": 0.0006948071172345022,
      "loss": 6.2517,
      "step": 1885
    },
    {
      "epoch": 0.2471719087164062,
      "grad_norm": 2.25,
      "learning_rate": 0.00069289349260818,
      "loss": 6.2196,
      "step": 1890
    },
    {
      "epoch": 0.24782580265480939,
      "grad_norm": 1.171875,
      "learning_rate": 0.0006909769898188447,
      "loss": 6.2582,
      "step": 1895
    },
    {
      "epoch": 0.2484796965932126,
      "grad_norm": 1.296875,
      "learning_rate": 0.0006890576474687264,
      "loss": 6.2445,
      "step": 1900
    },
    {
      "epoch": 0.24913359053161577,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006871355042172493,
      "loss": 6.3192,
      "step": 1905
    },
    {
      "epoch": 0.24978748447001897,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006852105987802534,
      "loss": 6.263,
      "step": 1910
    },
    {
      "epoch": 0.2504413784084222,
      "grad_norm": 1.296875,
      "learning_rate": 0.000683282969929215,
      "loss": 6.19,
      "step": 1915
    },
    {
      "epoch": 0.2510952723468253,
      "grad_norm": 1.171875,
      "learning_rate": 0.000681352656490465,
      "loss": 6.1907,
      "step": 1920
    },
    {
      "epoch": 0.25174916628522853,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006794196973444078,
      "loss": 6.2582,
      "step": 1925
    },
    {
      "epoch": 0.25240306022363174,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006774841314247372,
      "loss": 6.2187,
      "step": 1930
    },
    {
      "epoch": 0.25305695416203494,
      "grad_norm": 1.25,
      "learning_rate": 0.0006755459977176532,
      "loss": 6.1929,
      "step": 1935
    },
    {
      "epoch": 0.2537108481004381,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006736053352610756,
      "loss": 6.2847,
      "step": 1940
    },
    {
      "epoch": 0.2543647420388413,
      "grad_norm": 1.328125,
      "learning_rate": 0.0006716621831438588,
      "loss": 6.2249,
      "step": 1945
    },
    {
      "epoch": 0.2550186359772445,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006697165805050039,
      "loss": 6.1595,
      "step": 1950
    },
    {
      "epoch": 0.2556725299156477,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0006677685665328701,
      "loss": 6.2418,
      "step": 1955
    },
    {
      "epoch": 0.25632642385405086,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006658181804643862,
      "loss": 6.1652,
      "step": 1960
    },
    {
      "epoch": 0.25698031779245406,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006638654615842593,
      "loss": 6.2339,
      "step": 1965
    },
    {
      "epoch": 0.25763421173085727,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006619104492241848,
      "loss": 6.2378,
      "step": 1970
    },
    {
      "epoch": 0.25828810566926047,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006599531827620523,
      "loss": 6.2416,
      "step": 1975
    },
    {
      "epoch": 0.2589419996076636,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006579937016211547,
      "loss": 6.1776,
      "step": 1980
    },
    {
      "epoch": 0.2595958935460668,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006560320452693922,
      "loss": 6.1817,
      "step": 1985
    },
    {
      "epoch": 0.26024978748447003,
      "grad_norm": 1.25,
      "learning_rate": 0.0006540682532184785,
      "loss": 6.2917,
      "step": 1990
    },
    {
      "epoch": 0.26090368142287323,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0006521023650231444,
      "loss": 6.2002,
      "step": 1995
    },
    {
      "epoch": 0.2615575753612764,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006501344202803415,
      "loss": 6.2898,
      "step": 2000
    },
    {
      "epoch": 0.2622114692996796,
      "grad_norm": 1.421875,
      "learning_rate": 0.0006481644586284442,
      "loss": 6.1994,
      "step": 2005
    },
    {
      "epoch": 0.2628653632380828,
      "grad_norm": 1.40625,
      "learning_rate": 0.0006461925197464516,
      "loss": 6.2312,
      "step": 2010
    },
    {
      "epoch": 0.263519257176486,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0006442186433531879,
      "loss": 6.2085,
      "step": 2015
    },
    {
      "epoch": 0.26417315111488915,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0006422428692065038,
      "loss": 6.2685,
      "step": 2020
    },
    {
      "epoch": 0.26482704505329235,
      "grad_norm": 1.28125,
      "learning_rate": 0.0006402652371024727,
      "loss": 6.1797,
      "step": 2025
    },
    {
      "epoch": 0.26548093899169556,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006382857868745929,
      "loss": 6.2338,
      "step": 2030
    },
    {
      "epoch": 0.26613483293009876,
      "grad_norm": 1.2890625,
      "learning_rate": 0.000636304558392982,
      "loss": 6.1684,
      "step": 2035
    },
    {
      "epoch": 0.2667887268685019,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0006343215915635762,
      "loss": 6.219,
      "step": 2040
    },
    {
      "epoch": 0.2674426208069051,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006323369263273246,
      "loss": 6.2483,
      "step": 2045
    },
    {
      "epoch": 0.2680965147453083,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006303506026593865,
      "loss": 6.1543,
      "step": 2050
    },
    {
      "epoch": 0.26875040868371153,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0006283626605683249,
      "loss": 6.2272,
      "step": 2055
    },
    {
      "epoch": 0.2694043026221147,
      "grad_norm": 1.375,
      "learning_rate": 0.000626373140095301,
      "loss": 6.1111,
      "step": 2060
    },
    {
      "epoch": 0.2700581965605179,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006243820813132683,
      "loss": 6.2162,
      "step": 2065
    },
    {
      "epoch": 0.2707120904989211,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0006223895243261644,
      "loss": 6.2731,
      "step": 2070
    },
    {
      "epoch": 0.2713659844373243,
      "grad_norm": 1.234375,
      "learning_rate": 0.000620395509268104,
      "loss": 6.1539,
      "step": 2075
    },
    {
      "epoch": 0.27201987837572744,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0006184000763025703,
      "loss": 6.1705,
      "step": 2080
    },
    {
      "epoch": 0.27267377231413065,
      "grad_norm": 1.265625,
      "learning_rate": 0.0006164032656216059,
      "loss": 6.1859,
      "step": 2085
    },
    {
      "epoch": 0.27332766625253385,
      "grad_norm": 1.3125,
      "learning_rate": 0.0006144051174450039,
      "loss": 6.2442,
      "step": 2090
    },
    {
      "epoch": 0.27398156019093706,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0006124056720194963,
      "loss": 6.1768,
      "step": 2095
    },
    {
      "epoch": 0.2746354541293402,
      "grad_norm": 1.234375,
      "learning_rate": 0.000610404969617945,
      "loss": 6.1013,
      "step": 2100
    },
    {
      "epoch": 0.2752893480677434,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0006084030505385298,
      "loss": 6.2312,
      "step": 2105
    },
    {
      "epoch": 0.2759432420061466,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000606399955103937,
      "loss": 6.1479,
      "step": 2110
    },
    {
      "epoch": 0.2765971359445498,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006043957236605465,
      "loss": 6.1309,
      "step": 2115
    },
    {
      "epoch": 0.27725102988295297,
      "grad_norm": 3.21875,
      "learning_rate": 0.0006023903965776203,
      "loss": 6.2399,
      "step": 2120
    },
    {
      "epoch": 0.2779049238213562,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0006003840142464886,
      "loss": 6.1833,
      "step": 2125
    },
    {
      "epoch": 0.2785588177597594,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0005983766170797363,
      "loss": 6.218,
      "step": 2130
    },
    {
      "epoch": 0.2792127116981626,
      "grad_norm": 1.515625,
      "learning_rate": 0.0005963682455103892,
      "loss": 6.1316,
      "step": 2135
    },
    {
      "epoch": 0.27986660563656574,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005943589399910996,
      "loss": 6.235,
      "step": 2140
    },
    {
      "epoch": 0.28052049957496894,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005923487409933316,
      "loss": 6.1523,
      "step": 2145
    },
    {
      "epoch": 0.28117439351337215,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005903376890065452,
      "loss": 6.1137,
      "step": 2150
    },
    {
      "epoch": 0.2818282874517753,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005883258245373818,
      "loss": 6.1378,
      "step": 2155
    },
    {
      "epoch": 0.2824821813901785,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005863131881088478,
      "loss": 6.142,
      "step": 2160
    },
    {
      "epoch": 0.2831360753285817,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005842998202594984,
      "loss": 6.1987,
      "step": 2165
    },
    {
      "epoch": 0.2837899692669849,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005822857615426207,
      "loss": 6.1106,
      "step": 2170
    },
    {
      "epoch": 0.28444386320538806,
      "grad_norm": 1.8203125,
      "learning_rate": 0.0005802710525254179,
      "loss": 6.1704,
      "step": 2175
    },
    {
      "epoch": 0.28509775714379126,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005782557337881911,
      "loss": 6.1612,
      "step": 2180
    },
    {
      "epoch": 0.28575165108219447,
      "grad_norm": 1.328125,
      "learning_rate": 0.0005762398459235225,
      "loss": 6.2012,
      "step": 2185
    },
    {
      "epoch": 0.2864055450205977,
      "grad_norm": 1.203125,
      "learning_rate": 0.0005742234295354577,
      "loss": 6.1321,
      "step": 2190
    },
    {
      "epoch": 0.2870594389590008,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005722065252386881,
      "loss": 6.1904,
      "step": 2195
    },
    {
      "epoch": 0.28771333289740403,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005701891736577318,
      "loss": 6.1983,
      "step": 2200
    },
    {
      "epoch": 0.28836722683580723,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005681714154261168,
      "loss": 6.1008,
      "step": 2205
    },
    {
      "epoch": 0.28902112077421044,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005661532911855621,
      "loss": 6.1466,
      "step": 2210
    },
    {
      "epoch": 0.2896750147126136,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0005641348415851577,
      "loss": 6.1397,
      "step": 2215
    },
    {
      "epoch": 0.2903289086510168,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005621161072805483,
      "loss": 6.1541,
      "step": 2220
    },
    {
      "epoch": 0.29098280258942,
      "grad_norm": 1.375,
      "learning_rate": 0.0005600971289331123,
      "loss": 6.1941,
      "step": 2225
    },
    {
      "epoch": 0.2916366965278232,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005580779472091438,
      "loss": 6.1416,
      "step": 2230
    },
    {
      "epoch": 0.29229059046622635,
      "grad_norm": 1.21875,
      "learning_rate": 0.0005560586027790334,
      "loss": 6.1109,
      "step": 2235
    },
    {
      "epoch": 0.29294448440462956,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0005540391363164491,
      "loss": 6.1435,
      "step": 2240
    },
    {
      "epoch": 0.29359837834303276,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005520195884975165,
      "loss": 6.1458,
      "step": 2245
    },
    {
      "epoch": 0.29425227228143597,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00055,
      "loss": 6.1721,
      "step": 2250
    },
    {
      "epoch": 0.2949061662198391,
      "grad_norm": 1.234375,
      "learning_rate": 0.0005479804115024836,
      "loss": 6.1512,
      "step": 2255
    },
    {
      "epoch": 0.2955600601582423,
      "grad_norm": 1.265625,
      "learning_rate": 0.000545960863683551,
      "loss": 6.0768,
      "step": 2260
    },
    {
      "epoch": 0.2962139540966455,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005439413972209664,
      "loss": 6.2096,
      "step": 2265
    },
    {
      "epoch": 0.29686784803504873,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0005419220527908563,
      "loss": 6.1144,
      "step": 2270
    },
    {
      "epoch": 0.2975217419734519,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005399028710668879,
      "loss": 6.1411,
      "step": 2275
    },
    {
      "epoch": 0.2981756359118551,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0005378838927194518,
      "loss": 6.0838,
      "step": 2280
    },
    {
      "epoch": 0.2988295298502583,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0005358651584148423,
      "loss": 6.1795,
      "step": 2285
    },
    {
      "epoch": 0.2994834237886615,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000533846708814438,
      "loss": 6.1312,
      "step": 2290
    },
    {
      "epoch": 0.30013731772706465,
      "grad_norm": 1.25,
      "learning_rate": 0.0005318285845738831,
      "loss": 6.0628,
      "step": 2295
    },
    {
      "epoch": 0.30079121166546785,
      "grad_norm": 1.4921875,
      "learning_rate": 0.0005298108263422685,
      "loss": 6.1802,
      "step": 2300
    },
    {
      "epoch": 0.30144510560387106,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0005277934747613122,
      "loss": 6.0886,
      "step": 2305
    },
    {
      "epoch": 0.30209899954227426,
      "grad_norm": 1.3125,
      "learning_rate": 0.0005257765704645423,
      "loss": 6.1344,
      "step": 2310
    },
    {
      "epoch": 0.3027528934806774,
      "grad_norm": 1.359375,
      "learning_rate": 0.0005237601540764777,
      "loss": 6.0617,
      "step": 2315
    },
    {
      "epoch": 0.3034067874190806,
      "grad_norm": 1.296875,
      "learning_rate": 0.0005217442662118091,
      "loss": 6.0991,
      "step": 2320
    },
    {
      "epoch": 0.3040606813574838,
      "grad_norm": 1.28125,
      "learning_rate": 0.0005197289474745822,
      "loss": 6.0716,
      "step": 2325
    },
    {
      "epoch": 0.304714575295887,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0005177142384573793,
      "loss": 6.1183,
      "step": 2330
    },
    {
      "epoch": 0.3053684692342902,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0005157001797405019,
      "loss": 6.1452,
      "step": 2335
    },
    {
      "epoch": 0.3060223631726934,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005136868118911523,
      "loss": 6.1009,
      "step": 2340
    },
    {
      "epoch": 0.3066762571110966,
      "grad_norm": 1.34375,
      "learning_rate": 0.0005116741754626184,
      "loss": 6.0759,
      "step": 2345
    },
    {
      "epoch": 0.3073301510494998,
      "grad_norm": 1.2734375,
      "learning_rate": 0.000509662310993455,
      "loss": 6.0946,
      "step": 2350
    },
    {
      "epoch": 0.30798404498790294,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0005076512590066686,
      "loss": 6.1445,
      "step": 2355
    },
    {
      "epoch": 0.30863793892630614,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0005056410600089004,
      "loss": 6.1435,
      "step": 2360
    },
    {
      "epoch": 0.30929183286470935,
      "grad_norm": 1.25,
      "learning_rate": 0.000503631754489611,
      "loss": 6.1043,
      "step": 2365
    },
    {
      "epoch": 0.30994572680311255,
      "grad_norm": 1.265625,
      "learning_rate": 0.0005016233829202638,
      "loss": 6.0846,
      "step": 2370
    },
    {
      "epoch": 0.3105996207415157,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0004996159857535115,
      "loss": 6.0687,
      "step": 2375
    },
    {
      "epoch": 0.3112535146799189,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0004976096034223799,
      "loss": 6.0996,
      "step": 2380
    },
    {
      "epoch": 0.3119074086183221,
      "grad_norm": 1.359375,
      "learning_rate": 0.0004956042763394536,
      "loss": 6.1266,
      "step": 2385
    },
    {
      "epoch": 0.3125613025567253,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004936000448960631,
      "loss": 6.0878,
      "step": 2390
    },
    {
      "epoch": 0.31321519649512847,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0004915969494614701,
      "loss": 6.0788,
      "step": 2395
    },
    {
      "epoch": 0.3138690904335317,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0004895950303820552,
      "loss": 6.0546,
      "step": 2400
    },
    {
      "epoch": 0.3145229843719349,
      "grad_norm": 1.1875,
      "learning_rate": 0.00048759432798050397,
      "loss": 6.1952,
      "step": 2405
    },
    {
      "epoch": 0.3151768783103381,
      "grad_norm": 1.15625,
      "learning_rate": 0.00048559488255499626,
      "loss": 6.1099,
      "step": 2410
    },
    {
      "epoch": 0.31583077224874123,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004835967343783941,
      "loss": 6.0544,
      "step": 2415
    },
    {
      "epoch": 0.31648466618714444,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0004815999236974298,
      "loss": 6.0583,
      "step": 2420
    },
    {
      "epoch": 0.31713856012554764,
      "grad_norm": 1.265625,
      "learning_rate": 0.00047960449073189605,
      "loss": 6.0961,
      "step": 2425
    },
    {
      "epoch": 0.31779245406395085,
      "grad_norm": 1.140625,
      "learning_rate": 0.0004776104756738359,
      "loss": 6.0921,
      "step": 2430
    },
    {
      "epoch": 0.318446348002354,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0004756179186867319,
      "loss": 6.1023,
      "step": 2435
    },
    {
      "epoch": 0.3191002419407572,
      "grad_norm": 1.171875,
      "learning_rate": 0.000473626859904699,
      "loss": 6.106,
      "step": 2440
    },
    {
      "epoch": 0.3197541358791604,
      "grad_norm": 1.3125,
      "learning_rate": 0.0004716373394316752,
      "loss": 6.1096,
      "step": 2445
    },
    {
      "epoch": 0.3204080298175636,
      "grad_norm": 1.171875,
      "learning_rate": 0.0004696493973406135,
      "loss": 6.0056,
      "step": 2450
    },
    {
      "epoch": 0.32106192375596676,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0004676630736726754,
      "loss": 6.092,
      "step": 2455
    },
    {
      "epoch": 0.32171581769436997,
      "grad_norm": 1.34375,
      "learning_rate": 0.00046567840843642387,
      "loss": 6.0769,
      "step": 2460
    },
    {
      "epoch": 0.32236971163277317,
      "grad_norm": 1.265625,
      "learning_rate": 0.00046369544160701807,
      "loss": 6.0364,
      "step": 2465
    },
    {
      "epoch": 0.3230236055711764,
      "grad_norm": 1.21875,
      "learning_rate": 0.0004617142131254073,
      "loss": 6.0695,
      "step": 2470
    },
    {
      "epoch": 0.3236774995095795,
      "grad_norm": 2.109375,
      "learning_rate": 0.00045973476289752727,
      "loss": 6.1327,
      "step": 2475
    },
    {
      "epoch": 0.32433139344798273,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00045775713079349636,
      "loss": 6.0117,
      "step": 2480
    },
    {
      "epoch": 0.32498528738638593,
      "grad_norm": 1.25,
      "learning_rate": 0.00045578135664681196,
      "loss": 6.1191,
      "step": 2485
    },
    {
      "epoch": 0.32563918132478914,
      "grad_norm": 1.25,
      "learning_rate": 0.00045380748025354844,
      "loss": 6.1042,
      "step": 2490
    },
    {
      "epoch": 0.3262930752631923,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000451835541371556,
      "loss": 6.0728,
      "step": 2495
    },
    {
      "epoch": 0.3269469692015955,
      "grad_norm": 1.234375,
      "learning_rate": 0.0004498655797196586,
      "loss": 6.0861,
      "step": 2500
    }
  ],
  "logging_steps": 5,
  "max_steps": 4000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3342538488545280.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}